Development and Application of Dual Atmospheric Tracer Techniques for the Characterization of Pollutant Transport and Dispersion Citation Lamb, Brian Kent (1978) Development and Application of Dual Atmospheric Tracer Techniques for the Characterization of Pollutant Transport and Dispersion. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/6djp-n724. https://resolver.caltech.edu/CaltechTHESIS:12022025-232421338 Abstract Environmental concerns over urban development and industrial growth have resulted in increased interest in atmospheric tracers. Expanding urban areas in coastal regions make it particularly important to understand the effects of complex meteorological patterns and rough terrain upon the transport and dispersion of atmospheric pollutants. The purpose of this thesis was to further develop single and double atmospheric tracer techniques which would be economical and efficient for large-scale field studies. Sulfur hexafluoride and bromo-trifluoromethane were selected as tracers since they are both inert gases detectable by electron capture gas chromatography in the parts per trillion range . A tracer system was developed which permitted the collection and analysis of as many as 1000 air samples per day. This system was then used to demonstrate the application of tracer techniques to important problems of current interest in both coastal and noncoastal zones over distances as great as 100 km. Atmospheric tracer studies, conducted in Southern California during September, 1975, investigated the possible transport of pollutants between the Los Angeles Basin and the Oxnard Plain. The SF 6 data clearly showed that pollutant transport occurs from the Oxnard Plain along the Malibu coast into the Los Angeles Basin. A second series of tracer studies using SF 6 and CBrF 3 were completed within the California Delta Region during September, 1976. The purpose of these tests was to analyze the impact of possible industrial developments in the Montezuma Hills of the Delta Region. The relative impact of existing and proposed industrial sources was investigated by simultaneously releasing CBrF 3 from the Montezuma Hills and SF 6 from upwind of the Delta Region. Results indicated that plumes from both areas were carried southeast over Stockton into the San Joaquin Valley. Data from both of the coastal studies indicated that because of periods of steady coastal winds, the commonly used Hino correction grossly underestimated hourly averaged concentrations computed from 10-second averaged concentrations. Tracer concentrations predicted from the Gaussian plume model overestimated values observed in the Oxnard study because of increased dispersion over rough terrain. Predicted concentrations showed good agreement with observed values in the Delta study where the terrain was relatively flat. When sufficient data were available, surface air parcel trajectories determined from numerical solutions of the two-dimensional conservation equation were found to be more accurate than trajectories drawn from meteorological streamline maps. Five large-scale tracer studies utilizing SF 6 and CBrF 3 were performed in urban St. Louis during August, 1975, to probe the relative impact of emissions from low-level and elevated sources. Releases of SF 6 were used to simulate elevated sources, while simultaneous releases of CBrF 3 were used to simulate low-level sources. The dispersion of emissions from low-level sources was initially enhanced in contrast to the dispersion of emissions from elevated sources. Analysis of the St. Louis t racer data resulted in the development of the concept of a limit model of air quality. By considering the relative impact of different sources and formulating necessary assumptions in terms of an upper or lower bound , the limit model approach offers a new perspective upon contemporary air quality problems. The application of this concept was demonstrated in the development of an upper limit model of the relative impact of elevated rural sources and low-level urban sources upon urban SO 2 levels in St. Louis. Results of this application indicated that rural sources contribute no more than 25% of the average total urban SO 2 concentration, even though rural sources can dominate urban air quality in portions of the urban region during relatively large portions of the time. As a means of applying experience gained from atmospheric tracer studies to liquid systems, a method was developed for the determination of the solubility of halogenated compounds in liquids. Solubilities of SF 6 in water measured with this technique were in good agreement with literature values. The method was used to determine, for the first time, values of the solubility of SF 6 in a homologous series of n-alcohols. The method allows SF 6 to be used as a tracer in a number of unique liquid systems including the characterization of the interfacial exchange rate of gases across the pulmonary membrane of the lung. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: (Chemistry) Degree Grantor: California Institute of Technology Division: Chemistry and Chemical Engineering Major Option: Chemistry Thesis Availability: Public (worldwide access) Research Advisor(s): Shair, Fredrick H. Thesis Committee: Unknown, Unknown Defense Date: 11 October 1977 Funders: Funding Agency Grant Number California Air Resources Board UNSPECIFIED U.S. Environmental Protection Agency UNSPECIFIED Record Number: CaltechTHESIS:12022025-232421338 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:12022025-232421338 DOI: 10.7907/6djp-n724 ORCID: Author ORCID Lamb, Brian Kent 0000-0001-7641-5227 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 17781 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 10 Dec 2025 22:41 Last Modified: 10 Dec 2025 22:43 Thesis Files PDF - Final Version See Usage Policy. 102MB Repository Staff Only: item control page

DEVELOPMENT AND APPLICATION OF DUAL ATMOSPHERIC TRACER TECHNIQUES FOR THE CHARACTERIZATION OF POLLUTANT TRANSPORT AND DISPERSION Thesis by Brian Kent Lam~ In Partial Fu1fi11ment of the Requirements for the Degree of Doctor of Philosophy California Institute of Technology Pasadena, California 1978 (Submitted October 11, 1977) ii This thesis is dedicated to my parents and, with love, to Margaret. iii Acknowledgments I am very grateful for the experiences afforded me by my research advisor, Professor Fred Shair, during the course of this research. I thank him for trusting my abilities and for supporting me with his enthusiasm. It was extremely rewarding to me to share with him in the problems and the successes of this research projecto I wish to thank Dr. Peter Drivas for introducing me to the capabilities of the electron capture detector. I also thank Dr. Kevin Donohoe, Caye Johnson, and all of my office mates for their insight int~ problems , real and imaginary, which arose in the confines of our office. The expertise of George Griffith and John Yehle along with the capable, willing help of Ray Reed and Henry Smith provided the necessary instrumentation and equipment for each field study. Their ability and willingness to lend a hand during times of need are greatly appreciated. The patience of Sharon ViGario during the preparation and typing of pounds and pounds of reports is also appreciatedo I particularly thank the members of the St . Louis, Oxnard, and California Delta Field study teams for their enthusiasm and dedication. The success of each study was a direct result of their efficiency and hard worko A note of thanks is also given to the staffs of Chemical Engineering, the Purchasing Department, and Graphic Arts for all of their help. This work was supported by grants from the California Air Resources Board and the Uo S. Environmental Protection Agency . iv Abstract Environmental concerns over urban development and industrial growth have resulted in increased interest in atmospheric tracers. Expanding urban areas in coastal regions make it pa rticularl y important to understand the effects of complex meteorological patterns and rough terrain upon the transport and dispersion of atmospheric pollutants. The purpose of this thesis was to further develop single and double atmospheric tracer techniques which would be economical and efficient for large-scale field studies. Sulfur hexafluoride and bromotri- fluoromethane were selected as tracers since they are both inert gases detectable by electron capture gas chromatography in the parts per trillion range . A tracer system was developed which permitted the collection and analysis of as many as 1000 air samples per day. This system was then used to demonstrate the application of tracer techniques t? important problems of current interest in both coastal and noncoastal zones over distances as great as 100 km. Atmospheric tracer studies, conducted in Southern California during September, 1975, investigated the possible transport of pollutants between the Los Angeles Basin and the Oxnard Plain. The SF 6 data clearly showed that pollutant transport occurs from the Oxnard Plain along the Malibu coast into the Los Angeles Basin. A second series of tracer studies using SF 6 and CBrF 3 were completed within the California Delta Region during September, 1976. The purpose of these tests was to analyze the impact of possible industrial developments in the Montezuma Hills of the Delta Region. V The relative impact of existing and proposed industrial sources was investigated by simultaneously releasing CBrF 3 from the Montezuma Hills and SF 6 from upwind of the Delta Regiono Results indicated that plumes from both areas were carried southeast over Stockton into the San Joaquin Valley. Data from both of the coastal studies indicated that because of periods of steady coastal winds, the commonly used Hino correction grossly underestimated hourly averaged concentrations computed from 10-second averaged concentrationso Tracer concentrations predicted from the Gaussian plume model overestimated values observed in the Oxnard study because of increased dispersion over rough terrain. Predicted concentrations showed good agreement with observed values in the Delta study where the terrain was relatively flat. When sufficient data were available, surface air parcel trajectories determined from numerical solutions of the two-dimensional conservation equation were found to be more accurate than trajectories drawn from meteorological streamline maps. Five large-scale tracer studies utilizing SF 6 and CBrF 3 were performed in urban Sto Louis during August, 1975, to probe the relative impact of emissions from low-level and elevated sources. Releases of SF 6 were used to simulate elevated sources, while simultaneous releases of CBrF 3 were used to simulate low-level sources. The dispersion of emissions from low-level sources was initially enhanced in contrast to the dispersion of emissions from elevated sources. vi Analysis of the St. Louis t racer data resulted in the development of the concept of a limit model of air quality. By considering the relative impact of different sources and formulating necessary assumptions in terms of an upper or lower bound , the limit model approach offers a new perspective upon contemporary air quality problems. The application of this concept was demonstrated in the development of an upper limit model of the relative impact of elevated rural sources and low-level urban sources upon urban so 2 levels in St. Louis. Results of this application indicated that rural sources contribute no more than 25% of the average total urban so 2 concentration , even though rural sources can dominate urban air quality in portions of the urban region during relatively large portions of the time. As a means of applying experience gained from atmospheric tracer studies to liquid systems, a method was developed for the determination of the solubility of halogenated compounds in liquids. Solubilities of SF 6 in water measured with this technique were in good agreement with literature values. The method was used to determine, for the first time, values of the solubility of SF 6 in a homologous series of n-alcohols. The method allows SF 6 to be used as a tracer in a number of unique liquid systems including the characterization of the interfacial exchange rate of gases across the pulmonary membrane of the lung. vii Table of Contents Acknowledgments iii Abstract iv Table of Contents vii List of Figures ix List of Tables xiii lo The Transport and Dispersion of Atmospheric Pollutants 1.1 Introduction 1.2 Meteorology of Transport and Dispersion 1.3 Models of Atmospheric Diffusion 1.31 The Gaussian Plume Model 1.32 K Theory 1.33 Urban Diffusion Models 1.4 Application of Tracer Techniques 1 2. Experimental Procedure 2.1 Electron Capture Gas Chromatography 2.2 SF5 and CBrF3 as Atmospheric Tracers 2.3 Field Study Techniques 2.31 Tracer Release System 2.32 Sampling Systems 2.33 Analysis of Samples 2.4 Calculation of Plume Parameters from Crosswind Traverses 2.5 Mass Balance of Tracer Data 37 3. Atmospheric Dispersion and Transport Within Coastal Regions ~ Part I. Tracer Study of Power Plant Emissions from the Oxnard Plain 72 4. Atmospheric Dispersion and Transport Within Coastal Regions Part II. Tracer Study of Industrial Emissions in the California Delta Region 105 5. Tracer Studies for Characterizing the Transport and Dispersion of Plumes Emitted from Low-level and Elevated Sources 161 6. A Limit Model for Determining the Relative Impact of Rural and Urban Emissions Upon Urban Air Quality Part I. Development and Sensitivity Analysis of the Model 191 Part II. The Relative Impact of Rural Power Plants and Urban Sources Upon St. Louis S02 Concentrations 227 viii 7. Determination of Concentrations of Halogenated Compounds Dissolved in Various Liquids by Electron Capture Gas Chromatography 271 Literature Cited (Chapter's 1 and 2) 275 Abstract of Propositions 289 Propositions: Heats of Reaction for Oscillating Chemical Systems 290 Determination of the Solubility of Gases in Liquids at High Pressures 300 Determination of Trace Amounts of Atmospheric Sulfur Compounds 314 Experimental Characterization of Accidental Spills of Liquified Natural Gas 323 The Flux of Naturally Emitted Hydrocarbons into High Pressure Systems over the United States 332 ix hi st of Figures l. Pasquill-Gifford crosswind horizontal and vertical dispersion parameters as functions of downwind distance (from Turner, 1970). 17 2. Crosswind horizontal and vertical dispersion parameters as functions of downwind distance in terms of Pasquill stability classes as modified by Mc Elroy ( 1969). 21 3. Crosswind horizontal dispersion parameter determined from •the standard deviation of the wind direction (Pasquill, 1976). 23 4. Parameters for the calculation of vertical dispersion from Pasquill (1974) as presented by Fabrick et al. (1977). 25 5. Potential energy curves and temperature dependence for different electron capture mechanisms (Wentworth and Steelhammer, 1968). 44 6. Tracer Release System; Sequential 3-hour and 12-hour Air Samplers; and, Tracer Analysis System. 54 7. Typical automobile traverse SF 6 data and best-fit Gaussian curves. 56 8. Hourly averaged SF 6 data from fixed sampling locations. 57 9. Typical dual tracer calibration curves and dual tracer chromatogram. 60 10. SF6 concentration as a function of height above the surface from the California Delta tracer study . 68 .CHAPTER 3 1. Map and topographical overview of the Southern California coastal region. 77 2. Overview of automobile traverse tracer data collected between 1631 and 1721 PDT (9/21/75). 84 3. Crosswind profile of tracer plumes observed 28 km downwind of OBGS. 85 4. Hourly averaged SF 6 data as a function of time. 86 X 5. Hourly surface wind vectors, 1200 PDT, 9/21/75 (W. Goodin, 1977) 90 6. Surface forward air parcel trajectories. 91 7. Comparison of horizontal and vertical dispersion curves determined from 10-second averaged tracer concentrations with Pasquill-Gifford dispersion curves. 93 8. Centerline SF6 concentrations compared with centerline concentrations predicted from the Gaussian plume model. 96 CHAPTER 4 1. San Francisco Bay Area and the California Delta Region. 112 2. Prevailing west and northwest wind flow types, Smalley (1957). 115 3. Typical dual tracer calibration curves and typical dual tracer chromatogram. 123 4. Overview of automobile SF 6 data, Test 1, 8/31/76. 126 5. Overview of hourly averaged SF 6 data, Test 3, 9/5/76. 127 6. Crosswind hourly SF6 and CBrF 3 profiles measured along Highway 160 during Test 2, 9/2/76. 128 7. SF5 concentration as a function of distance along an airborne traverse taken at three altitudes. 130 8. Average plume centerline positions for plumes emitted from the Montezuma Hills. 132 9. Horizontal and vertical dispersion parameters calculated from automobile traverse tracer data obtained during the designated meteorological periods. 135 10. Centerline tracer concentrations compared with centerline concentrations predicted using the Gaussian plume model. 137 11. Crosswind tracer profiles compared with crosswind profiles predicted using the Gaussian plume model and either experimental California Delta dispersion terms or PasquillGifford dispersion terms. 139 12. 147 Hourly surface and upper air wind vectors calculated from data measured between 1700 and 1800 PDT. xi 130 Forward air parcel trajectories. 149 CHAPTER 5 1. Location of tracer release sites and hourly average air sampling stations during St. Louis tracer tests. 166 2o Automobile traverse SF6 data and best-fit Gaussian curves. 173 3o Overview of automobile traverse SF6 data collected between 0200 and 0300 CDT. 174 4. Overview of hourly average SF6 data collected between 0800 and 2000 CDT. 175 5. Sto Louis Test 3 auto traverse. 176 6. Crosswind horizontal and vertical dispersion parameters versus downwind distance in terms of source height compared to Pasquill-Gifford dispersion curves. 178 7. Crosswind horizontal and vertical dispersion parameters versus downwind distance in terms of source height compared to McElroy (1969) dispersion curves. 179 8. Crosswind horizontal and vertical dispersion parameters versus downwind distance in terms of bulk Richardson number compared to McElroy classeso 180 9. Comparison of observed and calculated SF6 data along crosswind automobile traverses. 184 CHAPTER 6 - Part I 1. Urban regi on of St . Louis and five surrounding rural power plants. 197 2. Sensitivity of percentage impact area to variations in values of the independent variables. 208 3. Determination of the rural tra ns port distance, Xmax, where the percentage impact area , A, is a maximumo 211 4o S02 emissions in St. Louis as a function of the di~tance from the edge of the urban zone. 216 5o Urban concentration of S02 as a function of downwind distance for the three fo rms of the urban emission distribution function. 217 xii CHAPTER 6 - Part I I 1. The urban region of St. Lou i s and five surrounding rural power plants. 232 2. Description of the portion of the urban area where >~0 which results from Baldwin emissions under D'n conditions. 238 3. Yearly averaged frequency of wind direction versus wind direction for the St . Louis region (Martin et al., 1967). 243 4. Percentage impact area as a function of frequency of occurrence during specified atmospheric stability conditions. 246 5. Percentage impact area as a function of frequency of occurrence for a range of specified concentration ratios. 248 6. Percentage impact area as a function of impact frequency by power plant. 250 7. Impact of rural power plants upon urban St. Louis in terms of fractional impact area , total fractional impact occurrence, and concentration ratio. 252 8. Percentage impact area as a function of wind direction for~~ 1. 255 ~ CHAPTER 7 1. Typical SF6 chromatogr am . 273 2. 274 Solubility of SF6 in water. 3. Solubility of SF6 in various solvents. 274 xiii List of Tables lo Summary of Gaussian Dispersion Parameters 18 2. Physical Constants of SF 6 and CBrF 3 47 CHAPTER 3 lo Hourly Averaged Dilution Factors, ·Predicted and Observed so 2 Concentrations 99 CHAPTER 4 1. Occurrence of Wind Patterns which Included Trajectories from Richmond through the Carquinez Strait into the Sacramento or San Joaquin Valley , 1952-1955 114 2. 120 Tracer Release Data CHAPTER 5 1. Tracer Release Information 169 2. Average Meteorological Data Observed During Tracer Release Periods 171 3. Best-fit Power Law Coefficients for St. Louis Tracer Dispersion Data 182 Comparison of Calculated and Measured Concentrations of SF 6 CHAPTER 6 - Part I 185 l. Emission Rates of S02 Associated with Urban and Rural Sources 198 2. Comparison of Relative Sensitivity of the Percentage Impact Area to the Input Variables 213 4. CHAPTER 6 - Part II 1. Emission Rates of so 2 Associated with Urban and Rural Sources 233 2. Percentage of the Urban Area where¢~ ¢0 239 3. Average S02 Concentrations and Concentration Ratios in the St. Louis Urban Region 259 1 ,. The Transport and Dispersion of Atmospheric Pollutants 1. 1 Introduction The transport and dispersion of atmospheric pollutants must be understood in order to describe and predict the effects of pollutant emissions upon our environmento The complexity of atmospheric turbulent processes and the diversity and number of pollutant sources often require tha unique labeling of a polluted air parcel in order to trace its path and measure its dilution in the atmosphere. Gases, such as SF 6 and CBrF 3 which are sensitive to detection by electron capture gas chromatography at levels as low as one part per trillion, provide an ideal means of quantitatively tagging the emissions from a particular source under complex meteorological conditions. The application of atmospheric tracer techniques yields an unambiguous method of correlating the transport and dispersion of airborne material with properties of the atmosphere. The purpose of this work was to further develop and integrate single and double atmospheric tracer techniques into an efficient and economical method for performing large-scale, detailed tracer field studies of timely air quality prob,lems. Large- scale in this case includes distances as great as 100 km while detailed refers to spatial resolution of plume cross sections on the order of 800 m, 50 km downwind of a release, and temporal resolution of plume behavior on time scales ranging from 10 minutes to several hours. ·The ability to obtain detailed tracer data on a large-scale was demonstrated by applying the field study techniques to the investigation 2 of the transport and dispersion of pollutants in complex coastal zones and noncoastal urban zones. In coastal areas, strong diurnal variations in complicated meteorological patterns as well as abrupt changes in terrain often require that analyses of the transport and dispersion be based upon experimental tracer observations . In both coastal and noncoastal urban areas , a problem of fundamental interest to regulatory agencies is the relative impact of different sizes and types of sources. Dual tracer techniques offer a unique means of measuring the relative impact of two sources under identical meteorological conditions. In the remainder of this chapter, pertinent descriptions of atmospheric processes and current atmospheric modeling practices are given. Techniques and results of several past tracer studies are also described. The second chapter includes a discussion of electron capture gas chromatography and the atmospheric and analytical chemistry of the tracerso Chapter 2 also outlines the experimental procedures and data reduction methods employed in the tracer field studies. In Chapter 3, results are presented from two large-scale atmospheric tracer studies conducted from the Oxnard Plain of Southern California during September, 1975. This study was motivated by controversy regarding the possible transport of pollutants between the Oxnard Plain and the Los Angeles Basin. This controversy arose over a proposal to form a Ventura Air Quality Management District separate from the South Coast Air Quality Management District. Because of the rough coastal terrain separating the two zones and the existence of complex land-sea breeze flow patterns, the Oxnard tracer study was 3 expected to provide experimental data describ ing the transport and dispersion of pollutants under these conditions. In a second coastal tracer investigation, described in Chapter 4, eight large-scale studies were performed utilizing CBrF 3 and/or SF 6 within the California Delta Region of Central California during September, 1976. The purpose of these tests was to characteri ze the impact of possible industrial developments in the Montezuma Hills of the Delta Region. The Delta Region is located in a major channel for marine air flow from the San Francisco Bay Area into the Sacramento and San Joaquin Valleys. The construction and operation of huge chemical complexes in the Montezuma Hills could cause significant degradation of the air quality in the inland, agricultural valleys. Dual tracer techniques were used in these tests to determine the relative impact of existing industrial sources upstream of the Delta Region and proposed industrial sources in the Montezuma Hills. This application of dual tracer techniques in the Delta Region appears to be the first, successful use of these methods in a large-scale field study. Chapter 5 reports the results of five extensive tracer studies conducted in St. Louis , Missouri during August, 1975. These investigations were designed to determine the relative impact of lowlevel and el evated sources on urban air quality. Regulation of fuel supplies and emission rates among different types and sizes of low- level urban sources and large urban and rural sources is dependent upon knowledge of the relative impact of these different sources. Sources of different heights were simulated by releasing tracers from towers 4 at heights as great as 305 mo St . Louis was t he site of the Lio S. Environmental Protection Agency's Regional Air Pollution Study (RAPS) which was an intensive effort to obtain comprehensive emissions, meteorological, and air quality data for use in the development and validation of large urban air quality models . The tracer studies complemented this effort by providing detailed data of the transport and dispersion of an inert, nonreactive species with accurately known emission rateso The tracer data from each of these field studies were reduced in such a manner as to facilitate the validation of diffusion modelso A brief review of the validation of a model using Sto Lou i s tracer data is given in Chapter 5. Similar use of t he Cal i fornia Delta tracer data is currently being carried out by personnel from the California Ai r Resources Boardo Because of the large amount of data collected during these field studies, extensive tabulations and descriptions of the tracer and meteorological data are not included i n this dissertation; a series of detailed reports are available elsewhere (Lamb et alo, 1975, 1977, Lamb and Shair , 1977)0 Consideration of the relative impact of low- level and elevated sources during the St . Louis field study l ed t o the i ntroduction of a lim i t model of air qualityo By definition , a limit model is an approach in which uncertainties in the analysis are treated so as to place upper or lower limits upon the results of a description of the relative impact of different emission sourceso These unique features simplify an impact analysis and give a different perspective upon the 5 relationship of the sources. Limit model expressions are derived so that the impact of one source is forced to be an upper bound while the impact of a second source is calculated to be a lower bound. If the results of this procedure indicate that the impact of the first source is less than that of the second, in spite of the upper and lower limit assumptions, then more detailed calculations will only confirm this description. In Chapter 6, the limit model concept is introduced. As an example, a particular limit model was developed to determine the relative impact of low-l evel urban sources and elevated rural sources upon urban so 2 levels in St. Louis. The results of the St. Louis li mit model were discussed in terms of the maximum percentages of time and urban area where the impact of rural sources is greater than that from urban sources. Chapter 7 is a demonstration of the versatility of electron capture gas chromatography. A method was developed for the determination of the solubilities of halogenated compounds in liquids based upon electron capture detection. The work is expected to form the basis for the development of tracers for liquid systems. Measured values of the solubility of SF 6 over a range of temperatures were compared to previously reported results. This technique was then used to determine, for the first time, the solubility of SF 6 in a homologous series of n-alcohols. These volatile solvents present formidable difficulties for typical volumetric-manometric methods of measuring solubilities. The analytical procedure allows the determination of solute concentrations over many orders of magnitude. 6 It has recently served as the basis for charac terizing the interfaci~l exchange rate of gases in the pulmonary membrane of the lung (Paciorek, 1977). 7 1.2 Meteorology of Transport and Dispersion Extensive discussions of the effects of meteorology upon the transport and dispersion of atmospheric pollutants have been given by Slade (1968), Seinfeld (1975), and Wanta and Lowry (1976). Generally these authors describe the structure of the atmosphere and the measurement of turbulent processes in terms of the atmosphe r ic energy balance, vertical temperature gradient, horizontal and vertical turbulence parameters, and vertical profiles of the wind. Aspects of these topics which apply to the interpretation of tracer data are discussed in this section. Among the manifestations of the sun ' s energy in the atmosphere is the development of an atmospheric vertical structure. A typical measure of this structure is the vertical temperature gradient. If a dry parcel of air rises 100 meters through the atmosphere and expands adiabatically, the resulting decrease in temperature will be approximately 1° C. If this dry adiabatic lapse rate of -1° C/100 m is identical to the existing temperature gradient, the atmosphere is considered to possess neutral stability. Since no buoyant forces act upon the parcel, it will remain at its new position . However, if the parcel becomes warmer than its new surroundings, buoyancy forces will cause the parcel to continue rising . Such an atmosphere is termed unstable and has a lapse rate which is smaller or more negative than the dry adiabatic lapse rate. If the rising adiabatic parcel becomes coo ler than its surroundings, buoyancy forces will cause the parcel to return to its Qriginal position. In this case, the atmosphere is termed stable and has a lapse rate which is larger or less negative than the 8 adiabatic lapse rate. rate is positive. An atmosphere is termed very stable if the lapse It is evident that as the lapse rate becomes less negative , the dispersive potential of the atmosphere decreases. The existence of a positive lapse rate, better known as a temperature inversion, is a particularly important condition since it effectively prohibits surface air from mixing aloft. Edinger (1969), in a general discussion of the effects of meteorology upon the transport and dispersion of atmospheric pollutants, described the formation and role of the inversion layer in the Los Angeles Basin . The Pacific anticyclone pos i tioned off the coast of California causes a persistent sunmer inversion to occur over the coast. by warming of the surface layer. This inversion is elevated The resulting shallow marine layer , when advected over an urban area, acts as a lid on the vertical transfer of pollutants . Miller (1968, 1973), Miller and Ahrens (1969) , and Lovil l and Miller (1968) examined the structure and effects of low-level inversion layers over the San Francisco Bay Area . They reported t hat oxidant l evels increased at al l levels from the surface to 2500 m immediately after the early morning breakdown of low-level inversions. During the evening, oxidant concentrations decreased much more rapidly below the inversion than it did above the inversion. Furthermore, well-defined wave patterns were observed in the inversion; it appeared that these waves sometimes break and allow moisture and pollutant s to be injected into the layer. Morgan and Bornstein (1977) used data from slow-rise radiosondes to characterize the inversion layer over San Jose, California, with respect to frequency of occurrence , elevation, 9 thickness, and intensityo According to their data, elevated inversions during midday hours occur as often as 90% of the time over San Jose during the summer and only slightly less frequently during the winter. Atmospheric stability can also be measured in terms of the vertical and horizontal fluctuations of the wind direction. This is particularly convenient since the standard deviations of the fluctuations can be determined directly from the traces of a recording anemometer and are related to the intensity of turbulence. As the stability of the atmosphere increases, the magnitude of the standard deviation of the wind direction, 0 8 , decreases. Berg and Sivertsen (1977) reported the development and application of an electronic digital recording anemometer which yields as many as eight variables associated with the changing nature of the wind speed and direction. Data obtained from a coastal site indicated that 0 8 was linearly correlated with the amplitude of the wind fluctuations . Furthermore, measured values of 8 were approximately equal to measured values of the intensity of turbulence, defined as 0 u/u where 0 u is the standard deviation of the 0 wind speed and u is the mean wind speedo Gifford (1968) categorized values of 0 8 , the horizontal standard deviation of the wind direction, in terms of atmospheric stability classes . These classes, designated classes A through F, were suggested by Pasquill (1961) as one means of determining stability solely from wind velocity and insolation considerations when direct measurements of turbulence were not available. Class A corresponds to an unstable atmosphere and class D represents neutral stab ility. Class Fis related 10 to very stable conditions. Turner (1961) presented a straight- forward classification system for determining Pasquill stab ility classes. It is of particular interest to describe the effects of the interrelationship between meteorology and topography upon the transport and dispersion of air pollutantso Defant (1951) has discussed the flow patterns associated with a number of geographical situations. In coastal regions, the temperature and pressure gradients which exist between air over land and over water cause convective mot ions of the air to occur . During the day when the land warms faster than the water , a sea breeze occurs from over the water towards the land . At night when the water cools more slowly than the land, the reverse occurs, and a land breeze is directed toward the sea. studied in a number of locales. This system has been Edinger (1959) modeled the depth of the marine layer over the Los Angeles Basin as a function of time of day and penetration inlando The marine layer over Los Angeles typically extends vertically to approximately 300 meters along the coast and as much as 600 meters farther inland . At the coast, the marine layer increases in depth early in the day and then decreases during the afternoon o At inland points , the marine layer reaches a maximum during midday o Fosberg and Schroeder (1966) described the characteristics of t he sea breeze in Centr al California noting that marine ai r is warmed considerably after transport inland . Frenzel (1962) found that during the summer in Central California the land breeze does not reverse the marine flow, but does reduce its strength. Sivertsen (1975) reported wind, temperature, and air quality data 11 measured in the Frierfjord of Norway during the land-sea breeze cycle. Because of industrial facilities located near the coast, levels of NOx increased and levels of ozone decreased as the sea breeze front passed various inland stations. Sivertsen concluded that moderate pressure gradients of 1-2 mb and upper level winds less than 5 m/sec were characteristic of land-sea breeze occurrences. These studies generally indicate that diurnal variations in the flow over coastal regions will transport pollutants inland during the day within a shallow, well-defined marine layer. Smalley (1957) analyzed wind patterns over the San Francisco Bay Area and determined that the Bay Area topography tended to channel the marine flow to a greater extent as the depth of the marine layer decreased. The diurnal variation of heat flux which drives the land-sea breeze system also causes diurnal variations to occur in the flow patterns associated with complex mountain-valley terrain. Edinger (1969) described the patterns associated with the San Gabriel Mountains of the Los Angeles Basin. During the day, the heated slopes of the mountains force the surface air to rise while drawing valley air up slope. This convective motion acts as a chimney for pollutants by injecting surface air into and above the inversion layer. At night, radiative cooling of the slopes cools the surface air and causes it to drain into the valley. This phenomenon was reported upon in early studies by Hewson (1945) and Holland (1953). More recently, Kao et al. (1974) conducted a similar study of mountain-valley flow in the Salt Lake Valley of Utah. Kao et al. concluded that dispersion of pollutants 12 would be greatest during the afternoon when valley winds were divergent flowing toward the mountains. Davidson (1963) and Ayer (1961) both reported observations on the erosion of nighttime drainage by early morning air flow above the ridges of the valley slopes. Lorenzen (1974) and Unger (1974) noted the frequent occurrence of mountain-valley winds associated with flow in the Sacramento and San Joaquin Valleys, respectivelyo Pasquill (1974) summarized a number of studies by noting that there is a lack of ventilation in deep valleys during stable nighttime conditions. A nighttime phenomenon, termed the urban heat island, is caused by different rates of cooling between an urban area and the surrounding countryside. Martin et al. (1967) pointed out that this causes the nighttime surface inversion to be lifted above the ground over the city . They described the heat island as a bubble of air with a neutral to isothermal lapse rate and a recirculating flow pattern centered over the urban areao Vukovich et al . (1976) developed a model of the urban heat island using the equations of motion and found that the structure of the heat island became markedly distorted by high winds . It appears that pollutants emitted into the heat island are prevented from mixing beyond the boundaries of the heat islando 13 lo3 Models of Atmospheric Diffusion Most atmospheric diffusion models can be categorized in one of three classes. Statistical models, based on the work of Taylor (1921), Sutton (1932), Cramer (1957), Pasquill (1961), (1974) , and Gifford (1960), (1961), among others, considers the history of the motions of particles and develops the statistical properties necessary to describe the dispersion. Richardson (1926) and Schmidt (1925) independently introduced the gradient transport concept of atmospheric diffusion. This approach, in analogy to molecular diffusion, assumes that atmospheric diffusion proceeds as a function of the concentration gradient . In assuming that turbulent diffusion is related to the concentration gradient, the concept of an eddy diffusivity , K, is introduced. For this reason, the gradient transport approach is commonly called K theory. A third category of diffusion models is based on consideration of dimensional arguments involving atmospheric parameters associated with the planetary surface layer . similarity theory. This method is generally termed Ellison (1959) and Batchelor (1959) first reported the use of dimensional arguments to determine statistical properties of the diffusion of material. Because of the widespread use of the Gaussian plume model which results from statistical theory and the atmospheric diffusion equation which results from K theory, the application of these models is discussed in the following sections. 1. 31 The Gaussian Plume Model The simplicity of the Gaussian plume model has lead to utilization of the model in a large number of situations. Turner (1970) writes 14 the Gaussian expression for a continuous point source as: ( 1) x(x,y,z;H) where x(x,y,z;H) is the concentration of airborne material observed at coordinates x,y,z resulting from emissions into a uniform wind field with velocity u from a source with effective height Hand release rate Q. The horizontal and vertical crosswind standard deviations of the distribution of dispersed material are given by cry and crz, respectively. The values of these statistical terms increase with distance downwind of the source. The use of Equation (l) is restricted in that turbulence is assumed to be stationary and homogeneous. Since this is seldom if ever the case in the atmosphere, rigorous application of the formula is severely limited. However, Batchelor (1949) theorized that because of the random nature of the atmosphere, the Gaussian plume model may be used to describe average plume dispersion . A great deal of effort has been expended in order to realistically specify values of the terms used in Equation (1). Effective source height is equal to the sum of the actual stack height and the height that a plume rises because of buoyancy forces acting on hot exhaust gases. Briggs (1969, 1972) developed a series of expressions which describe the plume rise under a variety of conditions. The value of u is typically taken from measurements at the surface or at the effective source height. In other cases, the value of u can be determined by averaging data over height or over the area of interest. 15 Realistic specification of the wind speed is particularly important because concentration is inversely proportional to wind speed i n t he Gaussian dispersion expression. Values of cry and oz are typically more difficult to specify accurately. Hay and Pasquill (1959) derived the relationship cry= cr 0 (s,T) where cr 0 (s,T) is calculated from wind direction fluctuations averaged over periods and sampled over interval T, They suggested that the sample periods should equal T/8 where Tis the transport time,x/u,and Sis the ratio of the Lagrangian to Eulerian time scales. The conversion from a Lagrangian to a Eulerian time scale is necessary since the dispersion of the particles is a Lagrangian phenomenon while measurement of the wind fluctuations is a Eulerian function. Results from Hay and Pasquill (1959) and Islitzer and Dumbauld (1963) indicate that the value of S ranges from approximately 1 to 10 with a nominal value of 4 suggested. Pasquill and Smith (1971) suggested that Bis inversely related to the intensity of turbulence, B = p/i, and further suggested that p = 0.44 appeared to be a reasonable value of the constant. Thus, the values of cry and, presumably, cr 2 , can be determined for a particular region by measuring horizontal and vertical fluctuations of the wind over appropriate time periods. MacCready et al. (1974) correlated horizontal and vertical standard deviations of fluctuations in wind velocity, av and ow, and fluctuations in wind direction, a 0, with the mean wind speed from measurements taken over rough terrain. Under very stable conditions, av (minimum) was independent of wind speed up to 5 m/sec. Under stable conditions values 16 of a0 were much greater than would have been predicted from the Pasqu i1 1 stability classes. MacCready et al. also reported that the strength of turbulence increases with increasing roughness of terrain. Corre- lations of av and aw with a terrain roughness parameter, atr' suggested that a = K ao::U'zY v l tr (2) where U is the wind speed, z is the measurement height , and K , ex: ,,, 1 and y are constants for a given stability class and must be determined empi ri ca lly. In cases where turbulence measurements are not available, Pasquill (1961) suggested the use of a series of empirical relationsh ips which give cry and az in terms of transport distance according to the stability of the atmosphere. Using dispersion data from flat, open country, he categorized dispersion with respect to wind speed and insolation in terms of the stability classes A through F. These data were modified by Gifford (1961) to yield the curves shown in Figure 1. Coefficients of best-fit lines of these data are tabulated in Table 1. Although the Pasquill-Gifford dispersion terms are widely used in predicting pollutant concentrations or as a standard for comparison of measured dispersion terms, the restricted data base from which they were derived requires care in their application . Turner (1970) reminds users that values of cry are more accurate than oz and that values of both terms become less accurate as distance increases beyond a few kilometer~ 17 (f) a: w w 1- L >I a: L L'.) ._.. (f) 10 3 I 04 105 X, METERS (f) a: w 1- w L • 10 3 N I a: L L'.) ._.. (f) 10 2 10 1 L-----1..----1.......L...JL.LL-1...LL---L---L-.L.L.J...J...L ,------'----'-_._._-'--'--'..Ll 1o2 104 1o3 5 1o X, METERS Figure l. Pasquill-Gifford crosswind horizontal and vertical dispersion parameters as functions of downwind distance (from Turner, 1970). 0.836 o. 716 1.43 Ri I > 0. 01 , 0 0 c 8 = 8 - 13 0.770 1.41 Ri' < - 0.01, a = 30°+ 8 Ri' < -0.01, a 8 = 24 0 -29 0 1. 36 0.900 _ 0.066 F very stable Ri' = ±0. 01, a 8 = 150 -20 0 0.901 0.097 E stable 1. 51 0.904 0.128 D neutral Turner (1970) Ri' < -0 . 0l , a = 18° -22° 8 -4.07 0.896 0.207 C slightly unstable from data in McElroy (1969) o. 572 0.878 0.348 unstable 0.667 0.671 0.689 0.483 0.122 0.050 0. 008 C -1.40 -3.11 -2 . 33 -8.46 B b (x in meters) 2.54'1 □- 4 7.63·1 □- 3 9.58·1 □- 2 -5.33·1 □- 2 0.700 -6.11•10 -2 0.657 0.924 1. 14 1.50 d -1.01·10 -3 -4.81·10- 3 -2.16·1 □- 2 1.17 d a = c x 2 -1.19·10- 3 5.56·10- 4 2. 79·10- 2 0.882 1.33 0 0 2.11 exp [a 0 + a 1(ln .x) + a 2(ln x) 2 + a (ln x) 3J 3 a2 a3 al ao (x in meters) = Gifford ( 1961); 2 0.869 a 0.510 axb A very unstable = Pasquill (1961); a ml-b y Classification Source a SUMMARY OF GAUSSIAN DISPERSION PARAMETERS TABLE l - ~-o Pasquill (1976) from Fabrick et al. (1977) where F(x) is given by: x(m) 100 200 400 1000 2000 4000 10,000 >10,000 k F(x) 0.8 0.7 0.65 0.60 0.50 0.40 0.33 0.33(10/x) 2 F(x) ~ 2.96 x- 0 • 253 (best fit of data) which yields: ay = 2.96 x0 • 747 a 8 For 0 in terms of Pasquill stability classes, ay = ax 0 • 747 8 Class a 8(deg) a(average) A > 23 1. 29 B 18-23 1,07 C 13-18 0.800 8-13 D 0.533 E 4- 8 0.296 < 4 F 0.193 For x > 20,000 m ay = (F(x) 2x2a 82 + .03 68 2x2)½ 68=total change in mean wind direction over vertical dept h of the plume ay = FW • X • a 8 0.656 o. 968 E'F' stable l 0.567 0.706 1.06 1. 22 d = a (x) • f(x) • g(x) lo a l = C Xd 6H = bouyant plume rise . 0 g'(x) = g(x)/g(x)l =3 cm· al (x) = al for neutral 0 stability, l 0 =10 cm (Fig. 4). f(x) = stability correction term (Fig. 4). g(x) = surface roughness correction term (Fig. 4). a l 0 (x) = a l from PasquillGifford (Turner,1970). or to revise PasquillGifford curves: al= [al 0 (x) 2·g'(x) 2+0.16H 2J½ a 0.867 0.837 0.736 0.843 neutral D' 0.111 o. 723 1. 29 slightly unstable C' 0.068 1. 36 unstable B' McElroy (1969) C 0.737 a = axb a y b 1-b ( . m x 1n meters Cl ass ifi cation Source SUMMARY OF GAUSSIAN DISPERSION PARAMETERS (CONT.) __, \.0 20 Singer and Smith (1953), (1966) repor t ed dispers ion data col l ec ted at Brookhaven National Laboratory in terms of gustiness cat egories designated B2, B1, C, and D which were based on correlations of dispersion and wind fluctuations. Carpenter et al. (1971) presented results of dispersion analyses conducted from Tennessee Valley Authority (TVA) power plants in terms of six stabi lity categories . These categories were identified in terms of the vertical temperature gradient for stabilities ranging from neutral to strong inversion . A review of the preceding data is given by Strom (1976). McElroy (1969) discussed the results of low- l evel dispersion tests over urban St. Louis. He correlated the results wi th a number of meteorological indices including cr8 and the bu l k Richardson number : Ri' = g[~T/62 + f]z 2/Tu 2 (3) where r is the adiabatic lapse rate, 6T is the temperature difference measured over ~z = z-z 0 , Tis the mean absolute temperature over 62, and u is measured at the upper height, z. McElroy also correlated the data i n terms of Pasquill stability classes. Best-fit lines of these data are given in Figure 2; the best-fit power law coefficients for both sets of correlations are given in Table 1. Data for dispersion over complex terrain have been obtained by Hinds (1970) in tests conducted at Vandenberg Ai r Force Base in California. His data are represented within a factor of 2 by the single expression: 0 cry= 0.65x • 75 . Hinds also noted that cry could be related to values of cr 0 smoothed (4) 21 1011 (/) a: w 1w :a: 102 10 1 L--------l.------'-----L-...........-L-L..L-4--------1-'---+--4-'-....._......_.______...__.........._~~ I o2 l o3 I o4 I as X, METERS 1011 (/) a: w Iw ::E . 103 N I CI ::E c...' ,_,J (/) 102 10 1 L.____j___JLJ...--1-LJ...U..L____j_L.....L---L.l...J....L.J..L____JL_L....J..-L.J.....U....U 102 1o3 1011 1as X, METERS Figure 2. Crosswind horizontal and vertical dispersion parameters as functions of downwind distance in terms of Pasquill stability classes as modified by McElroy (1969) . 22 with respect to travel time as suggested by Hay and Pasquill (1959). Hinds found that values of S ranged from approximately 8 to 60 for these data. Although empirical dispersion curves such as those developed by Pasquill, Gifford, and McElroy have received widespread use, renewed emphasis has recently been placed upon the application of the simple, physical relationship between cry and cr 0. As Fabrick et al. (1977) noted, Pasquill (1974, 1976) suggested the relationship: cry= F(x) • x • cr0 (5) in which F(x) is a correction term to cr 0 which ranges from 0.8 at 0.1 km to 0.33 at 10 km. The scheme shown in Table 1 was developed by Pasquil l for the Environmental Protection Agency as a revision to the methods summarized by Turner's Workbook of Atmospheric Dispersion Estimates (1970). In cases where cr 0 data are available, Equation (5) can be applied directly, irrespective of sampling time, surface roughness, and stability. It is suggested that for distances greater than 20 km, values of cry should be modified using the expression: cry= ( F(x) 2 • x2cr 2 + 0. 360 2x2 )½ 0 (6) where 60 is the total change in mean wind direction over the vertical depth of the plume. When measured values of cr 0 are not available, dispersion estimates can be made using the values of cr 0 corresponding to the Pasquill stability classes as specified by Gifford (1968) or Smith and Howard (1972). The horizontal dispersion curves thus obtained are shown in Figure 3; best-fit coefficients are given in Table 1. 23 104 (f) a: w fw L ... 10 3 >I a: L (.9 ~ (f) 102 A B C D E F 10 1 10 2 Figure 3. 10 3 X, METERS 10 4 105 Crosswind horizontal dispersion parameter as a function of downwind distance determined from the standard deviation of the wind direction in terms of Pasquill atmospheric stability classes (Pasquill, 1976) 0 24 Fabrick et al. (1977) reviewed a different approach to the determination of the extent of vertical dispersion introduced by Pasquill (1974) which takes into account the effects of surface roughness and atmospheric stability: (x) • f(x) • g(x) (7) zo is the vertical standard deviation corresponding to neutral cr 2 = cr where cr zo stability and a surface roughness equal to 10 cm, f(x) and g(x) are correction terms to account for differences in atmospheric stability and surface roughness, respectively. in Figure 4. Curves for these terms are shown Smith and Howard (1972) listed typical values of the surface roughness, z0 , for different types of terrain. In keeping with this new approach of estimating vertical dispersion, Pasquill suggested that the vertical dispersion curves given in Turner's workbook should be modified using: cr 2 = ( cr;(x) 2 g'(x) 2 + O.ltH 2 )½ (8) where cr z0 is the workbook value, g 1 (x) equals the correction term, g(x) normalized by the value of g(x) corresponding to a surface roughness equal to 3 cm. Equation (8) also incorporates the dispersion associated with buoyant plume rise, tH. Pasquill (1976) also suggested a different procedure for estimating vertical dispersion over urban areas. This method specifies that cr z be one-half of one stability class more unstable than that normally predicted to account for the increased stability caused by the urban heat island. the following example of this approach: cr z (D stability,urban) = Fabrick et al. (1977) gave (9) v& z (D stability,rural) • cr z(C stability,rural). 25 >< (km) 0 .1 10 100 uzo FOR z0 = 10 cm. NEUTRAL STABILITY 7 1000 0.2 a: UJ tUJ ~ <l a: <l Cl. >t:.::i iii <l t- Vl Cl. B 5 04 C 0.6 4 0 0.8 20 1.0 (cml 1.2 3 1.6 E x (km) 10 100 RECIPRICAL OF STABILITY CORRECTION FACTOR I 1 (KI 0 .1 -~_._............_ .........I.++'_-'-..............,. 100 1 x (km) 10 0.1 SURFACE ROUGHNESS CORRECTION FACTOR g (>< I Figure 4o Parameters for the calculation of vertical dispersion from Pasquill (1974) as presented by Fabrick et al. (1977). 26 A corresponding procedure for calculating cry over urban areas was not presented. In addition to the flexibility provided by the availability of a series of different dispersion parameters, a diverse number of forms of the Gaussian expression have been derived. Ragland (1976) obtained expressions for worst-case concentrations by taking the derivations of Equation (1) with respect to downwind distance and wind speed. He compared worst-case concentrations calculated with three different sets of dispersion terms and found that the different sets of terms yielded widely different results. Fosberg et al. (1976) included a mass divergence term in the Gaussian plume model and found that it caused concentrations to decrease by factors of two in areas of large divergence such as over mountain-valley terraino Several methods have been developed to account for the presence of an inversion layero An equation from Bierly and Hewson (1962) can be written (10) 2 N L 2 exp [ -'21 ( H+z+2nl) ] X = 2ncrQcr u exp[-½(:) ] az y z y n=-N where the summation terms take into account multiple reflections of the plume from the surface and from an inversion layer at height L. Turner (1970) approximated the Bierly-Hewson formula by assuming that cr = 0.8L and the concentration is vertically well-mixed beyond a 2 27 distance 2xl where xl is the distance at which 0 2 = 0.47 (L-H) . Yamartino (1977) has shown that the series in Equation (10) can be accurately approximated by an expression involving a Jacobi theta function. An extensive evaluation of the Gaussian plume model applied to power plant emissions has been reported by Fabrick et al. (1977). These authors validated the Gaussian plume model using data from tracer studies by Start et al. (1974) , Drivas and Shair (1975), and Giroux et alo (1974). Generally, predicted concentrations were within a factor of 2 to 3 of observed values in complex terrain, and within a factor of 3 to 4 of observed values in coastal terrain . The authors attributed the increase in error associated with coastal terrain to the changing nature of the wind$ in coastal regions. l . 32 ' K Theory The basic formulation of K theory is called the semi-empirical turbulent diffusion equation. Johnson et al. (1976) described its deri- vation beginning with the conservation of mass equation in one dimension: ac a a2c at= - ax (uC) + D Tx ( 11 ) where C is the instantaneous concentration of a species in a wind field with velocity u and Dis the diffusivity. The concentration and wind speed can be written in terms of a mean and the turbulent fluctuation about the mean: C = C + C' - ( 12) u = ~ + u' ( 13) 28 which when substituted in Equation 01) and averaged yields u. -~ 2ac- __ a(uc) _ a(u'C') + 0 at ax ax . ax ( 14) The second term on the right side is taken to represent turbulent diffusion and is assumed to equal u'c"' = -K 2.£ ax where K is the eddy diffusivity. (15) In the atmosphere, turbulent diffusion is assumed to be much greater than molecular diffusion. The mass conservation equation then becomes: ac - a(~C) + ...£... (Kac) at - - ax ax ax ( 16) or in its usual three dimensional form: where the overbars have been dropped. ( 17) In this form, the so-called atmospheric diffusion equation includes a source term, S, and a chemica l reaction term, R. Application of the equation involves determining values of the eddy diffusivities and formulating appropriate boundary conditions. In some simple cases, the equation can be solved analytically; Seinfeld (1975) has reviewed a number of solutions to the equation. 29 For a uniform wind velocity, u, in the x direction, v and w equal to zero, diffusion in the x direction assumed to be much less than the mean transport, and KY= K2 = constant, Equation (17) can be solved analytically to yield the Gaussian plume model expression (Equation (1)) where ay = a 2 = 12Kx/u. ( 18) In more complex situations the equation must be solved numerically. Bauer (1974) surveyed efforts to determine experimental values of eddy diffusivities. He found that the horizontal eddy diffusivity varies with the width of a dispersing cloud according to K = 0 1.13 y y over the range 100 m < ay < 100 kmo ( 19) (20) ' Fabrick et al. (1977) in their study of the Gaussian point source model also presented a detailed evaluation of a point source grid model based upon the atmospheric diffusion equationo They reviewed several different methods of specifying values of the eddy diffusivities in addition to the relation given by Equation (18) . The validation of the atmospheric diffusion equation as formulated by Fabrick et al. yielded predicted concentrations which were within a factor of 2 of observed concentrationso Fabrick et al. concluded that the K theory grid model was superior to the Gaussian point source model in terms of precision, realism and generality. However, they also noted that because of the cost involved in applying the grid model, both models were needed for cost-effective air quality modeling. 30 Drivas and Shair (1974) compared predictions of the Gaussian plume model and K theory for a quasi-instantaneous line source with tracer data. They found that the appropriately formulated K theory gave more accurate predictions of the data than did the Gaussian plume model. Stukel et al. (1975) solved the diffusion equation with a particulate settling term to yield an analytical Gaussian expression which was employed to formulate a model of vehicle emissions from a series of streets and highways. The authors validated the model using measurements of lead concentrations in Illinois and concluded that the model provided accurate predictions of particulate lead concentrations. Ragland and Dennis (1975) solved the diffusion equation numerically for an elevated point source with variable wind and diffusivity profiles and compared the results with concentrations predicted with the Gaussian plume model. Under neutral conditions, good agreement was found for concentrations at ground level. However, considerable differences were found in concentrations predicted at elevated levels. 1.33 Urban Diffusion Models Because of the nature of our largely urban society, application of the Gaussian plume model, the atmospheric diffusion expression, and other models to the problem of predicting and understanding urban air pollution has been widely explored. The Proceedings of Symposium .2D_ Multiple-Source Urban Diffusion Models (1970) includes discussion 31 concerning many of the problems associated with urban modeling . Calder (1977) has analyzed the general structure of multiple-source urban plume models, w1th particular emphasis upon the use of the Gaussian plume expression. Calder emphasizes that these models share an assumption of quasi-steady state , usually over a one-hour period, and that concentrations predicted from multiple sources result from superposition of the concentration fields caused by single sources . Calder noted that the latter is true only for nonreactive species . Koch and Thayer (1971, 1975) and Koch and Fisher (1973) in a series of reports have validated a multiple-source Gaussian plume model with respect to air quality measurements in St. Louis, Chicago, and New York City. They also evaluated a simple urban dispersion model developed by Gifford and Hanna (1971 , 1973) and Hanna (1971) which gives the concentration x with the expression: ( 21 ) X = IQ_ u where Q is the urban emission rate, u is the mean wind speed, and C is a constant. For the case of so 2, C is suggested to equal approximately 50. The evaluation indicated that the use of variable emission rate data improved the prediction of short-term concentrations by the Gaussian multiple- source model and that the Gi fford- Hanna model was improved by including large point source contributions calculated with the Gaussian plume model. For long-term mean so 2 concentrations, the various forms of the Gaussian plume model and the Gifford-Hanna model gave similar validation statistics. On the average, short-term concentrations predicted with the multiple-source Gaussian diffusion model were considerably larger than measured values. 32 Shir and Shieh (1974) used K theory, incl uding a chemical reaction term, to predict so 2 concentrations in St. Louis for comparison with 25 days of. data measured during February, 1965. They assumed that KY was constant, while K2 was set equal to a function of height in which the effects of surface roughness were taken into account. The correlation coefficients between computed and measured concentrations for 2-hour and 24-hour data were 0.899 and 0.813, respectively. Reynolds et al. (1973) formulated an urban model using K theory for application in Los Angeles which included a detailed chemical mechanism describing the production of photochemical smog. Again, the horizontal eddy diffusivity was assumed constant, while the vertical eddy diffusivity was determined from an algorithm involving surface components of the wind and the height above the surface. Validation of the model with air quality data measu red in Los Angeles gave qualitative and, in many cases, good quantitative agreement with the data. 33 lo4 Application of Tracer Techniques A common denominator .in all reports concerning the application of atmospheric dispersion models is a statement to the effect that modeling results should be validated using experimental observations. Furthermore, the development of atmospheric diffusion models requires dispersion data in order to specify values of either cry and cr z or Ky and Kz. As noted in the Introduction, a relatively simple, straight- forward means of obtaining the required experimental data is through the application of tracer techniques. Drivas (1974) reported an extensive list of the kinds of tracer techniques that have been utilized in the past. Neiburger (1955) measured the transport of fluorescent particles to test the accuracy of surface wind trajectories in the Los Angeles Basin. Sulfur dioxide was released in the Prairie Creek experiments from which the Pasquill-Gifford dispersion terms were developed (Barad, 1959). Islitzer (1961) released uranine dye as an aerosol to relate dispersion to fluctuations of the wind direction. Singer and Smith (1966) measured the dispersion of oil fog to determine the dispersion data used to formulate the gustiness categories given in Table 1. Angell et al. (1966) tracked tetroons out to sea at night and then followed their return to the coast the next day. In later studies, Angell et al. (1972, 1976) tracked tetroons over the Los Angeles Basin to determine transport paths caused by the local landsea breeze system. In general, tetroons tended to move rapidly inland during the afternoon and slowly seaward during the night. McElroy (1969) measured the concentrations of fluorescent particles to 34 determine dispersion data for urban St. Louis. His results indicated that urban dispersion is initially enhanced by the urban topography and the urban dispersion rates are greater than rates observed in open country. However, at distances where the eddies become the size of typical urban obstructions, dispersion approaches that measured in rural areas. Sandberg et al. (1970) used fluorescent particles as a tracer in the Bay Area of San Francisco. Initial enhancement of plume size, discontinuous dispersion rates between land and water, and marked topographical effects upon dispersion were observed. Raynor et al. (1974) released oil fog over the ocean and concluded that under stable conditions dispersion rates were much less over water than those observed over land. Land and water dispersion rates were similar under unstable conditions. Hinds (1970) released zinc sulfide through insecticidal foggers and analyzed samples by a scintillation counting technique. Hovind et al. (1974) compared results of fluorescent particle and oil fog releases over complex terrain with predictions using the Gaussian plume model. Their analysis suggested that centerline concentrations measured over rough terrain can be as much as 10 times smaller than predicted. Start et al. (1974) used SF 6 and oil fog to measure the dilution of effluents over rough terrain. They concluded that enhanced turbulence and thus, greater dispersion occurs during flow over mountain tops, across canyons, and in wakes behind terrain irregularities. Giroux et al. (1974) tracked SF 6 released from a power plant on the Oxnard Plain of Southern California as far as 40 km. Drivas and Shair (1974) followed a quasi-instantaneous puff of 35 SF 6 from Anaheim , California, through five neighboring co1TD11unities to Palm Springs, 124 km downwind . Drivas and Shair (1975) more recently reported results of large-scale SF 6 tracer tests conducted from power plants located at Moss Landing near Monterey Bay, California, and at Long Beach, California . The tracer plume was detected as far as 80 km downwind during these tests. Cadle et al. (1976) studied automotive sulfate dispersion by performing a freeway simulation test using new model vehicles equipped with catalytic converters. from moving vehicles. A portion of the test involved SF 6 releases Data collected downwind of the test track indicated that the EPA HIWAY model (Zimmerman and Thompson, 1975) grossly overestimated the impact of vehicular emissions close to a highway (Chock, 1976). Palmer (1977) released SF 6 from a moving vehicle on Southern California freeways and collected samples in trailing vehicles. The results indicated that vehicle-induced turbulence is far more important in the near field than wind speed or stability conditions . In cases where an individual point source i s fa r removed from other sources , the plume can be tracked by sampling for pollutants such as so 2. Vaughan et al . (1975) used correlation-spectrometer data to measure so 2 flow rates. Deposition rates as great as 40% over 43 km were measured for plumes from a power plant , while deposition rates as large as 30% over 23 km were measured for plumes from a low- level source . Vaughan et al. estimated so 2 deposition velocities in the range 3 to 6 cm/sec. 36 Efforts to improve the use of halogenated tracers have been reported by Dietz and Cote (1973} and by Orgill et al. (1974) . Through the use of a frontal elution technique . they were able to obtain a continuous recording of tracer concentrations in a plume for as long as 60 seconds. Simmonds et al. (1976) have recently reported the development of a continuous analyzer capable of measuring SF 6 concentrations as low as 3 • 10-14 parts SF 6 per part air . Thomsen and Lovelock (1976) used this analyzer to determine instantaneous crosswind concentration profiles of SF 6 released from a power plant stack. 37 2. Experimental Procedure 2.1 Electron Capture Gas Chromatography Since its introduction in 1960 by Lovelock and Lipsky (1960), electron capture gas chromatography by virtue of its extreme sensitivity has served as the analytical cornerstone in building arguments against the spread of halogenated pesticides in our environment and the distribution of fluorochlorocarbons in the atmosphere (for example, see Juret and Cram, 1974; McNeil -·et al., 1977; Bowman and Nony, 1977; Molina and Rowland, 1974; Crutzen, 1974; Hester et al., 1974, 1975; and Lawrence and Ryan, 1977). The sensitivity of electron capture detection to halogenated compounds (as low as ,o- 12 p/p by volume) also provides the basis for the use of certain halogenated compounds as atmospheric tracers over distances as great as 100 km. The principle of detection is simple. As a carrier gas flows through the detector, S- particles, emitted from a low-energy radioactive source, ionize carrier gas molecules to produce a distribution of thermal electrons. The electrons are periodically swept to the anode to produce a standing current. A~ electron absorbing material is introduced into the detector through a chromatographic column, the electron concentration and, in turn, the standing current decreases. The signal is inverted electronically so that the change in current appears as a peak on a continuous recording of the signal. The area under the peak in proportional to the amount of electron absorbing material eluted from the column. The coaxial detector used in the work described herein is typical of many electron capture gas chromatographs; other widely used designs, discussed by David (1974) 38 in a review of chromatographic detectors , include the parallel plate and concentric electron capture detectors. Prepurified N2 or the mixture 5% CH4-Ar is generally used as the carrier gas. The typical radioactive source consists of 200 mCi of tritium bonded to a titanium substrate deposited on a copper or stainless steel foil . Because of H3 emanation, this source is limited to operation below 225°C (Shoemake et al., 1963). Yauger et al. (1966) reported on a Ni 63 source that can be operated continuously at temperatures as high as 300°C. Use of high temperatures eliminates absorption of contaminants upon the source and increases its lifetime. Fenimore et al. (1971) tested a high temperature (~300°C) H3 source which used scandium as a substrate. An H3 source is generally preferred over Ni 63 on the basis of lower energy of the B particle (18 KeV compared to 67 KeV) and, hence, less noise in the signal. In addition to these three possibilities, other sources , including Te 99 , Ra 226 , Am 241 , sr 90 , Kr 85 , and Pm147 have been examined (Shoemake et al. , 1963) . Most recently , Dwight et al . (1976) suggested the use of Fe 55 as a source of S- particles. Wentworth et al. (1975) , in an effort to avoid the temperature restrictions associated with radioactive sources , designed and tested an electron capture detector equipped with an argon resonance lamp as an extermal source of electron- producing photons. Although this design does extend the operating temperature range of the electron capture detector, the complexity of construction and operation and the slightly smaller degree of sensitivity leaves its future usefulness in doubt. 39 Taylor (1962) reported guidelines for the safe use of the radioactive foils, and indicated that a tritium loss rate of 5 µCi/day at room temperature is not unusual . Since tritium is particularly harmful when inhaled, proper ventilation of a laboratory is especially important. Taylor noted that ambient levels of tritium as tritium oxide should not exceed 5 • 10- 7 µCi/cm 3 and, as molecular tritium, levels should not exceed 2 • ,o- 4 µCi/cm 3. In a laboratory with volume 150 m3 and ventilation rate equal to 15 m3/min, a radioactive source with a tritium oxide loss rate greater than 7.5 µCi/min will cause the steady state room concentration to exceed the recommended safety level. In the work presented herein, as many as 16 gas chromatographs were present in the laboratory at one time. The estimated combined loss rate from these gas chromatographs is approximately two orders of magnitude less than that calculated to yield harmful levels. Nevertheless, it appears very worthwhile to vent the exhaust from each gas chromatograph to a fume hood. Handling of the radioactive foils should always be done using forceps and gloves in a fume hood . Taylor noted that foils may need to be cleaned periodically to remove surface deposits. He strongly advised that foils should not be swabbed; cleaning should be accomplished only by rinsing in appropriate solvents. Pellizzari (1974) and David (1974) reviewed the development of the electron capture detector and summarized the current state of knowledge concerning the method. The electron capture detector can be operated in a d.c. mode in which electrons are continuously collect~d at the anode or in a pulsed mode in which electrons are periodically collected at the anode. Lovelock (1963) pointed out a number of 40 anomalous responses generated by d.•c. operation and introduced the pulse mode as a means of eliminating the anomalous behavior as well as improving the sensitivity. In 1971, Maggs et al. (1971) introduced a new mode of operation in which the standing current was held constant and the pulsing frequency was varied in response to the presence of an electron absorbing material. This method increases the linear range of the detector and reportedly increases the sensitivity. The mechanisms which are important in electron capture detection are not well understood. Simmonds et al. (1967) reported that the standing current of the detector increased with temperature. and Guiochon (1969) reported just the opposite. Devaux Van de Wiel and Tommassen (1972) carefully cleaned the carrier gas, removing traces of water and oxygen, and found that the standing current was independent of temperature. Apparently, when trace amounts of oxygen are present in the carrier gas, both oxygen and water take part in electron capturing processes which are dependent upon temperature .. Van de Wiel and Tommassen also found that the standing current was independent of carrier gas flow rateo Scolnick (1969), Donohoe (1975), and Kapila and Aue (1976) studied the effects of pressure upon the response of the electron capture detector. As pressure increases, the applied voltage necessary to collect the increased number of electrons also increases. More importantly, it appears that the response of the detector to a compound increases dramatically. Kapila and Aue suggested that the minimum detectable amount also increases with pressure. Donohoe varied the radius of the anode and found that use 41 of smaller /radii increased the standing current , but also appeared to increase the baseline noise. An increase in the standing current would increase the linear range of the detector, while an increase in baseline noise would limit the minimum detectable level of a compound. Many of the preceding authors caution that results obtained with a particular detector may not occur in another detector. However, a number of operating characteristics in the pulsed mode seem generally true: 1) the standing current, I0 , increases with voltage, V, until Vis large enough that all electrons are collected during a pulse; 2) 10 increases with the width of the applied voltage pulse, tw, until the pulse width is sufficient to allow collection of all electrons; 3) detector sensitivity increases with pulse period, tp' until the natural recombination rate of positive ions and electrons becomes important. Values of these parameters sufficient to yield optimal operating conditions are typically V = 20v, tw = 2µsec, and tp = 200µsec Typical flow rates equal 50 cm 3/min. Wentworth and co-workers {1962), (1966), (1967), (1969) have reported a number of studies of the physical and chemical mechanisms involved in electron capture detection. Two reviews of their findings have been given by Wentworth and Steelhammer (1968) and Wentworth (1971), respectively. Wentworth and co-workers proposed four possible reaction mechanisms to account for the response characteristics of the detector: kl (I) k_l 42 kl 2 A· + B- e + AB k2 AB-- ( II) A· + B- (III) A- + B• (IV) For long pulse periods, the electrons are assumed to be at thermal energies with a steady state concentration. The rate expressions for these mechanisms can be solved simultaneously to give b - [e-] = Ka (22) where band [e-] are the concentrations of electrons in the absence and presence of a capturing species, respectivelyo The instantaneous concentration of the capturing species in the detector is given by 11 a 11 , while the electron capture coefficient, K, is a function of the rate constants: (23) where k0 is the overall rate constant for reactions removing e and kl is the overall rate constant for reactions removing AB-. The linear relationship between concentration and the electron concentration (or standing current) function given by Equation (22) has been validated experimentally for a large number of compounds. The value of K for a particular compound under a particular set of conditions is determined from the strip chart record of the signal produced by a sample of known concentration. Since 11 a' 1 is the instantaneous concentration, both sides of Equation (22) must be integrated with respect to volume in order to relate the known sample concentration to the signal. 43 The temperature dependence of the electron capture response to a particular compound can be used to predict the dominant capturing mechanism at a given temperatureo Wentworth and Chen (1967) considered the relative magnitude of the rate constants at various temperatures to derive the temperature dependent expressions given in Figure 5 for the four mechanisms. A large number of compounds have been investigated which appear to fit one or another of the four mechanisms (Wentworth et al., 1968, 1975). Pellizari (1974) noted that the sensitivity of the detector to a compound was dependent upon the electron concentration, the electron capture cross-section, the time allowed for encounter between electron and capturing species, and the magnitude of the electron capture coefficient. Since the value of K can depend strongly upon temperature, the temperature of the detector can be optimized for a particular compound. Similarly, the time allowed for encounter between reactants can be optimized for a particular compound. Unlimited increase in the pulse period is restricted by the rate at which natural recombination of positive species and electrons takes place . The descriptions and theories of electron capture detection presented in the cited works can be summarized in terms of the analysis of tracer compoundso All traces of oxygen, water, and other contaminants should be removed from the carrier gas to prevent the production of interfering species. The magnitude of the collecting potential and the width of the potential pulse should be sufficient to collect all electrons; the standing current will reach a maximum Figure 5. k0 (k-1•kL) LOWT ~ l_ 1./T ' ,- s ' ~ k1.2. HIGHT K= " ~ ko IJ ? ~ ,I.E. k.k, k~(kJ ,, , SLOPE.=-(+). EA R- ACE.TAT E. SLOPE=(-}:£. " = Al =(OA 8 -EAA) I Cl Bf\,I AROMATICS 1/"T ______ / SLOPE=(-), E• ,, ~,..._ ~ s ,, ~,',,\Jl kAAl k 0 (k_ 1 +k 2 +kJ INTIBMEDIA:TE, LOW T kl ( ki + kJ (:):\ HIGH T K= .N Potential energy curves and temperature dependence for different electron capture mechanisms (from Wentworth and Steelhammer, 1968). SLOPE:(-), E.*o<.AE. c1 , ef\, r, AUPHAT1c5 1/T • LOWT (' J ~ .._,,, --~ s~3~~, ' ' SLOPE.:(+) =E.A AROMATIC. H'YDROCARBONS AROMATIC CARBON'YLS -F, -CH3, -CF3, - (X}\3 ,-C=N K5l,, - - - : - ; - : : ; - - - ,, ko (k, li<-, ,, ~ k1.CkJ HIGHT I n-\r- : K= ~ ( k.,) I -~ _r:,,, 45 under these conditions. The response of the detector to a compound should be maximized with respect to pulse period, temperature, and pressure. Finally, safe operating practices should always accompany enthusiastic experimenting. Recent developments in electron capture gas chromatography suggest that diversity of applicatiqn will ensure a secure future for the technique. Lovelock (1975) suggested that the sensitivity of the electron capture detector could be greatly improved by filtering noise from the detector signal using coherent modulation of the solute concentration. Simmonds, A. J. Lovelock, and J. E. Lovelock (1976) introduced a continuous tracer analyzer which used the solute signal modulation method. Sensitivity as low as 3. 10 -14 p/p for SF 6 was reported. The device chemically eliminates ambient oxygen and allows only tracer compounds into the detector. In other developments, Lillian and Singh (1974) obtained coulometric responses (every sample molecule captures an electron) by using two electron capture detectors in series. In the case of coulometric response, the sample concentration can be calculated directly from the detector signal. Willmott and Dolphin (1974) coupled the electron capture detector with a liquid chromatograph for use in the analysis of pesticides. Chamberlain and Marlow (1977) have reported further developments of this combination. A number of studies (Siegel and Fite, 1976; Siegel and McKeown, 1976) have been reported in which the electron capture detector was employed as an atmospheric pressure ionization source coupled to a mass spectrometer. The ion spectra thus obtained provide insight into the capture process. 46 2.2 ?¼) and CBrF 3 as Atmospheric Tracers Sulfur hexafluoride and bromotrifluoromethane are both colorless , odorless, nontoxic, and inert gases which can be detected at extremely low levels using electron capture gas chromatography. Pertinent physical data for SF 6 and CBrF 3 are given in Table 2 (Matheson Gas Data Book, 1966). Halasz and Glemsar (1971) in a review of SF 6 chemistry, state that "normal nuc l eophil i c reagents cannot attack the highly symmetric octahedral structure of six equal S-F bond distances, 1.56 ~. 11 Case and Nyman (1962) suggested that a strong electrophile could coordinate with a fluorine atom and found that the reaction: SF 6 + 2S0 3 ~ 3S0 2F 2 proceeded at 20% efficiency at 250°C over 24 hours. (24) At high temperatures, SF 6 also reacts with alkali metals (Roberts, 1968). However, in the atmosphere, it appears that SF 6 can be considered as an inert, long-lived species. It is one of the least water soluble substances known as demonstrated in Chapter 7. Neither SF 6 nor CBrF 3 occurs naturally; SF 6 is generally manufactured by burning sulfur in the presence of fluorine. Bromotrifluoromethane is produced by thermal bromination of fluoroform or by brominolysis of perfluoropropane . Krey et al. (1976) reported that the estimated world-wide SF 6 production rate increased from 15 megagrams/year in 1953 to 1180 megagrams/year in 1974. CBrF 3 production rates are available. No estimates of The amount of SF 6 released during the tracer studies reported herein equaled 0.2% of the 1974 * Matheson Gas Data Book, 1966 146.07 g 310 p.s.i.g. 2.5 ft3/lb Molecular Weight Vapor Pressure@ 70°F Specific Volume, 70°F, 1 atm. Boiling Point@ 1 atm. Sublimation Temperature Freezing Point@ 1 atm. Density, Gas 70°F, 1 atm. Density, Gas@ b.p. Critical Temperature Critical Pressure Critical Density Solubility in Water, 25°C, 1 atm. 45.55°C 37.11 atm. 0.730 g/cm 3 0.004% (by weight) 6.139 g/1 -82.8°F (-63.8°C) ~ Pro~ PHYSICAL CONSTANTS OF SF 6 AND CBrF 3* TABLE 2 67°c 39.1 atm. 0.745 g/cm 3 0.03% (by weight) 8. 71 g/1 -270.4°F (-168°C) 148.93 g 190 p. s . i. g. 2.6 ft3/lb -72.0°F (-57.8°C) CBrF 3 ......, .$:::, 48 estimated emission rate. The total amount of CBrF 3 released was equal to one-third the amount of SF 6 tracer released . Krey et al . measured tropospheric and stratospheric ambient levels and estimated the total atmospheric SF 6 inventory to equal 985 megagrams. Fur- thermore, Krey et al., considering loss through photolysis the only atmospheric sink, estimated the photolytic half-life to range from 1 to 3 years with a total atmospheric half-life ranging from 12 to 24 years. These authors reported that ratios of SF 6 levels to CC1 3F levels increase markedly with increasing altitudes or polar latitudes indicating that SF 6 is much more stable to photochemical decomposition than CC1 3F. Short-term studies by Saltzman et al. (1966) suggested that SF 6 and CBrF 3 are stable under ultraviolet radiation. Doucet et al. (1975) measured the photoelectron and far ultraviolet spectra of CBrF 3; they reported strong bands occurring at 120 nm and 80 nm. No indication was given concerning the dissociation products of CBrF 3 under uv radiation. Neckers (1967) in a review, indicated that CBrF 3 decomposes under photolysis to give CF or CBr radicals. details of the measurements were obtainable. However , no Noutary (1968) presented photoionization data which indicated that CBrF 3 decomposed to CF 3+ and Br- ions under photolysis at 106 nm. In comparison, decomposition of CC1 F and CC1 2F2 takes pl ace at wavelengths less than 230 nm (Cicerone 3 et a1., 1974). The principal use of SF 6 is as an electrical insulation medium in switching-gear and transformers. De Bartoli et al. (1976) reported 13 ambient SF levels over Oslo, Norway to be approximately 4 • 10- p/p. 6 49 This level is typical of most urban areas. The principal use of CBrF 3 is as a fire extinguishing agent. Ambient CBrF levels appear to be 3 11 less than the minimum detectable level (1 . 10- p/p). Sulfur hexafluoride is a nontoxic gas as evidenced by a number of studies using SF 6 to study ventilation rates of the lung (Wood et al., 1969; Martin et al., 1972; Paciorek, 1977). Lester and Greenberg (1950) exposed animals to an atmosphere consisting of 80% SF and 20% o2 6 for periods as long as 24 hours with no indication of irritation or intoxication observed. Bromotrifluoromethane, like most freons, is considered a nontoxic gas. However, several freons, including CBrF 3, when inhaled in sufficient quantity, sensitize the heart so that the presence of epinephrine (adrenaline) causes cardiac arrhythmias to occur (1971). Trochimowicz (1975) termed CBrF 3 as a relatively weak sensitizer which produced arrhythmias in the hearts of beagles at 75,000 ppm. A bulletin from the E. I. du Pont de Nemours Company (1971) characterized CBrF 3 as ''the safest fire extinguishing agent available." Levels of 5% to 6% may be inhaled for 4 to 5 minutes without risk of serious cardiac or central nervous system effects. The gas is considered to have an acute toxicity less than CO 2. However, at temperatures above 500°C , CBrF 3 decomposes into toxic materials including HF, HBr, and Br 2. Under the circumstances involved in tracer releases, CBrF 3 can be considered as a nontoxic gas. The chemistry of SF 6 within the electron capture detector at room temperature appears to be that of Mechanism I as described in the 50 previous section (Wentworth and Steelhammer, 1968). The stable, long- lived ion, SF 6-, has been the subject of a number of studies (Thynne et al., 1973; Odom et al., 1975; Sides et al., 1976). Hickam and Fox (1956) first measured the electron capture cross-section of SF6 and reported it equal to approximately 10- 15 cm 2 eVo Spence and Schulz (1973) found that the total electron capture cross-section of SF 6 was independent of temperature over the range 200°K to 1400°K. Spence and Schulz also noted that the kind of negative ion produced from SF was 6 strongly dependent upon the temperature. Christophorou et al. (1971) found that production of SF 6- was much greater than that of SF -+ Fat 5 room temperatures. The production of SF 5-+ Fat higher temperatures indicated that SF 6 also undergoes a Mechanism II dissociation. Spence and Schulz (1973) reported that the capture cross-section of CBrF 3 2 16 16 2 increased from 2 • ,o- cm eV at 300°K to 6 • ,o- cm eV at 1000°K. Wentworth and Chen (1967) classified halogenated aliphatic compounds as Mechanism II materials. However, apparently very little is known about the dissociation rates and products associated with the electron capture mechanisms of CBrF 3. Clemons and Altshuller (1966) measured the responses of a large number of halogenated compounds to electron capture detection. According to their study, the response to SF 6 was approximately 15 times larger than that of CBrF 3. The works cited above as well as those given by Drivas (1974) indicate that SF 6 is an inert, nontoxic gas detectable at extremely low levels and, thus, perfectly suited for an atmospheric tracer. 51 Similarly, CBrF 3 appears to be an inert, nontoxic gas detectable at low levels and also, well-suited for use as an atmospheric tracer. 52 2.3 Field Study Techniques The purpose and design of each field study was specified by the problem at hand and the region of interest . In every case, the experimental goal was the collection and rapid analysis of a sufficient number of air samples to yield an accurate description of the transport and dispersion of the tracer plume. The key factor in achieving this goal was the ability to obtain detailed spatial and temporal resolution of a tracer plume in the presence of changing atmospheric conditions . Experience proved that detailed descriptions of a plume's behavior could be drawn from a combination of grab samples collected during automobile and airborne traverses, and hourly averaged samples collected from a network of automatic stationary samplerso These sampling systems complemented each other by providing quasi-instantaneous 11 snapshots 11 and long-term averaged profiles of the tracer plume. Rapid analysis of traverse samples during the initial stages of a tracer release allowed midstream corrections to be made in subsequent surface and airborne traverses. Analysis of all samples within a day of a tracer test introduced a great deal of flexibility into the design of a field study; tracer data from one test could be tabulated and graphed before a second test was begun . Although the experimental methods are described separately in the following sections, each procedure was developed as an integral part of the field study program. Together, these techniques represent an efficient and economical means of attaining the experimental goal of a tracer field study. 53 2.31 Tracer Release System For the St. Louis and California Delta field studies, the tracer gases were released simultaneously from six cylinders into a copper manifold. The manifold was connected to a two-stage gas regulator (Matheson Model BH-590) which, in turn, was connected to a largevolume calibrated flowmeter (F. & P. Co. Model B6-35-10/27). release system is shown in Figure 6. The Personnel from Meteorology Research, Inc. conducted the Oxnard Plain tracer release using a similar system; tracer was injected into the stack at approximately ground level. In St. Louis low-level releases and in all California Delta releases, gas from the flowmeter flowed through½ inch copper tubing to the appropriate release height. High-l evel St. Louis releases were conducted from the 1000 foot KETC-TV tower; tracer was released from cylinders at ground level into 1 inch garden hose which was secured along the tower to the top. The release system was set up in a truck to provide mobility between release sites. In every case, tracer was released continuously at a constant rate; every release was monitored continuously. The average release rate was determined from the weight difference in the gas cylinders before and after the release. The average release rate thus determined agreed within 2% with the reading on the calibrated flowmeter. The cost of the SF 6 and CBrF tracer gases was approximately $6.60/kg ($3.00/lb) and $12.10/kg 3 ($5.50/lb), respectively. Typical release rates were 10 g/sec over periods ranging from 2 to 14 hours. 54 TRACER RELEASE SYSTEM AUTOMATIC 3- HOUR 12- HOUR AIR AND SAMPLERS TRACER ANALYSIS SYSTEM FIGURE 6. 55 2.32 Sampling Systems Air samples were collected in 30 cm 3 plastic disposable syringes (Becton-Dickinson Co). Detailed, quasi-instantaneous descriptions of the tracer plume were obtained with 10-second averaged samples collected at even intervals during automobile or airborne crosswind traverses. Figure 7 illustrates the type of data obtained from automobile or airborne traverses downwind of a tracer release. Intervals along automobile traverses were determined from the odometer; airborne intervals were calculated from sampling time intervals and the ground speed of the aircraft. Tests indicated that samples collected in the syringes did not change in concentration more than 5% over approximately one to two weeks. Hourly averaged samples were collected at fixed points using either sequential 12-hour samplers (115 v, AC) or sequential 3-hour samplers (6 v, DC). Small diameter needles were used on the syringes to prevent back-diffusion of the sample air. in Figure 6. The samplers are shown Typical hourly averaged data are shown in Figure 8. Drivas and Shair (1975) found that the reproducibility and accuracy of samples collected with the 12-hour samplers and analyzed immediately were excellent, within+ 1%. For samples analyzed after being outdoors in the sampler for 19.5 hours, the reproducibility and accuracy dropped slightly to+ 4%. 56 TEST CAL3 TRAVERSE 6 ORTE• 9/05/76 TIME• 0259-0306 TRAVERSE RCIUTE• NCIRTli ON HWY 160 STARTING PCIINT• NCIRTH SIDE OF SAN JOAQUIN RIVER. ON HWY 160 0 0 0 N "" 1-o CLo (LO ' - / (X) 5 7 DISTANCE N. ~F HWY 4~HWY 160 JUNC. (KM.) 9 TEST CAL3 TRAVERSE 8 ORTE• 9/05/76 TIME• CN1ij-05ij5 TRAVERSE RCIUTE• EAST ON HWY q, NCJRTtt CIN RT J3. HEST CIN HWY 12 STARTING PCIINT• JUNG. BYROO RO ANO HWY ij 0 0 :::J' "" 1(L CL '-/ 0 0 .N u z 0 u 25 50 DISTANCE RL~NG TRAVERSE CKM.) 75 TEST CAL3 TRAVERSE 9 DATE• 9/05/76 TIME• Qll30-Qijij5 TRAVERSE RCIUTE• NCIRTH ON HWY' 160 0 0 0 STARTING-PCIINT• (X) JUNG. HWY q ANO HWY 160 1(L (L '-../0 u 0 0 • :::J' z 0 u 5 10 DISTANCE N. QF HWY 4-HWY 160 JUNC. (KM.) Figure 7q 15 Typical automobile traverse SFG data and best-fit Gaussian curvesq Traverses were taken (from top to bottom) 10 km, 45 km, and 10 km downwind of the Montezuma Hills tracer release site. 57 1 12 2 12 a z n • .,,.,, n n . . -0 -0 -t 11 TIME a______ 23 5 12 11 12 TIME 6 -1 o______ 11 TIME 7 n 0 z n • .,,.,, .,, -0 -t -t TIME 11 TIME 23 9 12 n 0 zn 12 n 0 10 • • -0 -0 -0 -t -t TIME • .,, -0 -1 O'--------' 23 11 12 TIME 23 12 n 0 z n . . -0 -0 -0 -t 11 Figure Bo TIME 11 -1 23 Q.___ 23 Q.___ 12 n Cl z n 11 12 TIME .,, 0'----11 11 23 n 0 z n zn .,, -1 Q.___ ___._ ___, o~----- 8 12 . -0 01-.....1----11 TIME 23 23 n 0 z n . .,, -1 12 z n • .,,.,, 23 g n zn 0 z n • .,, .,, Q.___ _ ___. 0 0 z 4 12 n n zn 0 -t 3 12 ·nME 0-..-..... 23 11 TIME 23 Hourly averaged SF6 data from fixed sampling locations. California Delta tracer study, Test l, 8/31/76. Tracer released from the Montezuma Hills between 1200 and 1700 PDT . 58 2.33 Analysis of Samples Air samples were analyzed using electron capture gas chromatography. As many as 12 gas chromatographs were prepared for a field study. Each gas chromatograph was constructed by personnel from the Instrument Shop in the Chemical Engineering Department, California Institute of Technology. A stainless steel coaxial electron capture detector, electrically insulated with teflon and nylon plugs, was pulsed periodically with a 25 volt, 1 µsec wide pulse. The radioactive source used was a 200 mCi H3 source bonded to a titanium substrate (U.S. Radium Corp., Parsippany, New Jersey). Analysis for SF 6 was achieved using a stainless steel column (39 11 x 0.25" OD, 0.18 11 ID) packed with 5 A 80-100 mesh Molecular Sieve (Alltech Assoc., Arlington Hts., Illinois). Columns were filled with Molecular Sieve and lightly vibrated before being coiled. The columns were conditioned at 300°C overnight with N2 flowing continuously. Using prepurified N2 at 60 cc/min as a carrier gas, SF 6 eluted in 18 seconds and o2 in 45 seconds. In the St. Louis tracer studies, samples were analyzed for CBrF 3 using a stainless steel (2" x 0.25 11 OD, 0.18' ID) precolumn containing 5 A 80-100 mesh Molecular Sieve and a stainless steel 129" x 0.125" OD, 0.101" ID) column packed with 80-100 Porapak Q. The CBrF 3 tracer eluted in 90 seconds using N2 carrier gas at a flow rate of 30 cc/min. The detection limit of CBrF 3 under these conditions was approximately 1000 ppt. 59 Further work on the analysis of CBrF 3 fol l owing helpful suggestions from Dickson (1976) yielded columns in which SF 6 eluted in 18 seconds, CBrF 3 eluted in approximately 35 seconds, and o2 eluted in 40 seconds. The dual tracer column is identical to the Molecular Sieve column previously described for the separation of SF 6. However, the column must be packed under pressure by partially filling the column with Molecular Sieve, flowing N at approximately 2 10 psig, and vibrating the column for several minutes. This procedure is repeated until the column is full. Small amounts of glass wool are used to plug the ends of the columns. Dickson (1976) indicated that the dual tracer column must be conditioned at 175°C for exactly 12.5 hours under a continuous flow of N2. However, it appears that columns conditioned overnight at 300°C also separate SF 6 and CBrF 3. The dual tracer column extended the limit of detection for CBrF 3 to approximately 10 ppt. A typical chromatogram is shown in Figure 9. It also appears that pressure-packed Molecular Sieve columns can be used to separate as many as five halogenated compounds including SF 6 , CBrF 3, CC1 2F2, CC1 3F, and CBr 2F2 (Dickson, 1976). Because the separation efficiency is greater with the dual tracer columns, several halogenated solvents sometimes present in an urban atmosphere elute approximately 5 to 10 minutes following injection of the sample. Although these compounds do not directly interfere with analysis of the tracers, their presence can slow the analysis procedure considerably. on the SF columns. 6 These compounds are much less of a problem For this reason, a number of gas chromatographs 60 zg 1O-s CBrF3 SF6 CSF5) 10-9 (CBAF 3. 10-10 I I I 60 30 0 T i me (sec) 10-11 10-12 1o-13 .._.......... 10 1 102 J.J..l..~....:...O......&.J..~..___............L.L.1,.u.u,___,__...,_,_~...__..L...J....,_._..U..U..__._...,_,_........... 103 HJ" . 105 INTEGAAT(jR AREA 107 106 BKL 8/21/76 Figure 9. Typical dual tracer calibration curves and dual tracer chromatogram (inset} [SF5] = 6 ppt, [CBrF3] = 310 ppt. 61 were fitted with SF 6 columns and a number were fitted with SF 6-CBrF 3 columnso All of the gas chromatographs were equipped with 6-port sampling valves (Valeo, Inc., Houston, Texas)o The SF gas 6 3 chromatographs were equipped with 1.0 cm sampling loops; the SF 6CBrF 3 gas chromatographs used 2.0 cm 3 sampling loops. The pulse period used with the SF 6 instruments was 200 µsec, while the period used with the SF 6-CBrF 3 instruments equaled 400 µsec. These values represent optimal periods for the detection of SF 6 and CBrF 3, respectively. All columns and detectors were used at room temperature·. In the time that one analysis could be performed on each of four gas chromatographs, the signal on the first unit had returned to baseline and the procedure could be repeatedo With four gas chromatographs, as many as 60 samples could be analyzed by a single worker per hour. The signals from the 4 instruments were input into an electronic digital integrator (Spectra-Physics, Santa Clara, California) which automatically converted tracer peak area to concentration. The proportionality constant between peak area and concentration, termed the calibration factor (KF), was determined using an exponential dilution calibration method. For a well-mixed vessel, the concentration, C, decreases according to: C = C e-qt/V (25) 0 where C0 is the initial concentration, q is the constant flow rate, Vis the vessel volume, and tis the time since flow began. At any given time, the number of air c.tianges Nin the chamber since t=0, is qt/V. If the chamber is perfectly mixed and flow is steady, a plot 62 of lnC versus N will yield a slope of -1. Using a 611 x 611 x 611 lucite 3 cube (V = 3516 cm ) equipped with a magnetically driven fan and flowing clean, dry air through the cube at 120 cm 3/min typically yielded slopes within~ 0.01 of the prescribed value. A microliter syringe, accurate to approximately~ 1%, was used to inject 1.0 µl of SF and/or 6 3.0 µl of CBrF 3 into the cube. This method produced calibration samples ranging from approximately 1 ppm to 1 ppt. Samples were drawn from the cube exhaust line directly into the sample valve of each gas chromatograph. According to a standard error analysis (Bevington, 1969), errors associated with the calculated calibration concentrations ranged from less than 3% at high concentrations to less than 7% near the detection limit. Typical calibration curves obtained with the dilution method are shown in Figure 9. This procedure allows calibration of a gas chromatograph over six orders of magnitude of the concentration. The curves become nonlinear at high concentrations because the detector becomes saturated with sample at those levels. In some cases, the curves also become nonlinear near the detection limit. This results from the desorption of tracer from the walls of the cube. Gentle heating of the cube walls and constant purging with clean air prior to a calibration generally eliminates this problem. Donohoe (1975) reported that the degree of absorption and desorption of a number of freons was decreased considerably in a cube lined with Tedlar. A potentially serious problem associated with prolonged use of the gas chromatographs is contamination of the radioactive foil by depositon of eluted contaminates. As the foil becomes contaminated, 63 the detector operating characteristics change. The concentrations of samples analyzed under these conditions can be in error as much as 15% to 25%. One means of monitoring changes in the detector responses is to cross-check samples among the gas chromatographs. In cases of drastic changes in response, the cross-check results can be used to determine a new calibration factor in agreement with the results from the remaining gas chromatographs. Typically, field tests were designed so that a tracer release was performed every other day. during each release. As many as 1200 samples were collected Using an analysis system consisting of 8 gas chromatographs and 2 digital integrators as shown in Figure 6, two workers could analyze all of the samples collected during one tracer release before another release was conducted. 64 2.4 Calculation of Plume Parameters from Crosswind Traverses Plumes are often modeled by assuming they have a gaussian shape; that is, the concentration along a crosswind traverse follows an equation: C(y) = C0 exp [-½ (y-:0 ) 2 (26) i where Y0 is the distance coordinate of the center of the plume, C0 is the concentration at the center of the plume, and o is the standard deviation of the plume. From standard data analysis techniques, equations relating o and Y0 to the data obtained (C(y) and y) are: loo oo yC{y)dy y = 0 Ioooo (27) C(y)dy and (28) A value for C0 can also be calculated once Y0 and o have been calculated: J 00 00 C(y )dy = C cr/2TI ( 29) 0 or, rearranging terms, 10000 C(y)dy (30) C = ----o These parameters (C , Y , o) when used in Equation (26) represent a 0 0 best-fit of the data (C(y), y) to the equation. Typical results of this analysis are shown in Figure 7. Values of o were corrected 65 for the angle of the traverse path with respect to the wind direction as follows: o=osina. (31) where a is the specified angle. Three major sources of error appear in the actual use of the equations : one is the error inherent in the data themselves; another is due to the data being for discrete points rather than for ally; and finally, due to limitation on the sampling locations, one or both edges of the plume might be chopped off (the integration cannot be carried out to the limits, - 00 , 00 ) . The errors in the data cannot be reduced once the data are taken, but errors in application of the above fonnulas can be estimated and reduced to some extento The error involved in calculating the integrals is dependent upon the method used, but the error involved in chopping off the edges of the plume can be treated generally and is considered first . By assuming a perfect gaussian plume and calculating the parameters from equations (27)-(30), but with limits of integration Ya and Yb instead of - and 00 + 00 , . the fo 11 owing expressions for error of the results can be found: ' y.~ = y O - (32) (33) 66 where Y-Y· 0 ) u = ((J ( 34) and P - u e-½r2 l (u) - l2i" L dr (35) (note P(u) is the normal probability function) -oo where cr and Y0 are the actual parameters of the plume , and cr and Y0 1 1 are the parameters calculated using formulas (27)-(30) with limits of integration Ya and Yb. No simple expressions for Y0 and cr in terms of Y0 ' and cr' could be found; however, Equations (32) and (33) can be appl ied in an iterative manner so that successively better approximations of Y0 and cr can be found. i --'2U =y . cr n ·o(o) - 2 ub(n) e ub ua( ) l27r P (n) (u) ua(n) (36) n 1 -½u21ub(n) --'2U 2 lub( n) ) 2] -½ _ue ua ( ) _ _1 ( e ua(A) (37) ub n 2rr p ub(n) l2'rr P (n) (u) ua (u) ua (n) U n(n) (n) = ( YY - Yo ( n) ) a (38 ) n where the small subscript in parentheses refers to the number of times the iteration was performed to arrive at that approximation, (o) r'e fers to initially calculated values. Traverse data are usually taken at 67 even intervals along the traverse which simplifies the integration considerably. Simpson's method was selected for the integration. In cases where this best-fit procedure is applied to data obtained during airborne vertical spirals, errors can arise for two additional reasons. While it is assumed that the data represent a vertical profile at a constant crosswind distance from the plume centerline, in reality~ the aircraft generally spirals around a radius of several hundred meters. In some cases samples taken only a short vertical distance apart can in fact be taken in widely different regions of a plume. The error associated with this sampling pattern is a function of the diameter of the sampling spiral relative to the dimensions of the plume. At short distances downwind, the scale of the spiral and the plume may be relatively similar. be large. In such a case the error will At far downwind distances, although the relative scales may not introduce error, plumes are generally vertically well-mixed. The best-fit procedure is of little value under these circumstances. For low-level rele~ses, the vertical profile can also be changed considerably by reflection of material at the surface. In this case, calculated values of cr 2 are not related to a true Gaussian profile. Thus, it appears that vertical sampling spirals are of limited value in the determination of cr 2 • These sampling methods do provide qualitative insight into the vertical profiles of dispersing material. Typical results of this data collection procedure are shown in Figure 10. An alternate means of determining values of cr2 provides a larger set of vertical dispersion data than otherwise might be available. If one assumes that the Gaussian plume model can be used - I LU ___.___ __. 800 C~NC. ( PPT) ij00 1200 0 80 C~NC. ( PPT) L!O 120 <D - a I LU 0 ~ 200 lJ00 600 - ~ ---' C~NC. CPPT) a " -- U') aCr) TIHE• 7 SPIRAL 9/13n6 1008-1011 Figure 10. SF concentration as a function of height above the surface from the CaYifornia Delta tracer study. Spirals were taken approximately 21 km downwind of the tracer release site. 0 a ....____._____..____. a U') aCr) LU I aCr) U') ~ ~ ~ DATE• mT TEST U') - i ~ ~ ~ - '-" - '"' ~ OfN5-ClN9 1 '-" - A - -_ SPIRAL '"' <D 7 9/13n6 ~ . _ _ __ - a TI Hf• ORTE• '-" 1 '"' ~ <D a SPfRAL mT TEST 6 0>T TEST ORTE• 9/10n6 TIHE• 1158-1203 U') U') 3 °' 0'1 69 to describe the tracer results, and that tracer is conserved during transport, then a value of cr2 can be calculated in an iterative manner using the crosswind integral of horizontal crosswind traverse data: exp[-½ ( z+H 2•J + exp[-½ ( z-H ) 2 ) °z ] 02 (39) The value of cr2 thus obtained represents an effective cr 2 in cases where the mixing layer inhibits vertical dispersion. Since the crosswind integral of the automobile traverse data is calculated in the procedure to find cry, it is relatively simple to also calculate cr 2 • Values of the crosswind integral should be determined from crosswind traverses taken at the effective stack height when possible. 70 2.5 Mass Balance of Tracer Data The integration of the automobile and airborne traverse data over the crosswind traverse distance can be used to determine the average flux of tracer gas passing through the traverse vertical plane. This flux can then be compared to the amount of tracer released at the source to provide a mass balance on the experimental data: % tracer observed= f ( 00 l l u(z)C(y,z)dzdy) 00 0 • 100 (40) Q where Q is the release rate, Lis the height of the mixing layer, u(z) is the wind velocity, and C(y,z) is the tracer concentration. The accuracy of this mass balance is obviously subject to the uncertainties involved in determining the vertical profiles of the wind speed and tracer concentration. This expression may be simplified by considering a constant average wind velocity, u(z) = u, and a uniform vertical tracer distribution, C(z,y) = C(y), over a height l. For convenience in treating the data, the double integral is approximated by a series of single integrals: % tracer observed = u n 2 li ,-l (1 00 - 00 - ) 100 C(y)dy i • -Q- (41) n where L li equals the height of the mixing layer. For cases where i only surface tracer data are available, n is taken to equal 1, and it is assumed that the tracer is vertically well-mixed from the surface to the height of the mixing layer, l=L. Where traverse data are 71 available at more than one height, n is taken to be the number of traverses and li is taken to be a height arbitrarily assigned to each crosswind traverse. In this case, the tracer is assumed to be vertically well-mixed over a discrete portion of the mixing layer. 72 CHAPTER 3. (Submitted for publication in Atmospheric Environment) ATMOSPHERIC DISPERSION AND TRANSPORT WITHIN COASTAL REGIONS PART I. TRACER STUDY OF POWER PLANT EMISSIONS FROM THE OXNARD PLAIN by Brian K. Lamb, Arndt Lorenzen (California State Air Resources Board), and Fredrick H. Shair * Division of Chemistry and Chemical Engineering California Institute of Technology Pasadena, California 91125 August, 1977 * To whom all correspondence should be addressed. 73 Abstract Atmospheric tracer studies were conducted using the Ormond Beach Generating Station in Ventura County during September 21 and 22, 1975 . The purpose of this work was to probe the transport and dispersion of atmospheric pollutants between Ventura and Los Angeles counties . The area of interest is characterized by rough terrain and complex , coastal meteorological conditions. The sulfur hexafluoride tracer data clearly show that pollutant transport occurs from the Oxnard Plain along the Malibu coast across the Lennox area into the San Fernando Valley of the Los Angeles Basin, and along an inland route through the Thousand Oaks area into the San Fernando Valley as far east as Reseda. These tracer data were found to be consistent with air parcel trajectories constructed from available meteorological data. Pollutants emitted from the Ormond Beach Generating Station, under the test conditions, were determined to be diluted by approximately 10 5 upon reaching the San Fernando Valley and upon reaching the Los Angeles Basin . Calculations using the Gaussian plume model yielded a close fit to the tracer data when values of the dispersion terms cry and cr 2 were taken from the experimental results . Close agreement between the tracer data and the model predictions was also obtained using values of cry and cr 2 from the Pasqui.11 -Gifford atmospheric stability class which was one class less stable than that suggested to exist by wind speed and insolation measurements. 74 1. Introduction The meteorology and topography of urban coastal regions generally differ considerably from the features of inland urban areas. The transport and dispersion of pollutants over coastal terrain is usually more difficult to characterize than that over inland regions. In various coastal regions of California, meteorology and topography play a dominant role in the development of serious air pollution problems. Atmospheric tracer studies provide a unique and relatively inexpensive means of characterizing the transport and dispersion of pollutants over rough coastal terrain under complex meteorological conditions. The purpose of this series of papers is to describe the results of four large-scale atmospheric tracer studies conducted from the Oxnard Plain of Southern California, from the Delta Region of Central California, from the Moss Landing area near Monterey Bay, and from the Long Beach area of the Los Angeles Basin, respectively. The results of these studies provide intimate descriptions of transport paths and dispersion processes in each area. These tests offer insight into unique features of each region and into similarities among the regions as they affect pollutant transport and dispersion. The data obtained from these studies form a comprehensive data base for use in the development and validation of large-scale atmospheric diffusion models. In Part I of this series, the results of two tracer releases from the Ormond Beach Generating Station (OBGS) on the Oxnard Plain are presented. The results of eight tracer tests conducted in the Montezuma Hills of the Delta Region are described in Part II. Finally, data from 75 three tracer releases from the Moss Landing power plant and six tracer releases from Long Beach power plants are presented in Part III of this series. 76 2. Meteorology and Topography The Oxnard Plain of Southern California is a flat coastal region bounded north and east by inland mountain-valley terrain and west and south by the Pacific Ocean. The rough Santa Monica Mountains rise as high as 950 m along the coast to the southeast. A map of the region is shown in Figure 1. The meteorology of this area is dominated by a diurnal land-sea breeze pattern with strong on-shore winds occurring most of the day. Typically, a shallow (300 meters) mixing layer of marine air exists over the plain during the day (Giroux et al., 1974). Edinger (1959) found that the thickness of the marine layer over the Los Angeles Basin increases from approximately 300 m along the coast to approximately 640 mat El Monte, 40 km inland. DeMarrais et al. (1965) described the typical surface flow patterns in the area for different months of the year during different times of the day. In July from 1200-1800 PST, typical westerly marine flow diverges around the Santa Monica Mountains with the southern component moving east along the coast across the Santa Monica area and then northeast into the San Fernando Valley. The northern component moves to the northeast across Thousand Oaks into the San Fernando Valley. Angell et al. (1966), using results from tetroon releases in the Ventura area, suggested that 11 the heating of the interior may cause this area (Oxnard Plain) to act like a gigantic vacuum cleaner, sucking in air of a diversified origin. 11 At night a land breeze develops and, depending upon its strength, either completely reverses the direction of flow or reduces the strength of the westerly marine flow . The patterns from DeMarrais et al . suggest that 0 ~~-"·- ' - • . ~- " '-.J . F;gure 1, Map and topograph;ca) overv;ew of the Southern Ca);forn;a coastal reg;on. Contour l;nes are ;n 1000 ft ;nterva)s . Tracer sampJ;ng stations. wind data Overview grid is in 2 mile squares; vert;ca) scalestat;onsA.. has been exaggerated. "' "''Coo '-- SC4LE -·- ' -.! '-J 78 the land breeze dominates nighttime flow in October throughout the area, but during July the land breeze does not develop fully throughout the region. These patterns indicate that pollutants emitted along the Oxnard coast during the day will be transported either inland across the Simi Hills into the San Fernando Valley or along the coast across Santa Monica into the center of the Los Angeles Basin. This latter path may also involve trajectories into the San Fernando Valley. Angell et al. (1972) found from tetroon studies in the Los Angeles Basin that, during the evening, air from central Los Angeles frequently moved northwestward across the Santa Monica Mountains into the San Fernando Valley. Edinger and Helvey (1961) found that convergence of Oxnard and Los Angeles marine air in the San Fernando Valley causes rising motions to occur over a strip, estimated to be less than a kilometer in width, that stretches north-south across the valley. This zone, which moves from west to east during the day, prevents Oxnard marine air from reaching the eastern end of the San Fernando Valley. Edinger and Helvey reported that during only 4 of 75 summer study days did the westerlies from Oxnard reach Burbank at the far end of the valley. Edinger and Helvey further noted that the rising air in this zone, presumably heavily polluted during its surface travel, is transported to the east after obtaining heights of approximately 600 m. In a study of vertical ozone profiles over the Oxnard Plain, Lea (1968) found maximum ozone levels above the inversion layer and concluded that these levels were apparently the result of air transported from the direction of the Los Angeles Basin. Kauper and Niemann (1975) 79 used elevated wind measurements to construct upper air parcel trajectories which also suggested that measured levels of ozone over Ventura County resulted from advection of pollutants from the Los Angeles Basin. Tracer studies by Hinds (1970) in the Point Arguello area of Southern California indicated that atmospheric stabilty had a strong effect upon the importance of terrain in dispersion mechanisms. Hinds concluded that the effects of local climate and terrain upon dispersion are minimized under unstable conditions, while these effects become more pronounced under stable conditions. Giroux et al. (1974) described the results of an SF 6 tracer study conducted from Southern California Edison's Ormond Beach Generating Station. They found that the SF 6 plume became widespread across the region with no significant differences between concentrations measured in elevated terrain and those found in the flatlands. This was attributed to increased dispersion due to mechanically and thermally induced turbulence over the mountains of the area. In that study, the SF 6 plume was easily tracked as far as 40 km downwind. Work by Raynor et alo (1974) was aimed at determining the differences between dispersion over land and dispersion over the ocean. They concluded that land-sea differences in dispersion were appreciable and were largely determined by the air-water temperature difference. Under stable atmospheric conditions, dispersion over water was much less than that over land, while under unstable conditions, dispersion rates approached those over land. 80 The presence of urban-industrial centers along the coast of the Oxnard Plain and to the east and southeast in Los Angeles County poses difficult questions concerning the pollutant impact relationships between the two regions. The complex mountain-valley and coastal terrain that separates the areas, coupled with the diurnal wind system and variable marine layer, does not permit simple descriptions of pollutant transport, dispersion, and impact from either region. The purpose of the work described herein was to determine possible pollutant transport paths from the Oxnard Plain to the San Fernando Valley and central area of the Los Angeles Basin and to quantify the dilution of pollutants along such paths. The examination of tracer data was expected to provide information about the effects of diurnal winds upon pollutant buildup within the area. The tracer tests would also provide additional data concerning the effects of mountain-valley and coastal terrain upon the dispersion of atmospheric pollutants. 81 3. Experimental Procedure Two tracer tests were conducted from the Ormond Beach Generating Station (OBGS), operated by the Southern California Edi son Company (SCE), during September 21 and 22, 1975. During the first day (Test 1), SF 6 was released continuously from 0300 to 1705 PDT at a constant rate of 3.8 g/sec from one of the two 70 m stacks at OBGS. During the second day (Test 2), SF 6 was released continuously from 0300 to 1140 PDT at a constant rate of 3.8 g/sec from the same stack. The tracer stack gas concentrations were estimated from flue gas flow rate data to equal 1.6 ppm during Test 1 and 1.0 ppm during Test 2. During each release day, a series of automobile air sampling traverses was conducted using three two- person teams. Automobile traverses were made by having the passenger in each car take grab samples of 10-second duration in 30 cm 3 plastic syringes. Generally, samples were collected every 0.2, 0.3, 0.4, or 0.5 miles, depending upon the distance from the release point and the steadiness of the wind. Traverse paths were determined in the field based upon real time wind data obtained from the Point Mugu U.S. Navy Weather Center and the Ventura Air Pollution Control District. Additionally, the location and approximate plume shape were determined prior to each set of traverses by sampling along Highway 1 and Los Posas Road, immediately downwind of OBGS, with a portable electron capture gas chroma to graph. Hourly averaged air samples were obtained with 28 air samplers located at eight sampling sites within Ventura and Los Angeles counties. The air samplers (by Developmental Science, Inc.) permitted the 82 collection of hourly averaged air samples each hour for 12 consecutive hours. These samplers were maintained for seven 12-hour periods from midnight, September 21, through noon, September 24. The locations of the sampling stations and highways used during traverses are shown in Figure 1. Air samples were analyzed for SF 6 using electron capture gas chromatography. Analytical and operational details for electron capture detection of SF 6 are available elsewhere (Drivas, 1974). Five chromatographs and two digital integrators used in the study were set up at the California Institute of Technology, Pasadena, California. Field samples were returned each day to the lab for analysis. Five hundred and seventy automobile traverse samples were collected during the two test days, and 494 hourly averaged samples were collected during the four-day period. Approximately 75 samples could be analyzed per hour; all of the samples were analyzed within two days after the study . The gas chromatographs were calibrated using the exponential dilution method. Calibration results showed that concentrations down to ,o- 12 parts SF per part air could be detected at a signal6 to-noise ratio of better than 3 to 1. The chromatogram peaks were integrated with an Autolab System I electronic digital integrator (Spectra-Physics, Santa Clara, California). The uncertainty of the tracer data was estimated to be less than approximately~ 9% overall. Details of the calibration results, further explanations of experimental procedures, and summaries of all data are given by Lamb et al. (1977a). 83 4. Synopsis of Tracer Tests Test l September 21 , 1975 From 0100 to 0700 PDT, off-shore winds were observed to be light at the surface and stronger aloft over the Oxnard Plain. The sea breeze developed around 0800 PDT and lasted until 1900 PDT over coastal areas of both Oxnard and Los Angeles. In the San Fernando Valley westerly marine flow from Oxnard had penetrated as far as Reseda by 1200 PDT. However, during the remainder of the afternoon, southeasterly flow prevailed throughout most of the valley. The depth of the mixing layer, estimated from upper air wind measurements at Camarillo and Moorpark, increased from approximately 300 mat 1000 PDT to 600 m during the afternoon. According to radiosonde data taken over Los Angeles International Airport (LAX) at 1300 PDT, the base of the inversion layer was at 244 m. Atmospheric stability was estimated to be similar to Pasquill class D conditions according to Turner's method (1961). The tracer plume apparently bifurcated into separate plumes traveling east along either side of the Santa Monica Mountains. As shown in Figure 2, significant, although irregular, levels of SF 6 were observed along the coast as far as Santa Monica , 67 km downwind. The inland plume, shown in Figure 3, extended as far no r th as Moorpark and traveled east as far as Reseda, 62 km downwind. Hourly averaged data at Burbank, shown in Figure 4, along with available wind data indicated that the coastal plume was transported 94 km downwind into the San Fernando Valley where it remained at observable concentrations until 1000 PDT the following day. " 0 0 5 10 15 SCALE 10 20 20 MILES 2~__ _}() KI LOMETERS 15 Figure 2. Overview of automobile traverse tracer data collected between 1631 and 7721 PDT (9/27/75). Maxi~um [SF5J = 726 ppt. SF6 was released from OBGS from 0300 to 1705 PDT, 9/21/75. ANACAPA ISLAND ~9 ~ ~ <:? PASADENA 00 ..i:,. 0 0 GO TRAVERSE 8 LEG CROSSWIND PORTION l 111{ HWY IOI 20 ~ HWY IOI WEST --.oN ')a;I ~ ~ TO 1254 Crosswind prof ile of t racer plumes observed 28 km downwind of OBGS. (KM) 30 (X) u, 40 _ CROSSWIND PORTION~ TIME• FR~H 1215 --~~ NORTH ON 23 S DATE 9/21/75 DISTANCE AL(jNG TRAVERSE 10 ~ _...,~ NORTH ON 23 S Figure 3. 0 OL 0 N ~:::,,IHWY u . 1 z 0 u 1-o a:: 0 ....... la: zCD '.J a_ Q_ I- ,,..... TEST - 1...~ - - - - - - 0 0 1 ••••••••• ♦♦♦♦♦♦♦♦♦ TI f1E< POT> \ ORTE SANTA MONICA •••••••••••••• MALIBU ............... ..,,..,, z 60 ..,, n 0 z ..,, 60 TIHE<POT) I. DATE ♦♦♦♦♦♦♦♦♦ 1 TI f1E< POT> Cc ORTE ♦♦♦♦♦♦♦♦♦ /22) 2ll AESEOR •••••••••••••• ♦♦♦♦♦♦♦♦♦ = ,------------, (9/22 > 21l CX) a, Figure 4. Hourly averaged SF6 data as a function of time. -~ (9/21 > 12 (9/21) 2Y <9/22 > 12 (9/22 > 2ll TI HE( POT) l DATE ■ ♦♦♦♦♦♦♦♦♦ Tlf1E(POT> l DATE THOUSAND OAKS •••••••••••••• ru:;----------=- 60 <- INDICATES NO DATA, + INDICATES TRACER RELEASE PER'JCIO) ♦♦♦♦♦♦♦♦♦♦♦♦♦♦ BURBANK ♦♦♦♦♦♦♦♦♦♦♦♦♦♦ LENNOX Cljq (9/21 > 12 (9/21 > 2Y (9/22TT2 TIHE<PDT> e. DATE 60 87 A mass balance analysis of the automobile traverse data accounted for essentially all of the tracer. In particular , the total mass of SF 6 was conserved between the single plume observed immediately downwind of OBGS and the dual plumes measured 28 km downwind. It was necessary in this analysis to assume that the tracer was uniformly mixed from the ground to the top of the mixing layer. A mass balance of the Burbank hourly tracer data accounted for half of the released tracer; similar analysis of the Reseda hourly data accounted for one-third of the released tracer. In each analysis, the observed SF 6 levels were assumed to represent a uniform concentration in a plane extending vertically to the top of the mixing layer and horizontally as wide as the entrances to the San Fernando Valley. Test 2 September 22, 1975 Onset of the sea breeze occurred at 1000 PDT over both Oxnard and Los Angeles. On-shore flow was strong throughout the day until 1900 PDT. In the San Fernando Valley, Oxnard marine air reached Reseda around 1500 PDT. During the remainder of the afternoon, northerly winds pre- vailed in the western half of the area and southerly winds prevailed i n the eastern half. Los Angeles. After 1900 PDT, light off-shore flow developed over San Fernando Valley winds were l i ght and variable. The height of the base of the inversion, measured at Pt. Mugu, increased from 44 mat 0455 PDT to 111 mat 1615 PDT. Measurements over LAX at 1300 PDT were simi lar; the base of the inversion was 122 m above the surface. A single plume was well-defined in crosswind profiles obtained 7 km east of OBGS. The tracer reached Malibu around 1200 PDT and was 88 also observed in Santa Monica and Lennox during the afternoon. levels were measured in Burbank during the evening. Low No evidence of inland transport was obtained during the second SF 6 tracer release. Measurable concentrations of SF 6 were observed at Lennox during the two nights (9/22 and 9/23) following the second release day as shown in Figure 4. During both nights, the land breeze was relatively strong throughout the Los Angeles Basin. It appeared that air from the San Fernando Valley was swept back to sea during these nights. Although these wind patterns explain the reappearance of the SF 6 plume at Lennox, it seems somewhat surprising that the observed levels would be caused by air with such long transit timeso 89 5. Discussion of Results The transport paths observed in the tracer data were compared to surface air parcel trajectories based on the rather extensive set of surface wind data obtained for the test period. Wind data were measured at the stations shown in Figure 1. Surface air parcel trajectories were determined in two ways. First, hourly surface streamlines were drawn by hand from the wind data; these streamlines were then used to determine the trajectory of an air parcel, hour-by-hour . A second means, provided by W. Goodin (1977), was based upon the development of hourly wind fields from numerical solutions of the two-dimensional conservation equation. A typical wind field over a 40 by 30 grid system of 2 mile squares is shown in Figure 5. Digital terrain data were used as input for the boundary condition at the surface. Typical trajectories determined with the two methods are shown in Figure 6. Generally, the streamline trajectories traveled east along the coast into the Santa Monica-Lennox area. Depending upon where the transport paths crossed the Los Angeles coastline, trajectories either turned north into the San Fernando Valley or traveled east toward Ontario. The nighttime land breeze, in some cases, carried San Fernando Valley air offshore; these air parcels returned towards Ontario with the early afternoon sea breeze. test passed directly over Lennox. Similar trajectories for the second Tracer arrival times predicted from hand-drawn trajectories generally agreed with the observed tracer data. The meandering of the trajectories within the San Fernando Valley provide an explanation for the slow disappearance of tracer at Burbank. Figure 5. + + + + + + + + + + + + ">, ">, ' ">, ...., ' '>, .................... ">, ..... ..., ...,,,...,, .................... , 1 ~/ 7 :\..::::: ::,., , ::,., ::,., ::,.,::,.,>, 1 ........ + + + + + 777'l 'I 1\ \ t'l f ::,., > > > > >::,., >> >,,, > >::,.,::,.,::,.,..., ' . > ~ ~ ~~~~~ ..., ..., ..., ..... , >, ::,.., >, :,,, :::,., ,,, ' > >, :,., ' ::,,, ' >, ' ' ' , , , > , ., ~ Hourly surface wind vectors, 1200 PDT, 9/21/75 (W. Goodin , 1977). Reference locations : Ventura (VNT), Ormond Beach Generating Station (OBGS), Thousand Oaks (THO), Malibu (MLB), Reseda (RSD), Lennox (LNX), Los Angeles (LA), and Burbank (BUR). Vector of one inch equals 15 mph, and(+) indicates wind speed less than l mph. ' > >, > ' :,.,::,,,:,,.,::,,,::::,,::,,,'>,>,:,,,:,,,:::,,, ,,,> ~ ~ ; > ~ : = : ~ ~ Ii I >,">,>,',',> ~~ """-~ _ , . , , , , __ I_ ·1 ,,,, >,',,>,,>,>,>,> 2 >> ~~ I ' ::,,, >, ' :::,., >, > ~ ~ - . . . . . , ; ; - ~...t.::::li~· :.-i~~~;t..:~ / '71 >, :,,, >, ::,,, >, > t :!:!- ~ ~_:?.:-> ...,, ...,, ..., ..., ...., ...,, ',,>,>>,:::,.,>,!>,>,">,',,>, , .................... ...,, .......... ...,,,,...,,...,,...,,,,,,,...,~~, ~~~~ , , > , > > , > ..., ..., > > > ...,, ..., > > ~~~ LNX * ::- == -> > > ~ ..., >::,.,::,., >::,., >...,::,.,..., > > > > > > ~~_>_,=>~~ ~~:::,..-> > > ">~✓>>>>'>>'>>>>>> ~~~<UIJ , i j r'~ z~ 'I, ;;;~B;S~;~;;;:~~ : .,,_::,., ~ > .>~~~......... + ') ') + +7 + + + ~t-..' \ ,., ~~· - • + + + + + +"'- : \ ? ? \ ; : , . ~~THO : ~ ,: ,., . . :, ,.,:,...., '>>>> ._ R~O _* + + 7 7+ BUR *, ' ~;:a,,_,;;;~.....,;;;-....._;: ..... ~......., ... ~ - ==a?:~~-;:> ">, ">, ">, ...., ' + + + + + + + + + t + + + + + + + + + t t 'I t t _.....;..>~., ., ., ., ., ., ., .,., .,,,,, , , , , , , ~ ~ + + + + + + t 1 1 '1 1 1 ~ ..... ...,, ..... , ...., , ..... '>> .......... ...., ::,.,::,.,::,.,~ e'/~/ + + + + + + t 'I ~ ~ 1 1 1 ~ ~ _ _ . . > : : , . , >, ::,., > ::,., > ::,., >'>'> ~ (.f1~ + + + + + t 'I ~ 1 11 11 ~~ .......... ,. ..... ,. >> ~ ~ ~++++++'I 1 1 ~ ~ 1, 1, • ~ ~~ :::.,., ..... ..... ..... ,. ,> ~~~ + + + + • • • 1 1 1 ~VNT ........ ~~~-=:::::>,...., , > ~ ~• • t t • • • • 111,... - """ ~ ~ .;!.'>~ ' - ••• j ~"5"5 i + + + + + + + + + + + + + 1, /71,,,7 £ + t,% ............... ..., ~ - > ........................................ ...., >>1 + + > ::,., '>,">, ..... ...., ..... '>>'>1 + + ~~ > ::,., ..... > > UJ 0 II 12 ◊LENNOX I SCALE -=~~ol====ii,5_ _....., ,o====:,;15,..._ _ _ 2_0 MILES 0 10 15 20 25 30 KI LOMETERS Figure 6. Surface forward air parcel trajectories: (• ) determined from hand-drawn hourly streamline maps; ( ■) determined from hourly wind vector fields numerically calculated with the conservation equation (W. Goodin, 1977). Release times on 9/21/75: top, 0400 PDT; center, 0700 PDT; bottom , 1300 PDT. 92 The trajectories calculated from the wind vector fields generally, do not agree with either the streamline trajectories or the tracer data . Because of the lack of data along the coast between OBGS and Santa Monicat the inland wind stations influence the coastal trajectory to a greater extent than appears reasonable. As will be shown in Part II of this seriest in cases where sufficient data are availablet the numerical trajectories show excellent agreement with tracer transport paths. Dilution of pollutants along coastal and inland trajectories was quantified by determining values of the crosswind horizontal dispersion parametert cryt from best-fit Gaussian curves through the traverse tracer data. Values of the crosswind vertical dispersion parametert cr 2 t were determined from mass balance considerations using the Gaussian plume model expression. The direction of the wind with respect to the traverse route was taken into account in these calculations. Details of the calculation procedure are given elsewhere (Lamb et al.t 1977b) . The results 9 in terms of cry and cr 2 as functions of downwind distancet are shown in Figure 7. In terms of the usual power law dispersion coefficientst the Oxnard dispersion relationships are: 12 a = 0. 042 X1. y and a z = 18.8 x0 • 20 (1) (2) where Xis the downwind distance in meters. These dispersion relationships are compared to the dispersion curves for the Pasquill-Gifford stability conditionst as given by Turner (1970) 9 in Figure 7. The Pasquill-Gifford curves are based upon dispersion data which were obtained over open t flat country. 93 1011 (f) a: LU I- LU :E . 103 >-I a: :E C) ....... (f) 102 1Q4 1Q3 X, METERS 105 OXNARD AUTO TRAVERSE CORR.= .805 1011 (f) a: LU I- LU . :E 1Q3 N I a: :E C) ....... (f) 102 10 1 '----'---'--'---'--'-.J....L.1.-'------''--L-.L--'--L...J...J..J-L--L-.L.....J.-'--'U-L.l.J 102 105 103 1011 X, METERS OXNARD AUTO ·TRAVERSE CORR. = .1127 Figure 7. Comparison of horizontal and vertical dispersion curves determined from 10- second averaged tracer concentrati ons with Pasqui l lGifford dispersion curves . 94 Wind speed and insolation data suggested that Pasquill-Gifford class D conditions existed during the afternoon of Test 1. However, the rate and extent of the observed horizontal dispersion were greater than that given by the Pasquill-Gifford class D curve. Vertical dispersion was almost independent of downwind distance. In view of the low inversion layer and the presence of the rough terrain, the vertical dispersion results suggest that tracer became uniformly mixed in the vertical direction at relatively short downwind distances. Thus, the small rate of vertical dispersion is not an indication of stable conditions as the curve in Figure 7 implies. Both vertical and horizontal dispersion were apparently increased over that predicted from wind speed and insolution considerations because of increased turbulence over the rough terrain. The experimental dispersion relationships given in Figure 7, as well as results from previous studies (Drivas and Shair, 1974, Start et al., 1974, and Liu et al., 1976), indicate that the commonly used Gaussian plume model will serve as an .upper bound to pollutant dispersion over rough terrain. The Gaussian model may be more accurate when experimental dispersion data, such as that given in Figure 7, are available for a particular locale or for a particular type of terrain. As an example, meteorological data and tracer release rates from the afternoon of Test 1 were used in the Gaussian plume model to calculate tracer concentrations as a function of downwind distance. The presence of the inversion layer was accounted for by using the plume trapping expression given by Bierly and Hewson (1962). For centerline, ground-level concentrations, the expression simplifies to: 95 (3) where the summation terms are used to account for multiple reflections of the plume at the bottom of the inversion layer and at the ground. Bierly and Hewson noted that only a few of the terms in the series are necessary for practical applications. In this case, we have used N = 4. The tracer release rate, Q, was constant at 3.8 g/sec; the average wind speed, measured at five stations throughout the area between 1200 PDT and 2000 PDT, was 2.7 m/sec. The effective source height, H, was calculated to be 198 m, according to Briggs (1969), (1972). The inversion height, L, was 555 m, averaged at OBGS, Camarillo, Moorpark, and LAX during the afternoon. Observed and calculated centerline tracer concentrations are given as a function of downwind distance in Figure 8. The two highest curves in the figure were calculated using the Pasquill-Gifford dispersion relationships given in Turner (1970) for stability classes D and C, respectively. The lower curve was calculated using the Oxnard experimental dispersion relationships given by Equations (1) and (2). The results calculated with class D stability serve as an upper bound to the experimental concentrations as expected. The calculated concentrations determined with the Oxnard dispersion terms yield a reasonably close fit to the data. However, it appears that the best fit is obtained using dispersion values associated with class C, which is one category less stable than that inferred from considerations [!] - 20 - - 60 RESEDA -.- - - O~WNWIND DISTANCE CKM) 40 MALIBU --•-C!J- - ............................. ······ THOUSAND OAKS .•4. - ...... . [!) - [!] - - 80 LENNOX - - - --- 100 BURBANK - ····················rn···· • OBSERVED 10-SEC ORTA OBSERVED HOURLY DR TA 9/21/75 RELEASE SITE• C,BGS RELEASE RFlTE CG/SEC) 3.8 MIXING- HEIGHT CM) 555. WIND SPEED CM/SEC) 2.7 EFF. ST. HT. CM) 198. OXNRRO 1 Figure 8. Centerline SF 6 concentrations compared with centerline concentrations predicted from the Gaussian plume model: - - calculated using experimental values of cry and oz; - - calculated using Pasquill-Gifford stability class C values of ay and oz; ... calculated using Pasquill-Gifford stability class D values of ay and o2 • 10° • 0 10 1 102 103 PPT. C()NC 1011 10 5 106 \0 O"I 97 of wind speed and insolation. As can be seen , the Gaussian model does predict reasonably accurate concentration profiles, when modified to account for the influence of terrain. An important feature of the tracer data is that both 10-second and hourly averaged concentrations lie along the same curve. Hino (1968) has suggested that the relationship between centerline concentrations and sampling times follows a power law of the form: (4) where c1 and c2 are concentrations measured over sampling time t 1 and t 2, respectively. Hino noted that for sampling times less than 10 minutes, a= -1/5, while for sampling times between 10 minutes and five hours, a= -1/20 The Hino correction indicates that the 10-second averaged concentrations should be approximately 5 times that of the corresponding hourly averaged data. Figure 8 shows that the winds were steady enough to invalidate the Hino correction. In Part II of this series, a more detailed study in another coastal region gives further evidence that during certain steady conditions of coastal meteorology,the Hino correction is invalid. The SF 6 injected into the stack becomes diluted along with other stack gases by atmospheric turbulence; SF 6 concentrations measured at ground level downwind can be converted to equivalent ground-level pollutant concentrations associated with that particular stack. Thus, 98 (5) cp = cp(stack) • c SF6(stack) where CP is the ground-level pollutant concentration, Cp(stack) is the stack-gas pollutant concentration, and CSF and CSF are 6 6(stack) corresponding SF 6 concentrations. The ratio of the SF 6 ground-level concentration to the stack gas concentration can be defined as the dilution factor, DFo The magnitude of the dilution factor is a measure of the dilution which occurs during transport from the source to a particular receptor. Thus, the dilution factor is a function of transport distance and topographical and meteorological conditions. The OF values calculated from the maximum hourly averaged SF 6 concentration at each station are listed with downwind distance in Table 1. These calculations were based upon an average SF 6 stack concentration of 1.6 ppm for Test l and 1.0 ppm for Test 2. Generally, pollutants emitted from OBGS during Test l and Test 2 were diluted by approximately 10 5 after transport into the Los Angeles area. Since the stack concentrations of pollutants which existed during the tests were unavailable, downwind pollutant concentrations could not be predicted. However, if the so 2 levels in the two OBGS stacks had been 200 ppm , hourly averaged so 2 concentrations would have been less than those given in Table 1; the concentrations of so 2 in Table 1 were calculated assuming no losses due to deposition or reaction. On the average, the predicted so 2 levels account for 28.3 46.5 46.7 62.4 66.4 66.9 78.6 94 . 0+ Thousand Oaks Malibu Simi Valley Reseda Newhall Santa Monica Lennox Burbank 0.5•10- 5 l.6·10- 5 0.7·10- 5 1.2-10- 5 0.4•10- 5 2.3•10- 5 4.3·10- 5 1. 9.10- 5 l.8•10- 5 5.0•10- 5 0.7•10- 5 2.,-10- 5 0.6•10- 5 1.4· 10- 5 3.3•10- 5 - Test 2 Test l Minimum Dilution* Factor 13 8 17 9 6 2 5 3 6 2 2 8 3 20 7 Predicted S02** Concentration ppb Test l Test 2 20 10 20 20 20 60 20 20 Observed Maximum S02 Concentration Test l Test 2 *Corresponds to maximum observed tracer concentration divided by tracer stack gas concentration. **Assumes S02 stack gas concentration equals 200 ppm in each of the two OBGS stacks; so 2 is assumed to be unreactive during transport. +Measured along the coast through Lennox to Burbank. Distance from OBGS (km) Location HOURLY AVERAGED DILUTION FACTORS, PREDICTED S02 CONCENTRATIONS AND OBSERVED S02 CONCENTRATIONS TABLE 1 <.O <.O 100 one-third of the levels observed at the stations listed in Table 1. 101 7. Summary and Conclusions Data obtained during two days of tracer releases from OBGS indicate that pollutant transport occurs from the Oxnard Plain along the Malibu Coast into the Lennox area of the Los Angeles Basin and by an inland route into the San Fernando Valley as far east as Reseda. The tracer and wind data also suggest that transport along the coast moves pollutants north into the San Fernando Valley. Meteorological conditions during the test days were similar to those typically observed in the area. A mass balance analysis of the tracer data accounted for essentially all of the tracer. The tracer transport paths coincide with those predicted from air parcel trajectories determined from hourly streamline maps, but do not agree with trajectories based upon numerical solutions of the conservation equation. The streamline trajectories indicate that emissions from OBGS are ultimately transported towards Ontario. The dispersion of pollutants along these paths is greater than that predicted by the Pasquill - Gifford dispersion relationships for open, flat country. Evidently, the rough terrain downwind of OBGS increases atmospheric turbulence and, thus, increases the extent and rate of dispersion. As a result, concentrations estimated with the Gaussian plume model are greater than those observed experimentally. By using either experimentally determined dispersion terms or Pasquill-Gifford dispersion terms associated with conditions one class less stable than those inferred from wind speed and insolation, concentrations predicted with the Gaussian model agree quite well with observed values . 102 Acknowledgments We are happy to acknowledge Charles L. Bennett, California Air Resources Board; Hans Giroux, Meteorology Research, Inc.; Dr. William Thuman, Douglas Tubbs, Steve Mollnar, Dr. Thomas Wurzburger, Ventura County APCD; Dominick Mercadante, Los Angeles County APCD; Dr. Duane Lea, Geophysics Divis,on of the Pacific Missile Range; and George Griffith, Henry Smith, and Sharon ViGario, California Institute of Technology. We especially thank William Goodin for his cooperation and contribution to the numerical air parcel trajectory analyses. Personnel from Meteorology Research, Inc. conducted the releases of SF 6 from the Ormond Beach Generating Station. This work was supported under the California State Air Resources Board Contract No. ARB 5-306. 103 Literature Cited Angell, J. K., D. H. Pack, G. C. Holzworth, and C.R. Dickson (1966) ''Tetroon Trajectories in an Urban Atmosphere," J . .8.Q.Ql. Meteor. 5, 565-572. Angell, J. K., D. H. Pack, L. Machta, C. R. Dickson, and W. H. Hoecker (1972) "Three-Dimensional Air Trajectories Determined from Tetroon Flights in the Planetary Boundary Layer of the Los Angeles Basin," i• ~. Meteor. 11, 451-471. Bierly, E. W., and E. W. Hewson (1962) "Some Restrictive Meteorological Conditions to be Considered in the Design of Stacks," J. lill.Ql_. Meteor. 1, 383-390. Briggs, G. A. (1969) Plume Rise. USAEC Critical Review Series TID-25075, National Technical Information Service, Springfield, Virginia 22151. Briggs, G. A. (1972) "Discussion on Chimney Plumes in Neutral and Stable Surroundings," Atmos. Environ . .§_, 507-510. DeMarrais, G. A., G. C. Holzworth, and C. R. Hosler (1965) "Meteorological Surrmaries Pertinent ·to Atmospheric Transport and Dispersion over Southern California," Technical Paper No. 54, ESSA-Weather Bureau. Doc. # C30:28:54. Supt. of Doc., U.S. Gov't Printing Office, Washington, DC 20402. Drivas, P. J. (1974) "Investigation of Atmospheric Dispersion Problems by Means of a Tracer Technique," Ph.D. Thesis, California Institute of Technology, Pasadena, California 91125. Drivas, P. J., and F. H. Shair (1974) "A Tracer Study of Pollutant Transport and Dispersion in the Los Angeles Area," Atmos. Environ . .§_, 1155-1163. Edinger, J. G. (1959) "Changes in the Depth of the Marine Layer over the Los Angeles Basin," i• Meteor . .1§_, 219-226. Edinger, J. G., and R. A. Helvey (1961) "The San Fernando Convergence Zone," Bull. Amer. Meteor. Soc. 42, 626-635. Giroux, H. D., L. E. Hauser, L. H. Teuscher, and P. Testennan (1974) "Power Plant Plume Tracing in the Southern California Marine Layer," from Proceedings of the Symposium on Atmospheric Diffusion and Air Pollution. American Meteorological Society, Boston, Massachusetts, pp. 238-245. Goodin, W. (1977) Personal communication. Environmental Quality Laboratory, California Institute of Technology, Pasadena, California 91125. 104 Hinds, W. T. (1970) "Diffusion over Coastal Mountains of Southern California," Atmos. Environ . .i, 107-124. . ,. Hino, M. (1968) "Maximum Ground-1 evel Concentrations .and Sampling Times," Atmos. Environ. £, 149-165. • Kauper, E. K., and B. L. Niemann (1975) "Los Angeles to Ventura OverWater Ozone Transport Study, Vol. I," Technical paper , California Air Resources Board Contract No. ARB-4-1126, Sacramento, California 95814. Lamb, B. K., A. Lorenzen, and F. H. Shair (1977a) "Tracer Study of Power Plant Emission Transport and Dispersion from the Oxnard/Ventura Plain," final report for the State of California Air Resources Board Contract No. ARB-5-306, Sacramento, California 95814. Lamb, B. K., and F. H. Shair (1977b) "Atmospheric Tracer Studies to Characterize the Transport and Dispersion of Pollutants in the California Delta Region," California Air Resources Board Contract No. ARB A5-065-87, Sacramento, California 95814. Lea, D. A. (1968) "Vertical Ozone Distribution in the Lower Troposphere near an Urban Pollution Complex,"~-~- Meteor. I, 252-267. Liu, M. K., D. Durran, P. Mundkar, M. Yocke, and J. Ames (1976) "The Chemistry, Dispersion, and Transport of Air Pollutants Emitted from Fossil Fuel Power Plants in California: Data Analysis and Emission Impact Model," final report prepared by Systems Applications, Inc. for the State of California Air Resources Board Contract No. ARB-4-258, Sacramento, California 95814. Raynor, G. S., P. Michael, R. M. Brown, and S. SethuRaman (1974) "A Research Program on Atmospheric Diffusion from an Oceanic Site," from Proceedings of the Symposium on Atmospheric Diffusion and Air Pollution. American Meteorological Society, Boston, Massachusetts, pp. 289-295. Start, G. E., C. R. Dickson, and N. R. Ricks (1974) "Effluent Dilutions over Mountainous Terrain and within Mountai.n Canyons , " from Proceedings of the Symposium .Q!l Atmospheric Diffusion and Air Pollution. American Meteorological Society, Boston, Massachusetts, pp. 226-232. Turner, D. B. (1961) "Relationships between 24-hour Mean Air Quality Measurements and Meteorological Factors in Nashville, Tennessee," i• Air Poll. Contr. Assoc . .ll, 483-489. Turner, D. B. (1970) "~Jorkbook of Atmospheric Dispersion Estimates," Environmental Protection Agency. Research Triangle Park, North Carolina. 105 CHAPTER 4. (Submitted for publication in Atmospheric Environment) ATMOSPHERIC DISPERSION AND TRANSPORT WITHIN COASTAL REGIONS. PART II. TRACER STUDY OF INDUSTRIAL EMISSIONS IN THE CALIFORNIA DELTA REGION by Brian K. Lamb, and Fredrick H. Shair* Division of Chemistry and Chemical Engineering California Institute of Technology Pasadena, California 91125 August, 1977 * To whom al l correspondence should be addressed . 106 Abstract Eight large-scale atmospheric tracer tests, utilizing CBrF 3 and/or SF 6, were conducted during September, 1976, within the California Delta Region. The purpose of this study was to demonstrate the large- scale application of single and double tracer techniques to the study of pollutant transport and dispersion under complex, coastal meteorological conditions. The resulting tracer data base was comprehensive enough to permit accurate mass balances of the tracer; essentially all of the tracer was accounted for by this analysis. On the average, plumes emitted from the Montezuma Hills during the test periods were transported southeast over Stockton into the San Joaquin Valley. As a result of the steady nature of the winds, the commonly used Hino correction was found to grossly underestimate the hourly averaged tracer concentrations computed from 10-second averaged concentrations. A comparison of experimentally determined dispersion parameters with those associated with Pasquill-Gifford atmospheric stability classes indicated that atmospheric stability generally decreases with increasing distance downwind from the Montezuma Hills. In spite of the complex meteorology and terrain, estimates of tracer concentrations based upon the Gaussian plume model were found to be reasonably accurate. A method was developed for the conversion of tracer con- centrations to inert and, in some cases, reactive equivalent pollutant concentrations. 107 The relationship between the horizont al standard deviation of the wind , cr 0, and the horizontal dispersion parameter of the plume , cry, was found to follow cry~ Oo85 x cr 0. Calculated trajectories and arrival times of air parcels based upon a numerical solution to the two-dimensional mass balance equation were found to be in excellent agreement with the tracer datao 108 1. Introduction The economic trade-offs between industrial growth and environ- mental damage are of growing concern. Industrial development may lead to economic growth in one area while causing economic loss in another. For example, proposed industrialization of the Montezuma Hills, located in the California Delta Region, is expected to stimulate the local economy. However, the possible in,crease in air pollution associated with new industrial facilities may seriously depress the economy of the inland valleys. Therefore, it is important to quantitatively assess possible monetary losses due to degradation of health and damage to livestock and crops downwind of proposed industrial activities. In order to determine the impact of industrialization upon air quality, it is necessary to characterize the transport and dispersion of pollutants from existing sources as well as the transport and dispersion of pollutants from proposed sources. The transport and dispersion of pollutants under complex topographical and meteorological conditions have been the concern of a growing list of workers (Pasquill, 1974). Start et al. (1974) used SF 6 as a tracer to determine the amount of dilution associated with airborne material passing over and around mountains and canyons at transport distances up to 5 km. They concluded that enhanced mechanical turbulence associated with rough terrain increased dilution over that found in flat regions; furthermore, terrain-induced dispersion appeared to increase with increasing atmospheric stability. Similar conclusions were reached by Giroux et al. (1974) from a tracer test conducted in the Oxnard Plain of coastal Southern California. In that test, the SF6 109 tracer was followed as far as 40 km inland over the Camarillo and Los Posas Hills. Angell et alo (1966, 1976) used tetroons to study pollutant trajectories through the Southern California air basin. Tetroons released along the coast in some instances were carried out to sea and then, with the diurnal reversal of the winds, swept back into the interior of the basin. Drivas and Shair (1974) released 33.5 kg of SF 6 over 45 minutes from Anaheim, California, and followed the tracer as it passed through the Santa Ana Canyon, over Banning Pass, and finally into Palm Springs and the Coachella Valley at a distance of 124 km downwindo Fosberg et al. (1976) attempted to account for the effects of mass divergence upon pollutant dispersion by modifying the well-known Gaussian plume model. Inclusion of a divergence term in the Gaussian expression decreased calculated pollutant concentrations by as much as a factor of two. Liu et al. (1976) modified the Gaussian dispersion terms to account for the roughness of the terrain. Their derivation of the modified dispersion parameters was based upon the relationship of the eddy diffusivities and the Gaussian dispersion parameters. Liu et al. found that for a surface roughness of 50 cm, the modified dispersion parameters caused calculated concentrations to decrease by a factor of two from the conventional Gaussian model. Several authors have utilized forms of the diffusion equation to develop dispersion models which take into account the effects of terrain. Roffman and Grimble (1974) developed a three-dimensional model based on successive coordinate transformations where one coordinate was required to be parallel to the flow during each segment of the transport path . 110 Reynolds et al. (1973) used a coordinate transformation to take into account the irregularity of the floor of the Los Angeles Air Basin. The complexity of coastal topography and meteorology often requires that atmospheric tracer techniques be used to quantitatively determine the transport and dispersion of pollutants. The purpose of the work described herein was to demonstrate the usefulness of single and dual tracer techniques in probing the transport and dispersion of pollutants under complex meteorological conditions over distances as large as 100 km. The data obtained from a tracer field study can be used in the development of atmospheric dispersion models suitable for application in complex situations. Furthermore, tracer data are of intrinsic value to industrial designers and regulatory personnel whose aim is to insure acceptable air quality downwind of proposed industrial facilities. Results from dual tracer releases are of particular value in determining the relative impact of pollutants emitted from existing sources with those projected for emission from proposed sources. In order to accomplish these goals in the California Delta Region, eight full-scale tracer studies were conducted during September, 1976. The cost of this tracer study was less than . 02% of the capital investments required for the construction of a chemical manufacturing facility proposed for construction in the California Delta Region. Atmospheric tracers (SF 6 and CBrF 3) were released from property owned by the Dow Chemical Company in the Montezuma Hills, from Martinez, and from Pinole. Extensive air sampling and meteorological observations systems were employed during the field study in cooperation with 111 Meteorology Research, Inc. (MRI). The region in which the field study was conducted is shown in Figure 1. This paper is the second part in a seri es describing the results of four large-scale tracer field studies conducted in coastal regions . This series of papers seeks to determine the differences and similarities associated with the transport and dispersion of pollutants in these four coastal regions. In Part I (lamb et al . , 1977), the results of a tracer study conducted from the Oxnard Plain of Southern California were presented. Part III (Drivas and Shair , 1977) in this series deals with the findings of tracer studies conducted from Moss Landing near Monterey Bay and from Long Beach in the Los Angeles Basin. 112 ■ UPPER A SURFACE AIR STATIONS WIND STATIONS N ;;,;~ . 1,~~ffJ s~¼ . ., :C·•:[Q] AIR SAMPLERii:°ri'2:·H~urs) / :: •:•: .•: ., ELEASE 0 ELK GROVE@] ,o.,,o.~}Un t0 IJlll[S \{ --~ ) ' ' " ,,.,, a,, Figure lo San Francisco Bay Area and the California Delta Region . Contour lines are in intervals of 1000 feet . 113 2. Meteorology and Topography of the California Delta Region Surface wind patterns from Smalley (1957} indicate that the Carquinez Strait is the major channel for air flow over the Delta Region During the years, 1952-1955, wind patterns which included an air flow trajectory from Richmond through the Carquinez Strait into the inland valleys occurred during 49% of the time. As shown in Table 1, these patterns occurred during 71% of the time during September. Of these patterns, the most prevalent were the west and northwest wind flow types shown in Figure 2. According to Smalley, these patterns occurred 51% of the time during September. These surface patterns show how marine air can become laden with urban pollutants in the Bay Area, pass over the Delta Region, and turn either north into the Sacramento Valley or south into the San Joaquin Valley. Fosberg and Schroeder (1966) explained the onshore summer flow patterns as the result of strong pressure and temperature gradients developed from the coexistence of the eastern Pacific subtropical high pressure area off the coast and a thermal trough in the interior. Wind flow from the Bay Area through the Delta typically centers around the daily occurrence of an afternoon sea breeze. Fosberg and Schroeder (1966) pointed out that this sea breeze , which is caused by differential heating and cooling of the land and sea, is superimposed upon the summer marine air flow. Miller (1968) noted that the Central California sea breeze was accompanied by a sharp, low-level temperature inversion extending inland as far as 40 km. 114 TABLE 1 OCCURRENCE OF WIND PATTERNS WHICH INCLUDED TRAJECTORIES FROM RICHMOND THROUGH THE CARQUINEZ STRAIT INTO THE SACRAMENTO OR SAN JOAQUIN VALLEY, 1952-1955* Frequency(%) ** MONTH 0400 PST 1000 PST 1600 PST 2200 PST AVERAGE Jan 15 11 23 15 16% Feb 9 9 33 20 18% Mar 31 31 60 48 43% Apr 40 41 74 68 56% May 56 61 67 73 64% Jun 74 78 88 86 82% Jul 80 87 94 92 88% Aug 84 81 88 69 81% Sep 58 64 87 73 71 % Oct 27 40 73 51 48% Nov 13 8 27 18 17% Dec 6 6 14 9 9% AVERAGE 41% 43% 61% 52% 49% *WIND PATTERN TYPES: S3, S5, SW2, SW3 , SW4 , SW5 , Wl, W2, W3, W4, W5, W6 , NW2 , and NW5 (SMALLEY, 1957). ** Calculated from data presented by Smalley (1957). 115 ......... ,..._ "'C 3: I.O ~- com rd ,-- ,-- +> Vl "'C >, QJ C QJ 0, rd 3: .,.. ,-3: ,-C +> E •r- Vl V') ,-- QJ .,.. 3: . rd .c: Vl > +> QJ QJ s... a. s... 0 >, o..c+> 0 • ...u I • C 0 0 .2 N QJ s... ::::, 0, .,.. LL 0 • C w u 0 0 • .2 I ,~. . 0 • u w u .I • . C 0 0 2 0 -• • u C C w u I 0 • 0 2 0 ... C . C u 0 0 •• . "' 0 0 116 The diurnal land-sea breeze cycle can be divided into four periods of the day (Smith, 1976). During Sea Breeze conditions, from approximately 1300 to 1800, winds are relatively strong throughout the area. The average wind speeds at Martinez, Sacramento, and Stockton during the Sea breeze test periods in the work reported herein were respectively, 5 m/sec, 2 m/sec, 4 m/sec . Smalley noted that wind flow types W5 and W6 accounted for 40% of the patterns observed at 1600 hours. In the afternoon, the mixing depth reaches a maximum due to the heating of the land. Typical afternoon mixing depths during the test period were between 1000 and 3000 meters. This pattern is opposite to that which occurs during Nighttime conditions from approximately midnight to 0500. At night, the depth of the mixing layer drops to a minimum, typically between 100 and 500 meters. Air flow stagnates as drainage from the Sierra Nevada Mountains meets the weak sea flow over the Delta Region. According to the Smalley report, light and variable winds occurred during 38% of the measurements taken at 0400; west wind types accounted for 39% of the measurements in the early morning. Two transition periods separate the Sea Breeze and Nighttime regimes: PreSea Breeze conditions occur from approximately 0600 to 1100 and are characterized by an increase in the depth of the mixing layer and development of the marine air flow. Wind type W2 was observed in 13% of the patterns at 1000, and light and variable winds were measured during 28% of the time at 1000. Sea Breeze Tail conditions follow the afternoon period and are typified by a decrease in mixing height and a decrease in the strength of the coastal flow. West wind flows in the Smalley work existed in 48% of the observations taken at 2200. 117 As a result of this summertime meteorological cycle, a strong, relatively constant jet of air issues from the Carquinez Strait and fans out into the reaches of the Delta Region. Fosberg and Schroeder (1966) indicated that the wind surge associated with the invading marine air gains speed as it passes through the Carquinez Strait where it combines with the intensely channeled summer marine flow. It appears that part of this jet maintains its strength past the Montezuma Hills and then dissipates fairly rapidly just beyond the area. For example, during the two-week test period in early September described herein, the average surface wind speed in the Montezuma Hills was 7 m/sec; further downwind at Brentwood and Venice Ferry, the average wind speeds were 2 m/sec and 3 m/sec, respectively. The existence of this jet suggests that material emitted from the Montezuma Hills may be carried into the central valleys within a narrow, stable stream of air. Material emitted west of the Carquinez Strait into the marine flow may be widely dispersed by the divergence of air from the Strait. Pollutants emitted from sources located very near one another in the vicinity of the Carquinez Strait may be transported along widely different trajectories into the Sacramento or San Joaquin Valleys. Terrain effects on air flow appear to be very important in determining pollutant trajectories through the Delta Region. In a Bay Area tracer study utilizing fluorescent particles, Sandberg et al. (1970) found that the hilly Bay Area terrain, which rises as high as 450 meters, served to deflect westerly marine air flow into northerly and southerly trajectories. Sandberg et al. noted that the presence of low-level temperature inversions enhanced the effects of the Bay Area 118 terrain upon air flow. Their data indicate that when the width of the plume is larger than the width of the channels associated with the hills, then dispersion increases beyond that over flat terrain; however, when the width of the plume is smaller than the width of the channels, containment by the terrain can lead to decreased dispersion. 119 3. Experimental Procedure Eight tracer studies were conducted from August 31 , 1976, through September 14, 1976 , within the Delta Region during the four designated meteorological periods. During seven of the tests, either SF6 or CBrF 3 was released from property owned by Dow Chemical in the Montezuma Hills. During the two tests where CBrF 3 was used, the SF 6 tracer was emitted upstream; SF 6 was released from Martinez during Test 2, and from Pinole during Test 7, in hopes of determining the origin and dispersion of the air passing over the Montezuma Hills. release of SF 6 from Pinole. The final test involved only a Releases from the Montezuma Hills covered all four meteorological periods; the releases from Martinez and Pinole were conducted during Sea Breeze and Pre-Sea Breeze conditions. Tracer release rates, release times, and release locations are given in Table 2. Tracers were released at a constant rate from gas cylinders using a manifold, regulator, and large-volume flowmeter; the gases were released into the air at a height of approximately 5 m. The release system was secured in a truck which provided an easy means of moving the release system to new release locations. Release rates were set using the regulator and calibrated flowmeter. Average release rates were determined by weighing the gas cylinders before and after each release and averaging the weight of the released tracer over the release period. The total amount of tracer released was within 2% of that determined from the calibrated rotameter. The release system was continuously monitored during every release. During each test day, air samples were collected via automobile traverses of three to five , two- person teams . Automobile traverses 120 TABLE 2 TRACER RELEASE DATA Date Test Location of SF6 Release Release Period (PDT) Release Rate (grams/sec) 8/31/76 1 Montezuma Hi 11 s 1200-1700 10.6 1.01 tons/day 9/2/76 2 Martinez 1100-1600 Montezuma 11. 4 1. 08 tons/ day Hi 11 s 9/5/76 3 Montezuma Hil 1s 0000-0500 9. 5 0.90 , tons/day 9/6/76 4 Montezuma Hi 11 s 1800-2300 10 . 8 1.03 tons/day 9/9/76 5 Montezuma Hills 1130-1330 10 . 7 1.02 tons/day 9/ 10/76 6 Montezuma Hj 11 s 0600-1100 10.5 1.00 tons/day 9/13/76 7 Pinole 0600-1500 Montezuma 11. 5 1. 09 tons/day Hi 11 s Location of CBrF 3 Release Release Period Release Rate 1300-1500 16.6 1.58 tons/day 0900 - 1100 16.0 1300-1400 16.0 1. 52 tons/1ay 9/ 14/76 8 Pinole 0730-1300 10 . 9 1.04 tons/day 121 were made by having the passenger in each car take IO- second grab samples in 30 cm 3 plastic syringes. Generally, samples were collected every 0.1, 0.2, 0.5, or 1.0 miles, depending upon the distance from the release point and the steadiness of the wind. Traverse paths were determined in the field from real time wind data. Analysis of samples collected during the early part of each release was used to establish traverse routes for the rest of the test. Personnel from Meteorology Research, Inc. (MRI) obtained air samples in a manner similar to the automobile traverse teams from an airplane traveling downwind of the release at various heights and locations. Air samples were also obtained during vertical spirals through the mixing layer; samples were typically taken at vertical intervals of 100 or 200 feet. A total of 56 sequential hourly air samplers were located at 28 sampling sites; the locations of the sites are shown in Figure 1. Equipment which permitted the sequential collection of hourly averaged air samples for 12 hours was positioned at 14 locations, primarily along Highway 99. An additional 14 locations along Highway 160 (10 km east of the Montezuma Hills) were designated for the positioning of batterypowered samplers which permitted the sequential collection of hourly averaged samples for 3 hours. Tests 1 and 2. The 3-hour samplers were only used during During the eight tests, 4508 automobile traverse samples, 1258 airborne traverse samples, 330 airborne spiral samples, 1721 hourly averaged 12-hour board samples, and 153 hourly averaged 3-hour board samples were collected; the total number of samples collected was 7970. 122 Air samples were analyzed for SF 6 and CBrF 3 using electron capture gas chromatography. The tracers were separated in a stainless steel column (39" x Oo25" OD, 0.18" ID) packed with 80-100 mesh 5A Molecular Sieve. With prepurified nitrogen flowing at 57 cc/min (10 psig), the SF6 eluted in 18 seconds and the CBrF 3 eluted in 35 seconds. Both column and detector were operated at ambient temperatures. Additional details concerning the gas chromatography are available elsewhere ~(Drivas, 1974; Lamb, 1977). A typical chromatogram is shown in Figure 3o Twelve chromatographs and two digital integrators were set up in a room at the California Holiday Lodge, Fairfield, California. Only eight of the chromatographs were used during the test; four were used to analyze for SF 6 alone and four were set up to analyze for both tracerso Air samples were returned each day to the lab for analysis. All of the samples from one test were analyzed before the next began. Calibration was done using an exponential dilution method. Calibration results show that concentrations down to 10- 12 parts SF 6 11 per part air and 10- parts CBrF 3 per part air could be detected at a signal-to-noise ratio of better than 3 to 1. Typical calibration curves are shown in Figure 3. The gas chromatographs were calibrated before and after the field test period. During the tests, the instruments were cross-checked periodically for reproducibility. The calibration results changed by approximately 7% among the gas chromatographs used to analyze SF 6 alone. Calibration results for the remaining gas chromatographs changed by 25% for SF 6 and 20% for CBrF 3. Degradation of the columns and detectors due to atmospheric contaminants in the air samples was the probable cause for the changes in the 123 zg 10-e CBrF3 ( Sf 5) SFs 10-9 (CBAF3 . 10-10 I I I 60 30 0 Ti me (sec) 10- 11 10-12 1o-13 .___-'--"-........................~_._.~.....__~.L-&..l~.._-'--"-............ ........_--'--L..1....1..LLJ.....___.__L-L.L..u..LM 10 1 103 104 . 10 5 INTEGRRT(jR AREA 106 107 BKL 8/21/76 Figure 3. Typical dual tracer calibration curves and typical dual tracer chromatogram (inset): [SF 6 J = 6 ppt , [CBrF 3J = 310 ppt 124 calibrationso Previous experience has shown that the presence of halogenated solvents in air samples can, over an extended period of use of the chromatograph, contaminate the detector and column. Uncertainty in the tracer concentrations is estimated to range from less than 5% for most of the samples, to no more than 25% for a limited number of samples. Meteorological data, consisting of surface and elevated wind speeds, wind directions, mixing height estimations, cloud cover reports, and standard deviations of the horizontal wind, were collected from a variety of agencies. Figure l. Surface and upper air data stations are shown in Further discussion of the experimental procedure and presentation of all tracer and meteorological data are available elsewhere (Lamb and Shair, 1977). 125 4. Discussion of Results The extensive amount of tracer data collected during automobile and airborne traverses and airborne spirals, and from the hourly average sampling network allowed very detailed resolution of the tracer plumes in both time and space. For example, as shown in Figure 4 for Test l, tracer released from the Montezuma Hills was transported east across Highway 160 and then southeast into the Stockton-Tracy area. Data from repetitive traverses along Highway 99 indicated that the tracer plume reached the Stockton area shortly before 1700 PDTo According to hourly averaged data collected at stations along Highway 99, the entire tracer plume passed through the Stockton region . between 1700 and 2000 PDT. Similar data collected during the remainder of the releases from the Montezuma Hills provided detailed evidence that (excluding Test 5) tracer plumes were carried southeast through the Stockton area into the entrance to the San Joaquin Valley. During Test 5, the typical westerly marine flow was replaced by weak northerly winds which carried the tracer from the Montezuma Hills directly south past Antioch towards Livermore. During the evening and nighttime tracer releases, as illustrated in Figure 5, much higher concentrations were observed at greater downwind distances than during the daytime releases. In dual tracer releases during Test 2 and Test 7, emissions from Martinez and Pinole were apparently carried in wide spread plumes south of the Montezuma Hills into the Stockton-Tracy area. The hourly averaged crosswind profiles given in Figure 6 show that the sharp, CBrF 3 plume emitted from the Montezuma Hills during Test 2 was superimposed upon a 126 [fil "' \ "'i (- ... MONTEZUMA HILLS .!J Pl ~ a:: NTW r>--------..... 4 \ - - - - I BYRON ~ > iii ~>----- (131 Figure 4. OuvERMCRE Overview of automobile SF 6 data , Test l, 8/31/76. No. 6 1630 - 1641 PDT, SF 6(max) • 1223 ppt No. 7 1701 - 1753 PDT, SF 6 (max) = 64 ppt SF released from the Montezuma Hills from 6 1200 - 1700 PDT. 127 ~ 3 n rl \LJ LJ 1 2 SACRAMENTO 3 DAVIS 22 Tll1[ 10 22 Tll1[ 10 22 \ ';,.. 22 / * - ,-,--..:-,._ '-'---'--~<---1 i----+--- Tll1[ 10 7 I MONTEZUMA HILLS 10 (] L r TJl1[ ~ 3 ,\1r so ~ ~ ~l---=-~8~1n~9 I\~;-,,:,;' ~W 22 Tll1[ 10 22 Tll1[ 10 12 22 Tll1[ 10 0 L 1VERMC'RE ...____.___________________ Figure 5. {D 22 TJ11[ -i ------~ 10 ,_,_ ___. Overview of hourly averaged SF6 data, Test 3, 9/5/760 SF6 scale: 0-1000 ppt. Sampling time: 2200-1000 PDT, 9/4-5/76. SF6 released from the Montezuma Hills between 0000- 0500 PDTo 128 24 TIME: 1336-14 36 PDT [ SFs] ppt 12 24 [ SF6] ppt 12 1400 TIME:1436-1536 PDT 2800 [ CBr F3] 1400 ppt 18 24 TIME:1536-1636 PDT [SFs] ppt 12 2800 [ CB rF3] 1400 ppt OOL.a~~.__.__.~11-------_._..i.._._-&-_&J 0 9 18 27 DISTANCE N. OF HWY 4HWY 160 JUNC. (KM.) Figure 6. Crosswind hourly SF5 (.a) and CBrF 3 (e) profiles measured along Highway 160 during Test 2, 9/2/76. SF6 was released at Martinez from 1100-1600 PDT (33 km upstream); · CBrF3 was released from the Montezuma Hills from 1300-1500 POT (10 km upstream). 129 portion of the wide, SF 6 plume emitted from Martinez. During Test 7, the SF 6 plume emitted from Pinole diverged south past the Walnut Creek area and east through Pittsburg. It appeared that portions of the SF 6 plume were ultimately carried as far as Stockton. Mass balances were performed with the automobile and airborne traverse data. The accuracy of these analyses depended upon the accuracy with which the vertical profiles of the tracer concentration and wind speed could be determined. In cases where no airborne data were available and the tracer plume could not be assumed to be vertically well-mixed (i.e., under stable evening and nighttime conditions), as expected, the mass balances widely overestimated the percent tracer observed in a traverse. However, excluding the stable cases, the overall average percent tracer observed in 43 crosswind traverses was 95%. Hence, essentially all of the tracer was accounted for by the airborne and automobile traverse data. During Tests 7 and 8, two sets of airborne traverses provided detailed concentration profiles at several different heights. The airborne crosswind profiles from Test 8 are shown in Figure 7. Assuming that the tracer observed in each traverse was vertically wellmixed over a discrete portion of the mixing layer, we found that the tracer data from these two sets of data accounted for 113% and 102% of the tracer released. Considering the assumptions used in the analysis, these estimates are remarkably accurate. Due to the steadiness of the winds, the plume trajectories at 10 km downwind of the Montezuma Hills were found to be quite similar from test to test. Although the lateral dispersion decreased with 130 TEST CFL8 TMVERSE II ORTE• 9/11&/76 Tlf1E• FLT. ( N> 1&26 O!N2-095S TRAVERSE MUTE• COFIELIR sa.1111 TO 1680 fH> HWY 2'&(MRJ«JT CREEK> "' l- a.. co a..""-/ • Uco z tO u 13 26 DISTANCE AL~NG TRAVERSE (KM.) ~ l"' - 39 TEST CFL8 TMVERSE 5 FLT. ( N> 3(N ORTE• 9/11&/76 TlHE• 0959-1012 TRAVERSE MUTE• 1600-HI~Y 2'l ("8..NUT CREEK> NORTH TO COAOELlA a.. a.. 'Vi u • z tO u 13 26 DISTANCE AL~NG TRAVERSE (KM.) g 39 TEST CFL8 TMVERSE 6 FLT. ( N> 182 DATE• 9/11&/76 TlNE• 1016-1028 TRAVERSE MUTE• CtHELIR sa.1111 TO 1690 fH> HWY 2'&CWRUtJT CREEK> ::1' "' la.. a.. ""-/ 0 0 aN u z tO u 13 26 DISTANCE AL~NG TRAVERSE (KM.) Figure 7. 39 SF6 concentration as a function of distance along an airborne traverse taken at three altitudes 17 km downwind of Pinole release site. Note the difference in concentration scales. 131 increasing atmospheric stability , the plume centerline was found to pass over Highway 160 within a crosswind zone of 2. 1 km (excluding Test 5). A similar analysis of the data obtained along Highway 99, 50 km downwind, indicated that the average position of the plume centerlines was 5 km south of Stockton; the standard deviation associated with the average position was~ 14 km . A straight line can be drawn from the release point in the Montezuma Hills to the average centerline position of the plume crossing Highway 160. The extension of this line intersects Highway 99 approximately 2 km north of the observed centerline position. The average plume centerlines from the Montezuma Hills release site to Highway 160 and from Highway 160 to Highway 99 and the standard deviations associated with the centerline locations are shown in Figure 8. On the average, plumes emitted from the Montezuma Hills during the test periods were transported southeast directly over Stockton. The area enclosed by the standard deviations of the plume centerline positions corresponds to the area of major impact which was observed during t~e tracer tests. The magnitude of this area appears to be closely related to the area swept out by the average value of the horizontal standard deviation of the wind, 0 8 , during the field study. The angle associated with the standard deviation of the centerline locations is 8°; the average value of 0 8 measured at the Montezuma Hills release site during the seven Montezuma Hills releases equaled 9°. However, the average wind direction measured at the Montezuma Hills does not coincide with the average centerline locations along Highway 160 and Highway 99 . The average wind direction was 275°; 132 N 1 • AIR SAMPLERS: (3-Hours) ill] AIR SAMPLERS: ( 12- Hours) * O ELK GROVE@] RELEASE SITE 10 KllQM[T[RS 10 MIUS Figure 80 Average plume center.line positions for plumes emitted from the Montezuma Hills. The outside lines correspond to the standard deviations associated with the average centerline positions. The arrow indicates the average direction of the wind measured at the Montezuma Hills during the 7 tracer releases . 133 the angle associated with the average centerline was 285°. It appears that curvature in the wind field immediately downwind of the Montezuma Hills causes plumes to turn to the southeast more than expected. Although the winds are generally very steady at the Montezuma Hills, the wind directions measured there apparently cannot be used to accurately predict the location of plume impact at distances downwind greater than a few kilometers. A further result of the steady nature of the winds over the Delta Region is the relationship observed between maximum tracer concentrations in 10-second and hourly averaged data. The corrmonly used Hino (1968) correction suggests that CHR;c 10s = 0.18 where CHR is the hourly averaged centerline concentrations and c10s is the 10-second averaged centerline concentration. This relationship can be checked easily for the Delta Region by comparing the automobile traverse results with the hourly averaged data obtained during the hours in which traverses were completed. The overall average of the experimental ratio was CHR;c 10s = 0.7. No significant differences in this value were apparent among the various test periods. The results suggest that the Hino correction should not be used for describing variations of concentrations due to variations in sampling time in coastal zones where the marine flow is quite steady. The horizontal dispersion parameter, cry, was calculated for each crosswind traverse; the direction of the wind was taken into account in these calculations. These values were used to compare the rates of dispersion with those based upon Pasqui11-Gifford stability classes as given by Turner (1970). The comparison indicated that atmospheric 134 stability generally decreases with increasing distance downwind from the Montezuma Hills. The results of the dispersion analysis for the releases from the Montezuma Hills are summarized in Figure 9. It appears that during the afternoon the heating of the land decreases atmospheric stability and dominates the stabilizing effects of the jet of air passing over the Montezuma Hills. At other times, the steadiness of the winds causes horizontal dispersion (at short distances downwind of the release) to be less than that predictedQ The horizontal dispersion of plumes emitted from Pinole and Martinez during the PreSea Breeze and Sea Breeze periods was found to be similar to that associated with plumes emitted from the Montezuma Hills during the Sea Breeze periodQ Vertical dispersion parameters, estimated from surface traverses using mass balance arguments and the Gaussian plume expression, were found to be similar in value to those suggested by Pasquill-Gifford. However, as shown in Figure 9, for releases from the Montezuma Hills under nighttime conditions, vertical dispersion was relatively constant from 7 to 50 km downwind. It appears that vertical mixing is extremely limited under these conditions. In Part I of this series, the Gaussian plume model was found to yield reasonably accurate estimates of centerline tracer concentrations when dispersion terms were changed to account for the effects of rough terrain. In Part I, the use of either experimental dispersion values or Pasquill-Gifford dispersion values one stability class less stable than that inferred from wind speed and insolation during the test produced accurate predictions of tracer concentrations. 135 l0ll en a: w w I- . 10 L 3 >-I a: ~ en ....... A 102 10 1 102 B C D E f 103 x. METERS l0ll 105 DOW AUTO fRRV. 105 l0ll en a: w w I- L . 1Q3 N I A a: ~ ....... en 102 10 1 '-----'----'----L--J...l...J...J...U._-1....-1.......LJ...L..L...L.J..L_J........J........L.L.L.LU.J 1Q3 lQ'l 10 2 X, METERS DOW AUTO TRRV . Figure 9. Horizontal and vertical dispersion parameters calculated from automobile traverse tracer data obtained during the meteorological periods designated: Sea Breeze (SB) , Sea Breeze Tail (SBT), Nighttime (NT), and Pre-Sea Breeze (PSB)o PasquillGifford dispersion parameters are also shown. 136 The Gaussian plume model is often used by industrial planners and regulatory agency personnel as a means of estimating the impact of industrial emissions. Because of the widespread application of the Gaussian plume model, it is worthwhile to compare its predictions of tracer concentrations with observed values. In order to account for the effects of an inversion layer upon dispersion, we will use the Gaussian expression given by Bierly and Hewson (1962). For ground-level concentrations, their expression simplifies to: n=N 2 exp[-½(~) J (1) ~N For application in the Delta Region, cry and oz' the horizontal and vertical dispersion terms, were taken from best-fit lines through the dispersion data for each test; H, the effective stack height, was set equal to the release height, 5 m. The release rate, Q, was assumed to equal the average release rate. The wind speed, u, was obtained by vector averaging hourly wind data from all stations downwind of the tracer release point over the time of interest. The mixing height, L, was determined by averaging the hourly heights from the available data for each test period. Ground-level, centerline concentrations were predicted for every tracer release; typical results are shown in Figure 10 for Tests l and 4. Gaussian model predictions using Pasquill-Gifford values of cry and oz are also shown. Atmospheric stability conditions were estimated according to Turner (1961). Generally, the model predicted reasonably . ........ . .... . ....... TEST 4 TEST CAL4 9/ 6/76 ATM. STABILITY CLASS• E RELEASE SITE• DOW MIXING HEIGHT (M) 510. HIND SPEED (M/SEC) 3.3 HOURLY DATA~ TEST I -----~----------------------- □ TEST CALl 8/31/76 ATM. STABILITY CLASS• C RELEASE SITE• DOW MIXING HEIGHT (M) 960. HIND SPEED (M/SEC) 3.ij HOURLY DATA rn Figure 10. □ eWNWIND DISTANCE CKM) Ground-level, centerline hourly (o,~) and 10-second (■,•) averaged tracer concentrations compared with ground-level centerline concentrations predicted using the Gaussian plume model. Solid curves were calculated using experimental values of oy and Ozo Broken curves were calculated using Pasquill-Gifford values of ay and o2 . 10° ~ - - - - - - - - - - - - + - - - - - - - ' - - - - - - - - - - L - - - - - - - - ' 0 20 ij0 60 80 100 10 1 102 103 PPT. CONC. 1014 105 106 w ......, 138 accurate concentrations with either set of dispersion terms for releases from the Montezuma Hills. The estimated concentrations served as an upper bound to the observed data for releases from Martinez and Pinole. In view of the divergence of tracer across the Delta Region during these tests, this overestimation is to be expected. The concentration curves predict the tracer data most accurately under stable evening and nighttime conditions. During the daytime in Test 5, the observed data were perfectly predicted using either set of dispersion terms. However, weak northerly winds prevailed during the day, and the strong jet of air over the Montezuma Hills was not present. In the absence of the strong westerly flow, the dispersion process followed the classical open country dispersion pattern. The comparison of experimental and predicted centerline concentrations indicates that averaging the wind speed and mixing heights over the available data is a reasonable, straightforward method of supplying the necessary input data to the model. In view of the large wind velocity gradients present in the region, the apparent usefulness of average input values is somewhat surprising. The applicability of the Pasquill-Gifford sigmas in the Delta Region, as opposed to their accuracy in the region treated in Part I, may be due to the much smoother terrain which exists in the Delta Region. Equation (1) can also be used to predict crosswind concentration profiles at various downwind distances. As shown in Figure 11 for two traverses taken during Test 4, the predicted profiles can yield extremely accurate descriptions of the observed tracer data. In this case, wind speeds were taken from the nearest wind station to 139 TEST CALij TRAVERSE 2 DATE • 9/06/76 TIME• 2130-2217 TRAVERSE. ROUTE• SOUTH ON HWY 99, W EST ON HW Y 120, 1205 STARTING POINT• JUNC. HWY 12 ANO HWY 99 ATM. STABILITY CLASS• E SOURCE• OOH X <KM> 50.9 L <M> 510 . U CM/SEC> 2.6 SF6 ORTA C) C) - ('\I l- a_ □ - - - Calif. Delta a...~ "-/ Pasquill - Gifford 22 66 ijij DISTANCE AL~NG TRAVERSE CKM.) TEST CAlij TRAVERSE 3 DATE• 9/06/76 TIHE• 2132-21ij0 TRAVERSE. ROUTE• SOUTH ON HWY 160 STARTING POINT• THE BEAN POT ON HWY 160 ATM. smBILITY CLASS• E SOURCE• OOH X <KM> 7.3 L <H> 510. U (M/SE.C> 7.1 SF6 ORTA C) C) C) - ('\I r--. 1-a Calif. Delta a... □ a... C) "-/ (XJ • C) u □ z □ ~ ::r u --Pasquill-Gifford □ i..-111111.....111111-... . . . . . . . . . . . . . .'1£.:.._~- -........- -...- - -- 3 6 -___J 9 DISTANCE N. ~F HWY 4-HWY 160 JUNC. (KM . ) Figure 11. Crosswind tracer profiles compared with crosswind profiles predict~d using the Gaussian plume model and either experimental California Delta dispersion terms or PasquillGifford dispersion terms. The wind speeds for Traverse 2 and Traverse 3 were averaged from measurements taken at Venice Ferry and the Dow site, respecti vely. 12 140 the traverse. Calculations show that simil ar profi les can be predicted accurately for most of the automobile traverse data . Further analysis using Equation (1) with the plume rise model of Briggs (1969, 1972) and stack parameters typical of facilities proposed for construction in the Montezuma Hills show that after a characteristic distance downwind of the release , realistic variations in the effective stack height do not significantly influence groundlevel concentrations. The tracer data indicate that this characteristic distance is less than about 10 km during the day. However, at night when the extent of vertical mixing is minimized by stable conditions, this characteristic distance increases. For example, under stability class E, an effective stack height of 107 meters yields ground-level concentrations less than half those from a ground-level release for distances up to 20 km downwind. As a means of interpreting the results of the California Delta tracer tests in terms of existing or projected pollutant emissions, measured tracer concentrations tan be converted to equivalent pollutant concentrations for specified pollutant emission rates . The conversion of concentrations implies that either the pollutan t is essentially unreactive or that it reacts so rapidly that its produc t disperses like a tracer . For example, carbon monoxide can be considered unreactive over the time scale involved in tracer transport , and sulfur dioxide has been shown to have a relatively slow (15%/hr) reaction rate in coastal regions (Roberts, 1975) . In cases where oxidant con- centrations are high, nitrogen oxide is converted very rapidly to 141 nitrogen dioxide. For these cases, the pollutant concentration Cp can be determined from the tracer concentration Ct as foll ows : (2) where Qt and QP are the tracer and pollutant rel ease rates, respectively; Mt and MP are the tracer and pollutant molecular weights, respectively. Peters and Richards (1977) have devised a means of mathematically separating descriptions of the chemistry and the dispersion of pollutants in cases where pollutant concentrations can be assumed to be in local equil i brium. Their method can be easily extended to yield a straightforward means of converting tracer concentrations to equivalent pollutant concentrations while accounting for associated pollutant chemistry. As described in Appendix A, a series of linearly algebraic equations can be obtained in the form (3) Peters and Richards defin e the function ¢ij as: (4) ¢·. = v.C. + v.C. 1J J 1 l J where Ci and Cj are the concentrations of reactant i and product j, respectively; v. and v. are the related stoichiometric reaction 1 coefficients. J The term ¢ij is a corresponding function of the background concentrations. The expressions represented by Equation (3) and the chemical equilibrium expression can be solved simultaneously to yield the concentration of each species. The second term on t he 142 right side of Equation (3) describes the dispersion of pollutants as if each species were inert. Since the tracer disperses as an inert species, Equation (2) can be directly substituted for the dispersion term in Equation (3) to yield _ o CtMt [V. Q.1 ¢ .. - ¢ .. + -Q- J -M + lJ lJ t . 1 V. Q·] ...:J.. , M . J (5) 0 For a set of reactant and products, assuming the emission rates Qi and Qj are given, the series of linearly independent expressions given by Equation (5) relates observed tracer concentrations directly to reactive po11utant concentrations. The Gaussian plume model used with the empirical values of cry and oz provides a simple, accurate means of estimating pollutant concentrations in the California Delta Region. However, Pasquill and Smith (1971) have reminded users of the model that the preferred modeling approach employs data on turbulent fluctuations of the wind to determine values of cry and oz. In view of this reminder, we have examined the relationship between the horizontal standard deviation of the wind direction, 0 8, measured at the Montezuma Hills release site, and the horizontal standard deviation of the concentration profile, cry' measured at vari ous downwi nd distances. Islitzer and Dumbauld (1963) found from a similar analysis that (6) if 0 8S,T (in radians) was based on wind directions averaged over period 143 sand sampled during time interval T. Hay and Pasquill (1959) suggested that the value of s should be determined from the proper portion of the scale of turbulence by using s = T/S where Tis the transport time, x/u, and Sis the ratio of the Lagrangian to Eulerian time sca l es, taken to equal 4. Pasquill (1974) noted that the sampling interval T shou ld equal the duration of the tracer release. In the Delta field study , values of s were of order 9 minutes at 8 km and 63 mi nutes at 50 km. Wind measurements at the release site were available in terms of 5-minute average wind directions. The 5-minute values can be summed to give 10-minute and longer averaging times. These average values can then be used to determine the standard deviation of the wind during each release period. Vertical fluctuations of the wind were not measured. Values of cry obtained with the values of cr 0S,T correspond to dispersion observations taken during the release periodT. In the Delta field study, this would yield 5-hour averaged values of cry· Since we have 10-second averaged and hourly averaged values of cry, the experimental va l ues of cry are expected to be less than those predicted from the values of cr 0. This is, in fact, the result obtained by plotting values of cry versus xcr 0; the least-squares best fit through the data points for the 10-second averaged profiles is given by cry= 0.85 x cr 0 - 550 m with a best-fit correlation coefficient equal to 0.83. (7) A similar analysis using values of cry determined from the hourly averaged data measured along Highway 99 and Highway 160 yields the best-fit curve: cry= 0.93 x cr 0 + 1500 m (8) 144 with a best-fit correlation coefficient equal to 0.40. Although the scatter from the hourly averaged data is large, both sets of relationships indicate that values of cry can be determined rather simply by acquiring sufficient data on the fluctuations of the wind. In this manner, input dispersion data for the Gaussian plume model can be obta i ned rather easily. Although vertical wind fluctuation data were not ava i lable in these tests, presumably such data could be used to d~termine cr 2 in a similar manner. Pasquil l (1974, 1976) has developed a new set of dispersion terms which are given by cry= F(x) • x • cr 0 (9) where F(x) is a correction term which ranges from approximately 0.8 for small x to approximately 0.2 for large x. In comparison, the constants given in Equations (7) and (8) represent overall correction terms for distances from approximately 4 km to 50 km. Hay and Pasquill (1959) noted that there is some uncertainty associated with the value of Bused to calculate the sampling period . Their analysis cf available dispersion data suggested B = 4. The results presented for the Delta Region also suggest that B ~ 4. Islitzer and Dumbauld (1963) found B = 5 gave a better fit of dispersion data collected over flat terrain in Idaho. More recently, Pasquill and Smith (1971) reported that B appears to be inversely proportional to the intensity of turbulence, B = p/i. They suggested that p = 0.44 is a reasonable value of the constant. This value results from consistency calculations between statistical and similarity theory for 145 eddy diffusivity under neutral conditionso Since 0 0 ~ i, we can use wind fluctuation data measured at the Montezuma Hills to determine an average value of the proportionality constant, p = S. 0 . For the 0S,T related to the 10-second averaged data, the average values of 0 0 S,T value is p = 0.61 ~ 0.34, and for values of 0 0 related to the hourly averaged tracer data, p = 0.49 ~ 0.30. The overall average for both sets of cr 0 data is p = 0.55 ! 0.32. S,T Within the scatter of the S,T data, these values are consistent with the value suggested by Pasquill and Smith. A primary goal of the tracer field study was to obtain sufficient tracer, meteorological, and air quality data for the validation of large-scale atmospheric diffusion models. Such validations have been started and will be reported upon at a future date. One of the tests of a model's validity is to predict experimental tracer trajectori es using available meteorological data. For example, the two-dimensional mass balance equation can be solved numerically using hourly averaged wind data to give hourly wind vectors for each square in a grid system. The hourly wind fields represented by these vectors can then be used to determine forward ai r parcel trajectories, given a starting point and time. Methods for computing the wind fields and the trajectories have been developed and kindly provided by William Goodin (1977). As an example of the procedure, the numerical solutions of the mass balance equation were determined for a 32 x 30 grid system of 4 km x 4 km squares covering approximately the area shown in Figure l. 146 Surface wind data for August 31, 1976, from 1200 to 2300 PDT were used to calculate hourly wind fields. A typical hourly wind field is shown in Figure 12. The major features apparent in the complex flow patterns which occurred during the first tracer test include the turning of the marine air into the northern Bay Area and the subsequent channeling of air through the Carquinez Strait. From 1200 to 1500 PDT, flow over the Delta was directed to the southeast; after 1600 PDT, the winds at the Montezuma Hills straightened towards the east. Inmediately downwind of the Montezuma Hills, the flow diverged to the northeast and southeast. From 1700 to 2000 PDT, an exceptionally strong flow to the south appeared in the Stockton-Tracy area. Upper air data obtained for the first release day were used to determine the average wind fields in three layers extending from 300 to 900 feet, from 900 to 1500 feet, and from 1500 to 3000 feet above the surface. The surface wind field was assumed to extend to 300 feet above the surface. The resulting three-dimensional wind field is presented in Figure 12 for 1700 PDT. Comparison of the upper air wind fields with the surface wind field at 1700 PDT readily illustrates the effects of the complex topography upon the gradient westerly flow. Above the surface layer, flow is uniformly from the west throughout the region. Note that if digital topography data were used to specify the boundary condition at the surface (assumed to be flat in this analysis), then more complex flow would probably appear in all the upper air wind fields. 147 SA- = . . ."'"' . .. - P~ z i t - ·S:,· • ~ . -SF,!t- l_: .,, . . .. . ~ ~ LV1t' » "' ·- 300-900 Ft, 1700 PDT . . . . . - . . sA*---. . - . . . . . .. . .,. . . . . . . . . . : : : : '.::::: ::),1_tl~ -P~~Mi.- ~ MHlt .. ~ P~Mi.- ~:;.Si=!i , , .. _,___.,, s~:: _,___.,, -=-~- ::-:SF* ~ 90 0-1500 Ft., 1700 PDT Figure 12. Lv 1500-3000 Ft ., 1700 PDT Hourly surface and upper air wind vectors calculated from data measured between 1700 and 1800 PDT, 8/31/76. Reference locations: Martinez (MZ), Montezuma Hills (MH), Pinole (PN), Sacramento (CA), San Francisco (SF), Stockton (SK), Livermore (LV). 148 The terrain surrounding the Bay Area reaches to over 1000 feet in places; Mt. Diablo rises to an elevation of 3849 feet {16 km south of Pittsburg). Forward hourly air parcel surface trajectories beginning at the Dow site in the Montezuma Hills were calculated from the surface wind fields for starting times from 1200 to 1700 PDT. The surface tra- jectories shown in Figure 13 for starting times 1200 PDT and 1700 PDT are typical of the calculated transport paths. In every case, surface· trajectories moved southeast from the Montezuma Hills to the StocktonTracy area. The effects of stagnant conditions in Stockton at 2300 PDT are apparent in the later trajectory. The paths and transport times of the surface trajectories are in excellent agreement with observations made from the traverse tracer data along Highway 99. Tracer and calculated trajectories both ap~eared to pass between Tracy and Stockton beginning at 1700 PDT. However ; the maximum hourly averaged SF6 concentration occurred between 1800 and 1900 PDT 4 km north of Stockton. Crosswind hourly profiles indicated that the SF 6 plume fluctuated through a zone extending from Lodi to Tracy between 1700 and 2000 PDT. Although the calculated surface trajectories are in agreement with the automobile traverse data at 1700 PDT, it appears that the impact zone was wider than that predicted from the trajectories. We have seen that the calculated air flow above 300 feet was directed almost due east throughout the Delta Region. Airborne tracer data collected during the sixth tracer test indicated that tracer can be mixed vertically at least as high as 600 feet after being transported from the Dow site to Highway 99. One may envision a tracer plume at the 149 N 1 [IQ] AIR SAMPLERS: (12-Hou,s) * RELEASE SITE 10 KILC)'lll[T[FtS ro MILES Trajectories beginning at 1200 PDT, Test 1, 8/31/76 N 1 [IQ] AIR SAMPLERS: (12-Hours) * RELEASE SITE 10 KILOfrol[T[~ 10 •1u:s Trajectories beginning at 1700 PDT, Test 1, 8/31/76 Figure 13. Forward air parcel trajectories. Solid bar along Highway 120-1205 in top figure corresponds to the location of tracer plume observed during an automobile traverse taken from 1701 to 1753 PDT. SF6 was released from the Montezuma Hills from 1200 to 1700 PDT. 150 surface moving southeast towards Tracy. As tracer is mixed vertically due to the afternoon heating of the land, upper air flow from the west carries tracer east across Highway 99 in a zone from Lodi to Tracy. As Neiburger (1955) suggested nearly 20 years ago , including the effects of the winds aloft in the numerical analysis would improve the overall agreement between trajectories and tracer data . The surface wind fields were also used to construct air parcel trajectories beginning at Martinez and Pinole for two release times during the first tracer test . Even though tracer releases were not conducted from these points during the first test, it is of interest to determine the transport paths of pollutants emitted in the northern Bay Area for the sake of comparison. The results for trajectories starting at 1200 and 1700 PDT from each point are shown in Figure 13. There appears to be a significant difference in paths initiated at 1200 and those started at 1700 PDT. The earlier releases from both Martinez and Pinole traveled south of Concord into the Tracy area. Strong westerlies at 1700 PDT forced the trajectories beginning at that time to move further east through Pittsburg before turning southeast towards Tracy. Neither set of trajectories was carried directly over the Montezuma Hills. These patterns are very similar to those observed during Test 2 and Test 7. The trajectories which began at 1200 PDT reached Tracy at about the same time as those released from the Montezuma Hills at 1600 and 1700 PDT. Both the trajectories initiated from the Pinole-Martinez area and the trajectories started from the Montezuma Hills indicate that pollutants 151 emitted from these points into the afternoon sea breeze can be transported into the San Joaquin Valley. 152 5. Summary and Conclusions The main purpose of this investigation was to demonstrate techniques of using one and two atmospheric tracers to probe the transport and dispersion of pollutants over long distances in areas of complex topographical and meteorological conditions. These techniques were applied in 8 tests during September, 1976, in the California Delta Region. The meteorology of this coastal area is controlled by land-sea variations in temperature and pressure; surface ' flow patterns are strongly influenced by the rough coastal topography. The prevailing flow of marine air channels through the Carquinez Strait and diverges across the Delta Region into the Sacramento and San Joaquin Valleys. Although the flow patterns are very complex, these patterns were surprisingly reproducible and, in some cases, very steady. In fact, plume trajectories were so steady up to 50 km downwind that the commonly used Hino sampling time correction grossly underestimated hourly averaged tracer concentrations computed from 10-second averaged concentrations. A comparison of experimentally determined dispersion parameters with those associated with Pasquill atmospheric stability classes indicated that atmospheric stability generally decreases with increasing distance downwind from the Montezuma Hills. In spite of the complex meteorology and terrain, estimates of tracer concentrations based upon the Gaussian plume model were found to be reasonably accurate. A method was developed for utilizing tracer data and pollutant emission rate data to calculate equivalent pollutant concentrations in 153 cases where local chemical equilibria can be assumed as outlined by Peters and Richards (1977). A mass balance analysis of the tracer data accounted for essentially all of the tracer released. In cases where airborne tracer data were available at several different heights, the mass balance estimates were very accurate. The relationship between the horizontal standard deviation of the wind, cr 0, and the horizontal dispersion parameter of the plume, cry, was found to follow cry~ 0.85 x cr 0 where xis the downwind distance. Trajectories and estimated arrival times of air parcels based upon Goodin's (1977) numerical solution to the two-dimensional mass balance equation were found to be in excellent agreement with the tracer data. These positive results, along with the negative results in Part I of this series, indicate that when adequate meteorological data are available, air parcel trajectories based upon numerical solutions to the mass balance equation are more accurate than those drawn without the aid of a computer. • 154 APPENDIX A Conversion of Tracer Concentrations to Equivalent Pollutant Concentrations for Cases of Local Chemical Equilibrium Peters and Richards (1977) have devised a means of mathematically separating descriptions of the chemistry and the dispersion of pollutants. When atmospheric diffusion equations, including the reaction terms, for a reactant and product pair are summed, the reaction terms cancel and an equation of continuity can be written in terms of a function ¢ · . = V· C. + V• C. • 1J J 1 1 (A-1) J The terms Ci and Cj are the reactant and product concentrations, respectively, and vi and vj are the corresponding stoichiometric coefficients. With no reaction terms and assuming constant eddy diffusivities, the continuity equation can be solved for a continuous point source to give: Q.1 . ( ) 2 ¢1J .. = ¢1J ~. + J exp [ -½ !-Y ] exp [-½ TTOYO U v ·z 2 (:z) ] (A- 2) where ¢ij corresponds to background concentrations, and Qij = vjQi + v.Q .. 1 J For "n" reactants and "p" products, n+p-1 linearly independent functions¢·. can be written as above. lJ If the associated chemical reactions are fast, local equilibrium can be assumed to exist and an equilibrium expression can be written in terms of the reactant and product concentrations. The~'f-'1J .. functions plus the equilibrium expression provide n+p independent algebraic equations which can. be 155 solved simultaneously for the concentrations of each reactant and each product. These expressions provide a unique means of converting measured tracer concentrations to reactive pollutant concentrations since the dispers i on portion of Equation (A-2) represents the concentrations of the reactant-product pair i f they disperse l i ke an iner t species. We can conver t measured tracer concentrations to pollut ant concentrations using Equation (2) from Section 4 of the text, where it is assumed that the pollutant disperses like a tracer. Substitution of the resulting equ i valent pollutant concentratio n for the di spersi on term in Equation (A-2) yie l ds: (A-3) For a set of reactants and products, i and j, we now have a method of directly converting observed tracer concentrations to pol lutant concentrations where t he pollutants can be assumed to exist in local eq uilibrium. In the case of N0 2 , following Peters and Ri chards , the pertinent reacti ons and reaction rates are kl NO+ o3 + N02 + 02, (A-4) k2 N0 2 + hv + NO+ 0, and (A-5) k3 0 + o + M -+. 0 • 2 3 (A- 6) 156 Because these reactions proceed rapidly, the concentrations can be approximated by the steady state expression (A-7) The constant k2 is a function of the zenith angle of the sun, while the value of k1 = 20.8 ppm-l min. (Hecht et al., 1974). Peters and Richards suggest treating the following pairs of species: (A-8) (A-9) where obviously Q0 = 0. Peters and Richards note that typically 3 about 5% of the NOx is directly emitted as N0 2 and that an additional 5-10% is converted very near the source via the reaction: 2NO + o2 + 2N0 2. (A-10) Peters and Richards suggest that QNO ~ 0.12 QNO. 2 (A-11) X With this information, Equations (A-7), (A-8), and (A-9) can be solved for given values of Ct and Qt to yield the concentrations of N0 2, NO, and o3 corresponding to a specified set of background concentrations and emission rates. 157 Acknowledgments We are very happy to acknowledge Charles L. Bennett and Jack K. Suder, both Senior Air Pollution Research Specialists with the California Air Resources Board, for their interest, helpful suggestions, and their help in coordinating the field program. Personnel from Meteorology Research, Incorporated provided meteorological support and also collected airborne samples. In particular, we wish to thank Ted B. Smith, Hans D. Giroux, and William Knuth for their cooperation throughout the field study. We wish to thank C. Ray Dickson of the NOAA Laboratories in Idaho Falls, Idaho, for helpful suggestions regarding the development of chromatographic columns which were used to separate SF 6 and CBrF 3. Finally, we would like to acknowledge the efforts of the following members of the field study team: Bart Croes, Andrew Falls, Arthur Gooding, Jim Hickey, Eric Kaler, Wing Hong Lee, Ernest Sasaki, and Jim Westover. 158 Literature Cited Angell, J. K., D. H. Pack, G. C. Holzworth, and C. R. Dickson (1966) "Tetroon Trajectories in an Urban Atmosphere," J. Al2El• Meteor . §_, 565-572. Angell, J. K., C. R. Dickson, and W. H. Hoecker, Jr. (1976) 11 Tetroon Trajectories in the Los Angeles Basin Defining the Source of Air Reach i ng the San Bernardino-Riverside Area in Late Afternoon, 11 i• ~-Meteor.Ji, 197-204. Bierly, E. W., and E. W. Hewson (1962) "Some Restrictive Meteorological Conditions to be Considered i n the Design of Stacks, 11 ~- &2.2,1. Meteor . .l, 383-390. Briggs, G. A. (1969} Plume Rise. USAEC Critical Review Series TID25075, National Technical Information Service, Springield, VA 22151. Briggs, G. A. (1972) "Discussion on Chimney Plumes in Neutral and Stable Surroundings," Atmos. Environ . .§_, 507-510. Drivas, P. J. (1974) "Investigation of Atmospheric Dispersion Problems by Means of a Tr acer Technique," Ph.D. Thesis, California Institute of Technology, 16-26, Pasadena, California 91125. Drivas, P. J., and F. H. Shair (1974) i•A Tracer Study of Pollutant Transport and Dispersion in the Los Angeles Area," Atmos. Environ . .§_, 1155-1163 Drivas, P. J., and F. H. Shair (1977) "Atmospheric Dispersion and Transport Within Coastal Regions Part III. Tracer Study of Power Plant Emissions from the Monterey Bay and Long Beach Regions," Atmos. Environ . to be submitted for publication. Fosberg, M.A., D. G. Fox, E. A. Howard, and J. D. Cohen (1976) "Nonturbulent Di spersion Processes in Complex Terrain, 11 Atmos. Environ . .lQ_, 1053-1055. Fosberg, M.A., and M. J. Schroeder (1966} "Marine Air Penetration in Central California," i• &eEJ._. Meteor.§_, 573-589. Giroux, H. D., L. E. Hausner, L. H. Teuscher, and P. Testerman (1974) 11 Power Plant Plume Tracing in the Southern California Marine Layer, 11 from Proceedings of the Symposium on Atmospheric Diffusion and Air Pollution. Amer. Meteor. Soc., Boston, Mass. 02108, 238-245. Goodin, W. (1977) Personal communication. Environmental Quality Laboratory, California Institute of Technology, Pasadena, CA 91125. 159 Hay, J. S., and F. Pasquill (1959) "Diffusion from a Continuous Source in Relation to the Spectrum and Scale of Turbulence," Advances in Geophys . .§_, 345-364. Hecht, T. A., J. H. Seinfeld, and M. C. Dodge (1974) "Further Development of Generalized Kinetic Mechanism for Photochemical Smog, 11 Environ. Sci. Tech.§_, 327-339. Hino, M. (1968) "Maximum Ground-level Concentrations and Sampling Times," Atmos. Environ. l_, 149-165. Islitzer, N. F., and R. K. Dumbauld (1963) "Atmospheric Diffusiondeposition Studies Over Flat Terrain," Inter. J. Air Water Poll. 7, 999-1022. - -- Lamb, B. K. (1977) "Application of Atmospheric Tracer Techniques to the Study of Pollutant Transport and Dispersion," Ph.D. Thesis, California Institute of Technology, Pasadena, California 91125. Lamb, B. K., A. Lorenzen, and F. H. Shair (1977) "Atmospheric Dispersion and Transport Within Coastal Regions Part I. Tracer Study of Power Plant Emissions from the Oxnard Plain," Atmos. Environ. submitted for publication. Lamb, B. K., and F. H. Shair (1977) "Atmospheric Tracer Studies to Characterize the Transport and Dispersion of Pollutants in the California Delta Region," California Air Resources Board, Contract No. ARB-A5-065-87, 1709 11th Streetj Sacramento, California 95814. Liu, M., D. Durran, P. Mundkar, M. Yocke, and J. Ames (1976) "The Chemistry, Dispersion, and Transport of Air Pollutants Emitted from Fossil Fuel Power Plants in California: Data Analysis and Emission Impact Model," California State Air Resources Board, Contract No. ARB-4-258, 1709 11th Street, Sacramento, Calif. 95814. Miller, A. (1968) "Wind Profiles in West Coast Temperature Inversions," Report No. 4, San Jose State College, Dept. of Meteor., San Jose, California. Neiberger, M. (1955} "Tracer Tests of the Accuracy of Trajectories Computed from the Observed Winds in the Los Angeles Area,t' Air Pollution Foundation, Los Angeles, California. Pasquill, F. (1974) Atmospheric Diffusion. John Wiley and Sons, New York. Pasquill, F. (1976) "Atmospheric Dispersion Parameters in Gaussian Plume Modeling - Part II: Possible Requirements for Change in Turner Workbook Values," Environmental Protection Agency EPA-600/4-76-030b, Research Triangle Park, North Carolina 27711 . 160 Pasquill, F., and F. B. Smith (1971) "The Physical and Meteorological Basis for the Estimation of the Dispersion of Windborne Material," in Proceedings from the 2nd International Clean Air Congress, pp. 1067-72. Peters, L. K., and L. W. Richards (1977) "Extension of Atmospheric Dispersion Models to Incorporate Fast Reversible Reaction," Atmos. Environ. 11, 101-108. Reynolds, S. D., P. M. Roth, and J. H. Seinfeld (1973) "Mathematical Modeling of Photochemical Air Pollution-I," Atmos. Environ. 7, 10331061. Roberts, P. T. (1975) "Gas-To-Particle Conversion: Sulfur Dioxide in a Photochemically Reactive System," Ph.D. Thesis, California Institute of Technology, Pasadena, Calif. 91125. Roffman, A., and R. Grimble (1974) 11 A Time-Dependent Air Quality Model with Terrain Correction," from the Proceedings of the Symposium 2n. Atmospheric Diffusion .a.nil. A.ir_ Po 11 uti on, Amer. Meteor. Soc., Boston, Mass. 02108, 311-317. Sandberg, J. S., W. J. Walker, and R. H. Thuiller (1970) "Fluorescent Tracer Studies of Pollutant Transport in the San Francisco Bay Area, 11 i• Air Poll. Contr. Assoc. 20, 593-598. Smalley, C. (1957) "A Study of Air Flow Pat erns in the San Francisco Bay Area," Information Bulletin 6-15-70, Technical Services Division, Bay Area Air Pollution Control District, San Francisco, Calif. Smith, T. B. (1976) Personal communication. Altadena, California. Meteorology Research, Inc., Start, G. E., C.R. Dickson, and N. R. Ricks (1974) "Effluent Dilutions Over Mountainous Terrain and Within Mountain Canyons," from the Proceedings of the Symposium on Atmospheric Diffusion and Air Pollution, Amer. Meteor. Soc., Boston, Mass. 02108, 22S=°2"'32. Turner, D. B. (1961) "Relationships Between 24-Hour Mean Air Quality Measurements and Meteorological Factors in Nashville, Tennessee," i• Air Pol 1. Contr. Assoc. JJ_, 483-489. Turner, D. B. (1970) "Workbook of Atmospheric Dispersion Estimates," Environmental Protection Agency, Research Triangle Park, North Carolina. 161 CHAPTER 5. (Submitted for publication in Atmospheric Environment) TRACER STUDIES FOR CHARACTERIZING THE TRANSPORT AND DISPERSION OF PLUMES EMITTED FROM LOW-LEVEL AND ELEVATED SOURCES by Brian K. Lamb, and Fredrick H. Shair Division of Chemistry and Chemical Engineering California Institute of Technology Pasadena, California 91125 September, 1977 162 Abstract Five large-scale tracer studies utilizing SF 6 and CBrF 3 were conducted in urban St. Louis during August, 1975. The purpose of these studies was to investigate the relative impact of emissions from lowlevel and elevated point sources in and near an urban area. Dispersion results were correlated in terms of source height and the bulk Richardson numbero It appears that dispersion of emissions from low- level sources is initially enhanced in contrast to the dispersion of emissions from elevated sources. Data from low-level tracer releases agreed favorably with data previously reported for dispersion in urban St. Louis. Data from elevated tracer releases showed good agreement with the commonly used Pasquill-Gifford dispersion curves. The tracer results have been previously used to validate an urban air quality grid model (Overton et alo, 1976). The validation results showed that generally poor agreement existed between observed and simulated tracer concentrations. A brief review of these results is presented. 163 1. Introduction Differences in emission rates and geographical distribution as well as variations in the vertical profiles of the wind and temperature cause the relative impact of low-level and elevated pollutant sources to be particularly important to understand and particularly difficult to predict. As a result, many studies have been reported concerning experimental and theoretical approaches to the description of dispersion from low and high-level sources. Ragland (1976) used derivatives of the Gaussian point source expression to obtain values of the wind speed and source height associated with worst-case surface concentrations. He categorized experi- mental dispersion data from Carpenter et. al (1971), Smith (1973), and Turner (1970) on the basis of experimental release heights in terms of tall (200 m), moderately tall (100 m), and short (50 m) stacks, respectively. Comparison of worst-case conditions with these sets of dispersion data showed that, for equal stack heights, much higher worst-case concentrations were predicted with dispersion data from Carpenter and Smith than with data from Turner. McElroy (1969) analyzed results of low-level tracer releases in urban St. Louis and found that the urban area greatly enhanced the in iti al dispersion of plumes relative to the dispersion of plumes over flat, open country. Islitzer and Slade (1968) listed twenty recent dispersion studies; only one-third of the studies were conducted from elevated sources. Furthermore, the differences in meteorological and topographical conditions found among the studies cited by Ragland, the study by 164 McElroy, and those listed by Islitzer and Slade vary widely. For this reason, it is difficult, and, in some cases, pointless to compare dispersion results with respect to source height . In accordance with the large differences in source strength corresponding to sources of different heights, atmospheric diffusion models treat the dispersion of emissions from low-level sources differently from the dispersion of emissions from elevated sources . Koch and Thayer (1971) developed a multiple source Gaussian plume model in which low-level urban sources were described in terms of an area source integral expression and elevated sources were modeled with the Gaussian point source expression. One set of dispersion data were used for both source types. In the construction of an urban grid model, Shir and Shieh (1975) used a Lagrangian plume trajectory submodel to track the path and width of a point source plume until the plume width was comparable to that of the grid. Liu et al. (1976) incorporated large so 2 point sources into a K theory grid model by injecting emissions into the grid cell located at the effective stack height directly above each point source. In view of the diverse experimental observations and conditions and the specialized theoretical methods associated with low and highlevel sources, it appears that additional low and high-level dispersion data, obtained under conditions of similar topography and meteorology, would be of particular interest. Single and double atmospheric tracer techniques can be -used to quantitatively determine the differences in transport and dispersion which are associated with sources of widely 165 different heights under similar topographical and meteorological conditions. The information gathered during tracer releases provides insight into the relative impact of low and high-level sources. By using two tracers released simultaneously from sources of different heights in a particular area, differences caused by variations in the surface roughness and weather patterns are limited to only those existing during the tracer release. The purpose of this work was to utilize gaseous tracers, SF 6 and CBrF 3, to characterize the transport and dispersion of low-level and elevated plumes in the urban area of St. Louis, Missouri. A second purpose of the tracer study was to provide a well-defined data base for use in the validation of atmospheric diffusion models. The Regional Air Pollution Study (RAPS) conducted by the U.S. Environmental Protection Agency in St. Louis includes an extensive meteorological and air quality measurement system designed to provide detailed input data for large-scale atmospheric diffusion models. Urban St. Louis is surrounded by five large power plants located at distances as far as 49 km from the urban center. For these reasons, a series of large-scale tracer tests were conducted during August, 1975, in the metropolitan area of St. Louis. The test region is shown in Figure 1. Unfortunately, a utility worker strike at the power plants during the study period prevented the release of tracers from any of the power plants. For this reason, high-level sources were simulated by releasing tracer from the top of the 305 m KETC television tower. Although this simulation does not include the effects of initial / RAPS SITES lO, KILOMETERS O MISSOURI State Boundary Interstate Routes U.S. Highways State Highways 1 2 5 2 3 4 5 10 CREVE °COEUR vo. Figure 1. Location of tracer release sites (~ and hourly average air sampling stations (•) during St. Louis tracer tests. LABADIE POWER PLANT ~ @J @ -◊ 5 ◊ I EC POWER PLANT UNIVERSITY / CITYofl2 ¼ oFERGUSON oFLORISSANT ■ 115 ■ 117 RIVER REFINERIES 6 WOOD POWER PLANT EDWARDSVILLE . ,23 +I N 167 dilution of stack gases, it does eliminate uncertainty in the value of the effective stack height. 168 2. Experimental Procedure Five full-scale tracer tests were conducted from August 8, 1975 through August 15, 1975. The locations, times, and rates of the releases are listed in Table 1. Complete details of the experimental procedure and tabulations of all data are given by Lamb et al. (1975) . A total of 21 sequential air samplers were located at 21 different Regional Air Monitoring Stations (RAMS) as indicated in Figure 1. These air samplers permitted the collection of hourly averaged air samples for 12 consecutive hours. Ten-second averaged tracer data were collected during automobile traverses by having the passenger in each vehicle take grab samples in 30 cm 3 plastic syringes. Samples were collected every 0.1, 0.2, 0.3, 0.4, or 0.5 miles, depending upon the estimated width of the tracer plume. Airborne grab samples were collected similarly at various heights downwind of the release point. Airborne samples were also collected by means of a motor-driven syringe system; this system provided integrated air samples over distances equal to 16 km. Air samples were analyzed for SF 6 and CBrF 3 using electron capture gas chromatography. The analysis system, consisting of 12 gas chromatographs and 2 digital integrators , was set up in a motel room in St. Louis. The gas chromatographs were calibrated with an exponential dilution method. The detection limit of SF 6 was 1 part per trillion (ppt) ; the detection limit of CBrF 3 was 1000 ppt. The uncertainty of the test data was approximately~ 8% overall. In five tests, 2162 automobile traverse samples, 63 airborne grab samples , and 929 hourly averaged samples were collected and analyzed . 305 m, KETC-TV Tower 6. l m, 2040 to parking lot of 0300 Webster College 30. 5 m, RAMS Site No. 111 ( 3) (4) ( 5) 8/11 /75 8/12-13/75 8/15/75 0835 to 1425 l 040 to 1500 1030 to 1500 305 m, KETC-TV Tower (2) 8/9/75 7. l 6.2 6. l 9.2 6. l m, RAMS Site No. 111 (Instantaneous point sources released every hour on the half hour) 30.5 m, RAMS Site No . 111 30.5 m, RAMS Site No. 111 0930 to 1330 l 000 to 1200 l 040 to 1500 1100 to 1500 9.1 m, RAMS Site No. 111 9.6 Release Period (CDT) 1100 to 1500 (1) 8/8/75 Height and Location of SF6 Release 30.5 m, RAMS Site No. 111 Test Date TRACER RELEASE INFORMATION Height and Release Release Location of Period Rate CBrF 3 Release (CDT) GRAMS SECOND TABLE l 7.9 8. l 7.0 Release Rate GRAMS SECOND _, 0) I.O 170 Meteorological data, including wind direction, wind speed, temperature gradient , vertical wind profiles , and estimated mixing he i ghts were obtained through the RAMS network of surface and upper air measurement sites. The average meteorological conditions observed during each test are listed in Table 2. 8/8/75 8/9/75 8/11/75 8/12-13/75 8/15/75 1 2o2 * 3.5 4.5 * 2.3 140 1100-1500 1000-1500 1100-1500 2100-0300 0900-1400 210 200 310 (m/sec) (deg) CDT 400 >1000 1800 900 1530 (m) -0.038 -0.137 -0.173 0.088 0.052 A-B C-D B D-E D l Vector-averaged from RAMS surface data. 2 Estimated by personnel from Pedco, Inc. (1975). 3 Ri' = g[~T/~z + f]z2/Tu; data taken from RAMS 141 upper air radiosonde during each tracer release. 4 According to Turner's method (1961). * Because a storm front passed through Sto Louis during the tracer release, wind conditions were highly variable; no averages were calculatedo 2 3 4 5 DATE TEST ., AVERAGE METEOROLOGICAL DATA OBSERVED DURING TRACER RELEASE PERIODS ~/IND l WIND 1 ESTIMATED 4 TIME MIXING 2 Ri 13 DIRECTION SPEED HEIGHT PASQUILL-STABILITY CLASS TABLE 2 _. -....i _. 172 3. Discussion of Results A mass balance of the automobile traverse tracer data accounted for essentially all of the released tracer. In this analysis, it was assumed that tracer was vertically well-mixed from the surface to the estimated mixing height. Typical results of the tests are shown in Figures 2-4. The SF 6 tracer was easily tracked in every test as far as 26 km downwind. The CBrF 3 tracer due to its relatively high limit of detection was observed in only 3 automobile traverses. As · shown in Figure 5, these traverses were taken within approximately 3 km of the release site. The smooth curve in Figure 2 represents a Gaussian best-fit to the data. With the exception of traverses taken close to the release points (within 6 km of the source), agreement between best-fit Gaussian curves and the crosswind profiles was very good. The best-fit procedure, which corrected for the angle of the traverse path with respect to the wind direction, provided a means of calculating values of the horizontal standard deviation of the plume, cry. Airborne data were not sufficiently detailed to allow similar calculation of values of the vertical standard deviation, az ~ However, assuming that dispersion is given by the commonly used Gaussian plume model, values of oz can be determined in an iterative manner from the crosswind integrated tracer concentration: CCWI =1-oo C(y)dy = - 00 2Q IZTio exp[-½ z u (t)z 2]. ( 1) 173 TEST ST! TRAVERSE 10 DATE• 8/08/75 TIME• 1156-1202 TRAVERSE ROUTE• EAST ALONG HWY YY START!!'{; POINT• HWY YY AND ST. LCIUIS CITY LIHIT a .... co 1(L (L '-' CJ u • co z 0 u 3 6 DISTANCE ALONG TRAVERSE (KM.) TEST STl TRAVERSE 9 DATE• 8/08/75 TIME• 1358-1Y08 TRAVERSE ROUTE• EAST ALONG HWY YO START!!'{; POINT• INTERSECTION OF HHY YO Al'O LINDBERGH BLVD. a ~ r-.. ua zoo 0 u CJ...,......,........,....,...,._..........,._.,,,.ll!!!!!!!s.,__ _ _ ___JL-_..,..:::t1.....,...,.,_._..,_J 0 CJ ~ 5 10 DISTANCE ALONG TRAVERSE (KM.) 15 TEST STl TRAVERSE 5 DATE• 8/08/75 TIHE• 1259-1313 TRAVERSE ROUTE• SCIUTHEAST ALCING I-70 STRATI!'{; POINT• INTERSECT!~ OF I-70 AND I-270 r-.. IQ_ CJ (L (X) '-' ua z =:!' 0 u 7 lij DISTANCE ALONG TRAVERSE (KM.) Figure 2. Automobile traverse SF 6 data and best-fit Gaussian curves, Test 1, 8/8/75. 174 ,,~~ ;:·~~~ 0114 \ \ '' '' Figure 3. Overview of automobile traverse SF 6 data collected between 0200 and 0300 CDT, Test 4, 8/13/75. [SF6] = 1961 ppt. SF 6 released from Webster College between ~8~o and 0300 CDT. 175 3 - llfl OCl .11' -['-""] ~~-"" l•. ,(I.Ii, l e ,i0,1'\ r . --. ....,,,., oa .11~ -~ s1 . ,o.n-1 e,u1,~ Figure 4. Overview of hourly average SF 6 data collected between 0800 and 2000 CDT , Test 3, 8/11/75. SF6 scale= 0 to 100 ppt . SF 6 released from KETC tower between 1040 and 1500 CDT . °' , KM I 0 0.50 I I Q:58 - I 1:32 (CDT) [CBr F3] MAX.= 17ppb 8-11-75 AUTO TRAVERSE TEST 3 ST. LOUIS Figure 5. ..... ....... 177 The dispersion terms calculated from the automobile traverse data can be correlated with distance from the source in terms of source height, or, following McElroy's procedure, in terms of the bulk Richardson number: Ri' = g[6T/6z + rJ z2/Tu 2 (2) where r is the adiabatic lapse rate, 6T and 6z are the temperature difference and height difference, respectively, measured between an upper height z and a lower height z0 , 6T is the mean absolute temperature over 6z, and u is measured at z. McElroy correlated values of cry and crz in terms of Ri' and cr 0, the standard deviation of the wind direction. Unfortunately, cr 0 data were not available for the tracer release sites. For this reason, the tracer data were categorized only in terms of Ri 1 • Values of Ri' were determined from radiosonde data obtained from RAMS 141 near downtown Sto Louis. The dispersion results are shown in Figures 6-8 in comparison with curves from Pasquill-Gifford, and McElroy. The power law coefficients of best-fit lines through the tracer data are listed in Table 3. The relationship of source height to dispersion Cdn be examined in Figures 6 and 7o For tracer releases from 305 m, horizontal dispersion appears to be adequately described by the Pasquill-Gifford curve corresponding to Class A stability. For releases from lower heights, the observed horizontal dispersion agrees more favorably with the curves from McElroy. In Figure 8 where experi- mental dispersion curves are given in terms of the bulk Richardson number and compared to similar curves from the McElroy data, the slopes of the curves for the daytime unstable case (Ri' < -0.01) and for the 178 (f) a:: w w 1- :E • >- 103 1 (I :E Q ..... en 102 10 1 ~~~~~~~~~~~~~~~~~~ 10 2 103 X, METERS 104 ST. LOUIS TRAVERSE ORTA (/) a:: w 1- w :E • J03 N I (I :E Q • A ....., (/) 102 10 1 .____._____,.___,__.___._._...,_,._____,___._-'-.L.L-LU.J-----L----L----L-.L.....L..JU..U 10 2 10 3 104 X, METERS ST. LOUIS TRAVERSE ORTA Figure 6. Crosswind horizontal and vertical dispersion parameters versus downwind distance in terms of source height (H 1 = 6.1 m, H2 = 30o5 m, H3 = 305 m) compared to Pasquill-Gifford dispersion curves . 179 10" Cf) a: w r w :r . lQ3 >I a: :r ~ ,-, Cf) 102 lQ3 i 1011 X, METERS 105 ST. LIJJIS TRAVERSE DATA 10" Cf) a: w r w :r N . lQ3 I a: :r ~ ,-, Cf) 102 10 1 ._____.____.__._.L...J...JU..U--L-L-L..JJ...J..lll--_J__J__L__L..LI..JUJ 102 10s 103 1011 X, METERS ST. LCIUIS TRAVERSE DATA Figure 7. Crosswind horizontal and vertical dispersion parameters versus downwind distance in terms of source height (Hi = 6.1 m, H2 = 30.5 m, H3 = 305 m) compared to McElroy (1969) dispersion curves. 1011 (f) a:: w w r- . :::E >-I a: 103 :::E c..!) ...... (f) 102 R2 103 X, METERS 1011 1()5 ST. Ll1JIS TRAVERSE DATA 1011 (f) a:: w w r:::E N . 10 3 I CI :::E • c..!) ...... (/) 102 103 Figure 8. X, METERS 1011 ST. LOUIS TRAVERSE DATA 10s Crosswind horizontal and vertical dispersion parameters versus1 downwind distance in terms of bulk Richardson number (R1 = Ri > OoOl (daytime); R2 = Ri' < -OoOl; R3 = Ri' > 0.01 (nighttime)) compared to McElroy classes: 1) Ri' < -0.01, cr 0 = 30°+; 2) Ri' < -OoOl, cr 0 = 24-29°; 3) Ri 1 < 1 -OoOl, cr = 18-22°; 4) Ri' = ! OoOl, cr 0 = 15-20°; 5) Ri > 0.01, cr 00 = 8-13°0 181 nighttime stable case {Ri' > 0.01) agree closely with the corresponding McElroy curves. The dispersion data for Ri' > 0.01, daytime (Test 5) appear completely unrelated to the stable McElroy curve. It appears that conditions were much more unstable than suggested by the value of Ri'o Close examination of the calculations of Ri' during Test 5 indicated that small differences in the selected measurement heights could cause large differences to occur in the value of Ri'. Note that the curve for Ri' < -0.01 is based upon data from both low and highlevel sources. Yet, the curve closely parallels the curves from McElroy which are based solely upon surface releases. The scatter observed in correlations with the calculated values of a2 is generally much greater than for the corresponding correlations with measured values of ay. Because a storm front passed through St. Louis during Test 2, insufficient traverse tracer data were obtained to use in any of the correlations. In spite of the scatter in the data, a number of observations can be made. In comparison to Pasquill- Gifford vertical dispersion, the initial extent of dispersion estimated from the tracer data in each case is greater than that inferred from the stability classes given in Table 2. In comparison to McElroy's modified Pasquill-Gifford curves, however, the tracer data show better agreemento The grouping of points by source height in Figure 6 and Figure 7 indicate that the data collected during the highlevel releases are scattered about the Pasquill-Gifford curves associated with the stability conditions given for the tests in Table 3. Similarly, data from the low-level, daytime tracer releases are grouped 182 TABLE 3 BEST-FIT POWER LAW COEFFICIENTS FOR ST. LOUIS TRACER DISPERSION DATA az = ex d a = axb y x in meters Class ifi cation r2 a 1-b m b Test 1 Test 3 Test 4 Test 5 37 Oo74 0. 57 24 0.37 0.83 0.73 0.45 0.76 Oo93 0.78 0.86 H = 6.1 m H = 30.5 m H = 305 m 0.57 2.6 0.10 0.73 0.67 1.02 1.5 24 0.57 0.73 0.45 0.73 Ri ' < -0 .01 Ri' > o. 01 (day ) Ri 1 > 0. 01 (night ) d r2 1-d m r 2 = best-fit correlati on coefficient * data uncorrelated C * * * 208 l.8 2.3 Oo 17 0. 55 0.71 0. 42 0. 78 0.21 0.78 0.87 0.56 1.8 48 0.55 0.39 0.78 0.46 * * * 0.84 0. 86 0.78 224 2.3 1.8 0. 19 0.71 0.55 0.20 o. 21 0.78 183 about the McElroy-modified Pasquill-Gifford curves as expected from consideration of wind speed and insolation data. In terms of the bulk Richardson number, the tracer data show similar scattered groupings about the correspondtng curve from the McElroy data. One of the purposes of the tracer field study was to obtain data for the validation of urban atmospheric diffusion models. One such validation has been reported by Overton et al. (1976). It is worth- while to briefly review the results of the model validation. The model tested was developed by Shir and Shieh (1974, 1975) and is based on numerical integration of the atmospheric diffusion equation over a specified set of grid elements. The model is designed to accommodate point sources through the application of a Lagrangian trajectory model. Because the model was developed as a part of the RAPS study in St. Louis, it was of particular interest to evaluate its performance with respect to the tracer data. The model was tested by simulating the tracer releases during Test 1 (8/8/75, 30.5 m source height) and Test 3 (8/11/75, 305 m source height). During Test 1, all simulated tracer concentrations were greater than the experimental traverse and hourly average values. However, the position and widths of the simulated plume closely agreed with the experimental tracer profiles. Simulated values for Test 3 were scattered above and below the traverse and hourly average tracer concentrations. Figure 9 from Overton et al., shows the simulated and experimental concentrations for 3 crosswind traverses downwind of the elevated source. Calculated and experimental hourly averaged tracer concentrations are compared in Table 4. Correlation of all observed 184 TEST ST3 AUTO TRAVERSE 3 DATE• 08-11-75 TIME• 1215-1239 TRAVERSE ROUTE• EAST AU'.lNG HWY l!O STARTING POINT• 1270-HWY 40 JUNCTI I.I) r-- .,-,.. I-Cl... a Cl... I.I) ..._,, LJLI) zN D u 0 ..................................................»--4~..............- , 6 - - - - - - - J - - - _ _ _ ; ~ . . - - , . . . . . . . . J 0 l/) r-- 9 18 DISTANCE ERST ~F 1270-HWY 40 JUNC. CKM.) 27 TEST ST3 AUTO TRAVERSE 4 DATE• 08-11-75 TIME• 1256-1320 TRAVERSE ROUTE• WEST ON 155-70 ANO HWY 40 STARTING POINT• 155-70 AND BLACK LANE INTERSECTION .,-,.. l- a... 0 Cl... I.I) ..._,, u l/) zN D u o....,........,....,........,....,......,.....,......______----'--.-...:::.-........................ 0 9 18 27 DISTANCE EAST CF 1270-HWY 40 JUNC. (KM.) TEST ST3 RUT~ TRAVERSE 5 DATE• 06-11-75 TIME• 1342-1408 TRAVERSE ROUTE• EAST ALONG HWY 40 STARTING POINT• HWY 40-HWY 30 JUNCTION .,-,.. l- a... 0 Cl... I.I) ..._,, u l/) zN D u Figure 9. Comparison of observed(~) and calculated (o) SF 6 data along crosswind automobile traverses 18 km downwind of the tracer release point. 0 0 0 - 0/2 0 0 0 0 0 0 0 0 - 2/0 36/65 - - 0 0 46/189 2/0 0/32 - 0 6/15 0 0 0 5/0 0 0 0 4/0 12/279 5/0 0 6/9 0 0 0 0 0 0 0 - 0 0 0 0/1 0 0 0 0/1 0/8 - 0 0 0 0 5/0 4/79 1/0 0 63/172 1/0 0 0 0 0 4/0 0 2/0 6/0 0 0 0 13/206 0 11-12 2-3 12-1 1-2 11-12 0 22/18 0 0 0 0 0 0 2/7 0/17 0/4 0/0 0/1 27/26 12-1 0 18/22 0 0 9/6 7/12 0/3 0 0/1 26/17 0 0 0 6/11 1-2 0 44/18 0 0 7/21 13/29 0 0 8/4 0 0 0 7/1 0 2-3 Single entry: Both calculated and experimental concentrations are equal. Double entry: Experimental/calculated. Stations not included either had faulty data, no sample taken, or a value of zero . *calculated values from Overton et al. (1976); measured values from Lamb et al. (1975). l 01 102 103 104 105 106 109 110 112 113 114 115 116 117 118 119 120 121 Station Number TABLE 4 COMPARISON OF CALCULATED AND MEASURED CONCENTRATIONS OF SF 6 (PPT) * Test #3 (8/11/75) Test #1 (8/8/75) Time of Day - CDT Time of Day - CDT ..... co O'I 186 and calculated concentrations gave a correlation coefficient equal to -0.2. Because of the generally poor agreement between the calculated and experimental results, further testing of the model was not deemed worthwhile. 187 4. Summary and Conclusions The low-level tracer dispersion data agree favorably with the results from McElroy's low-level urban releases both in terms of modified Pasquill-Gifford classes and in terms of the bulk Richardson number. Since wind fluctuation data were not available, correlation of the dispersion data with the standard deviation of the wind direction was not possible. The elevated dispersion data, as might be expected, show better agreement with the curves of Pasquill-Gifford for flat, open country. At elevated levels, the initial enhancement of plume dispersion observed in dispersion of low-level emissions does not occur. These data strengthen the somewhat intuitive methods suggested for differentiating between dispersion from high and low sources. For Gaussian plume modeling purposes, the Pasquill-Gifford curves should apply to dispersion of emissions from elevated sources, and the McElroy curves should be used to model dispersion of emissions from low-level, urban sources. Two additional observations concerning the tracer data are also of interest for the application of air quality models. The effective wind direction determined from the position of the tracer plume centerline was within~ 10° of the wind direction obtained by vector-averaging data measured near the plume trajectory during the hour in which the tracer profile was measured. This suggests that use of the RAMS surface wind data in diffusion models will yield reasonably accurate descriptions of pollutant trajectories. The results from preliminary validation of an atmospheric diffusion model confirm this fact. The second observation, as noted previously, is that crosswind profiles of plumes emitted at low-levels do not appear Gaussian in form 188 for some distance removed from the release point. This distance appears to be related to the relative scale of the dispersing plumes and typical urban obstacles. Evidently, during the plume size enhancement process described by McElroy, the plumes are not Gaussian in form. Overton et al . concluded that further investigations of the sensitivity of the model with respect to the vertical eddy diffusivity were necessary since the position and widths of the simulated and observed plumes were in reasonable agreement, and concentrations were in poor agreement. Preliminary sensitivity analyses with respect to source location, source height, and mixing height did not significantly improve the agreement between simulated and observed results. 189 Literature Cited Carpenter, S. B., T. L. Montgomery, J.M. Leavitt, W. C. Colbaugh, and F. W. Thomas (1971} "Principal Plume Dispersion Models;' J. Air PolL Contr. Assoc. fl, 491-495. - --Isl itzer, N. F., and D. H. Slade (1968) "Diffusion and Transport Experi ments" in Meteorology and Atomic Energy 1968, D. H. Slade (ed.), U.S . Atomic Energy Commission, Oak Ridge, Tenriessee (from NTIS TID-24190), pp. 117-188. Koch, R. C., and D. S. Thayer (1971) "Validation and Sensi tivity Analysis of the Gaussian Plume Multiple Source Urban Di ffusion Model , 11 U. s. Environmental Protection Agency Contract No. CPA 70-94; available from NTIS PB-206-951. Lamb, B. K., J. D. Bruchie, and F. H. Shair (1975) "Tracer Studies for Characterizing the Transport and Dispersion of Plumes Emitted at Ground and Elevated Levels," U.S. Environmenta l Protection Agency Grant No. R802160-03-2, Research Triangle Park, North Carolina . Liu, M. K., D. Durran, P. Mundkar, M. Yocke, and J. Ames (1976) "The Chemistry, Dispersion, and Transport of Air Pollutants Emitted from Fossil Fuel Power Plants in California: Data Analysis and Em i ssion Impact Model, 11 Ca li fornia Ai r Resources Board Contract No. ARB-4-258, Sacramento, Cal i fornia 95814. McElroy, J. L. (1969) "A Comparative Study of Urban and Rural Dispers ion, 11 l• ~- Meteor ~' 19-31. Overton, J. H., Jr., B. K. Lamb, F. H. Shair, and W. E. Wilson, Jr. (1976) "A Dual Tracer Study for Validation of Models with Respect to High and Low Altitude Sources," presented at the 7th Internati onal Technical Meeting on Air Pollution Modeling and Its Applicati on, Airlie, Virginia. Pedco, Inc. (1975) Personal communication, Cincinnati, Ohio. Ragland, K. W. (1976) "Worst-case Ambient Air Concentrations from Point Sources Us ing the Gaussian Plume Model, 11 Atmos. Environ. J.Q, 371 -374 . Shir, C. Co, and L. J. Shieh (1974) "A Generalized Urban Air Pollution Model and Its Application to the Study of S02 Distributions in the St. Louis Metropolitan Area, 11 l• .6EE.!._. Meteor. J.l, 185-204. Shir, C. C., and L. J. Shieh (1975) "Development of an Urban Air Quality Simulation Model with Compatible RAPS Data , Vol. I," U. S. Environmental Protection Agency EPA-600/4-75-005a . 190 Smith, M. E. (1973) "Recommended Guide for the Prediction of the Dispersion of Airborne Effluents," ASMEo Turner, D. B. (1961) "Relationships Between 24-Hour Mean Air Quality Measurements and Meteorological Factors in Nashville, Tennessee," i• Air Poll. Contr. Assoc. ]J_, 483-489. Turner, D. B. (1970) "Workbook of Atmospheric Dispersion Estimates, 11 U. s. Environmental Protection Agency AP-26, Research Triangle Park, North Carolina. 191 CHAPTER 6. (Submitted for publication in Atmospheric Environment) A LIMIT MODEL FOR DETERMINING THE RELATIVE IMPACT OF RURAL AND URBAN EMISSIONS UPON URBAN AIR QUALITY PART I. DEVELOPMENT AND SENSITIVITY ANALYSIS OF THE MODEL by Brian K. Lamb and Fredrick H. Shair Division of Chemistry and Chemical Engineering California Institute of Technology Pasadena, California 91125 July, 1977 192 Abstract The concept of a limit model was introduced as a means of comparing the relative impact of urban sources and surrounding rural sources upon urban air quality. The model was developed using assumptions which forced the ratio of urban concentrations caused by rural and urban emissions to be an upper-limit value. Pollutant dis- persion was described in the limit model by the Gaussian point source expression for rural emissions and a Gaussian area source expression for urban pollutants. The model was formulated so that results could be discussed in terms of the percentage of urban area where the rural/urban concentration ratio was greater than a specified value. A sensitivity analysis indicated that the percentage of urban area where rural sources governed urban air quality was strongly dependent upon the ratio of rural and urban emission rates, the specified concentration ratio, and the stability of the atmosphere. In Part II of this series, the particular limit model developed herein was used to determine the relative impact of rural and urban so 2 emissions in the metropolitan area of St. Louis, Missouri. 193 1. Introduction One of the main goals of atmospheric dispersion modeling is to predict urban air quality solely from a knowledge of emission inventories and meteorological conditions. Starting with the urban air quality study of Frenkiel (1956), a great deal of effort has been expended towards this goal. Comprehensive reviews and/or summaries have been presented by Wanta in Stern's book (1968), Turner (1970), Moses (1969), Pasquill (1970, 1974), Neiberger (1971), and Seinfeld (1969, 1975). Much recent work has been described in the 11 Proceedings of the Symposium on Multiple-Source Urban Diffusion Models 11 (1970), 11 and in the Symposium on Atmospheric Diffusion and Air Pollution (1974). 11 As can be observed, the use of large-scale computer calcu- lations is increasing. Certainly, environmental quality questions often arise which require solutions to be generated (for practical reasons) through extensive use of computers. However, some questions arise in such a form as to permit the use of a 1i mi t mode 1. 11 11 A 1i mit mode 1 is based 11 11 upon a knowledge of the pertinent chemistry and physics and, by definition, upon assumptions which force the solution to be either a known upper bound or a known lower bound to the true solution. the calculated upper-bound solution 11 11 is still 11 quite low, 11 When then a more exact calculation may not be needed; when the calculated 11 upperbound solution 11 is 11 too high, 11 then the results are inconclusive and a more exact calculation is substantially justified. 11 Limit models 11 are also appealing because they lend themselves rather easily to an analysis of how uncertainties in the values of various parameters (associated 194 with the description of pollutant sources and meteorology) influence the final answer. This aspect of a 11 limit model 11 aids greatly in the rapid development of an overall perspective of the problem. This paper forms the first part of a two-part discussion of the use of limit models as an air quality analysis technique. In this initial article, a particular limit model will be constructed for comparing the impact of rural power plants relative to the impact of local urban sources upon urban so 2 air quality" The assumptions which form the foundation for such a model will be outlined and model nomenclature will be established. Sources of error will be examined and the effects of different urban emission distributions will be presented. Data obtained for metropolitan St. Louis will be used in this discussion. In Part II of this series, this particular limit model will be used to demonstrate that a progression of more complex questions concerning short-term and long-term pollutant impact upon air quality can be easily answered. of St. Louis. Application will again be illustrated for the case Specifically, the model will be used to determine the fractions of time and area that emissions from rural power plants dominate so 2 concentrations in the urban area. The analysis will also seek to determine if rural power plants dominate urban air quality on a long-term (yearly) basis throughout metropolitan St. Louis. 195 2. General Discussion of the Limit Model Consider, for example, the problem of constructing an upper limit model for determining the maximum impact of so emissions from rural 2 power plants, relative to that from local urban sources, upon so 2 concentrations in the urban region. of assumptions. Such a model would involve two types Assumptions of the first type concern the rural sources of so 2 and are either accurate (within the knowledge currently available) or are one-sided in such a manner as to always overestimate the corresponding so 2 concentrations within the urban region resulting from the rural sources. Assumptions of the second type concern the urban sources of so 2 and are either accurate or are one-sided in such a manner as to always underestimate the corresponding so 2 concentrations within the urban region due to local urban sources. If the question being asked is "Can rural sources dominate long-term average so 2 levels throughout the urban area? 11 , then the results of the upper limit model calculations are conclusive if, and only if, calculated concentrations due to rural sources are found to be much less than those due to urban sources. If, in spite of the one-sided assumptions, the S0 2 levels due to rural sources are still found to be much lower than those due to urban sources, then a more exact calculation (based on the same data) will merely confirm this conclusion. posed: Similarly, if a question is "For what fractions of time and urban area do the rural power plants dominate urban air quality? 11 , then the results will be described in terms of the maximum fractions of time and urban area where urban air quality can be governed by rural sources. 196 3. Importance of Geographic Context The appropriate assumptions for any particular upper-limit calculation should be chosen to minimize the effort of answering a particular question. Simple modeling assumptions appropriate to a windswept plain free of major topographic influences might be quite different from assumptions chosen to model a coastal air basin. Thus, the first step is to place the question in a given geographic context. In our example, the urban region of interest was taken to be the 18.9 km by 18.9 km zone of St. Louis outlined in Figure 1. The exact choice of the urban region was somewhat arbitrary, but includes essentially all of the known sources of so 2 emitted within the greater St. Louis area {Williams et al., 1967). Emission rates for so 2 are listed in Table l for the urban area and five surrounding power plants (Littman, 1975). Important geometric parameters for the five rural sources are also given in Table 1. Excellent summaries of the St. Louis meteorology and topography have been prepared by Martin et al. (1967), who also noted, "The St. Louis area experiences the meteorological conditions typical of most Great Plains cities that are not influenced by major natural features such as mountains or large bodies of water-" In this geographic context, a modified Gaussian plume model can be used to characterize the upperlimit impact of rural power plants upon the urban region. An area source expression based on a Gaussian dispersion model can be derived to characterize the lower-limit impact due to so 2 emissions from local urban sources. In other geographic settings, upper limit models might well be constructed based upon puff models, distinct point source models, Figure 1. LABADIE-POWER PLANT ...... 2 2 5 3 4 5 10 NT ILLINOIS \ 0117 POWER PLANT BALDWIN EDWARDSVILLE N 0123 y Urban region of St. Louis and five su r rounding rural power plants. KILOMETERS O 1 ------- MISSOURI State Boundary Interstate Routes U.S. Highways State Highways ('zy @ @ RAPS SITES o \ ~~ I ...:.1.' ....... u:i 198 TABLE l EMISSION RATES OF so 2 ASSOCIATED WITH URBAN AND RURAL SOURCES Average Sulfur Dioxide Emission Data for St. Louis ( Littman , 1975) Q Qu Grams so 2 Tons ~0 2 day second A. Urban Sources 9,300 885 1. Point Sources - mostly industrial and utility power plants located within the urban region of interest 370 2. Non-point Sources 35 9,670 Total Urban Sources 920 B. Rural Sources * 1. 2. 3. 4. 5. Qr Qr Baldwin-Illinois Power Labadie-Union Electric Sioux-Union Electric Meramec-Union Electric Wood River-Illinois Power 1,026 838 475 280 228 10,800 8,800 5,000 3,000 2,400 Total Rural Sources 2,847 30,000 RURAL POWER PLANT PARAMETERS Power Plant xr, Distance from urban area (km) s, Bearing angle with respect to urban area (clockwise from north) (deg) Hr, Stack height (meters) Baldwin Labadie Sioux Meramec Wood River 47.2 49.2 22.4 22.4 18.4 147 259 344 202 8 75 210 180 60 60 *Low-level industrial sources primarily located near the Wood River Power Plant have not been considered. 199 or upon biased solutions of the conservation equations. 200 4. Derivation of Limit Model Expressions The meteorology and geography of the St. Louis region suggest that pollutant dispersion over the area may be described with the Gaussian plume model (Turner, 1970). Tracer studies in the St. Louis region have shown that plumes are Gaussian in form (Lamb et al., 1975; McElroy, 1969). Thus, the Gaussian plume model provides a simple means of determining downwind concentrations due to the rural sources. Turner's (1970) point source expression may be written: ( 1) where Xr is the ground-level concentration, Qr is the release rate, u is the average wind speed, y is the crosswind distance from the plume centerline, and Hr is the effective rural stack height. The terms cr Y r and crz are the horizontal and vertical crosswind standard deviations, r respectively. McElroy (1969) has shown that for identical transport distances, dispersion is greater through urban regions than through open country; he compared tracer results for St. Louis with previous dispersion experiments for open country. McElroy concluded that for low- level point sources, the roughness of the urban area enhances the initial plume size and increases vertical and horizontal dispersion. At distances where the plume is larger than the size of eddies created by local obstructions, dispersion over urban areas approaches dispersion over open country. In view of this analysis, we evaluated cry and crz r r from dispersion curves given by Turner (1970) for distances less than or equal to xr, the distance from the rural source to the edge of the urban 201 zone. Inside the urban area, these parameters were calculated for x+x', where xis the downwind distance from the edge of the urban zone and x' is a virtual distance corresponding to the Turner sigma at the distance xr. These dispersion terms were determined at x+x' using McElroy's (1969) dispersion curves for urban St. Louis. Within the urban area, concentrations due to urban sources were calculated by modifying the Gaussian model of an infinite line source. If one considers a thin strip source at distance 11 a 11 from the edge of the zone, the ground-level concentration at x, also measured from the edge of the area, will be similar to Turner's (1970) expression for an infinite line source: (2) where Qu is the total emission rate, la is the width of the strip, o1 is the crosswind length of the box, D2 is the downwind length of the box, His the average urban effective stack height, u is the average wind speed, and crz(x-a) is the urban vertical dispersion parameter evaluated at (x-a), the source-to-receptor distance. The concentration at x due to an infinite series of thin strip sources will be: (3) Introducing the nondimensional (normalized) quantitie~ n =~and 2 Z02 He(n) = cr ~ n)' as shown in Appendix A, yields an expression for the 2 02 urban area source : (4) where xutx) corresponds to the ground-level concentrat ion at x. and and cr 2 is eval uated at x = n•D 2 + x" using best-fit l i nes to McElroy's data; x" i s a vi rtual distance corresponding to an initial plume size of 20 meters as suggested by McElroy (1969). The relative impact of rural sources to urban sources upon urban air quality depends upon the ratio of rural source to urban source concentrations: ( 5) In thi s form, it is apparent that the ratio is a product of the emi ssion rates rati o and a series of terms which are characteristi c of the atmospheri c dispersion process for the specified sources, terra i n, and meteorol ogy. The impact of a rural source upon urban air quality may be measured in terms of the fraction of the urban area where the rural/ urban concentration ratio is greater than a specified value. For example, the fraction of the urban area where 1/1: 1/10 corresponds to the portion of the urban area where a rural source has at l east 1/Jo t imes t he impact 203 that urban sources have upon air quality. This fractional area, A, can be derived from Equation (5) by setting ljJ = iµ0 and solving for y. The resulting function may be written: (6) 2 2·ln _I_ /2Tr [-½ :r ] 0 ) exp _H_ _ _ I __..,.,...._._z~r_,_______ , ) (02 ) r 2 2 °Jo(X lj,(n)[exp-½HJn) Jdn ljJ) (0 Y, where y0 (x,iµ 0 ) is the crosswind distance from the plume centerline at which the concentration ratio equals iµ 0 • Plotting y0 as a function of downwind distance yields an outline of the area where ljJ > ljJ. When the - 0 plume is completely within the urban box, the percentage of area where iµ ~ iµ 0 i s s imply (7) where x0 is the distance from the leading edge of the box along the centerline to where ljJ = iµ 0 (y 0 = 0). This distance may be determined by interpolation using a second degree polynomial fit to iµ(x,o). The integrals in Equation (7) and in all subsequent calculations were evaluated numerically using Simpson's rule in an iterative procedure. In the application of this model to St. Louis, we assume that (1) dispersion from rural sources can be adequately described by the Gaussian point source plume model; (2) dispersion from urban sources can be adequately described by an area source model in the form of a Gaussian expression for 204 an infinite series of line sources; (3) the effective stack height for rural sources is equal to the lowest actual stack height at each source (Hr); (4) the average effective stack height for urban sources equals 30 meters; (5) the wind speed, u, and the mixing layer height, L, are constant everywhere and are given as functions of atmospheric stability: class A'B 1 u = 3 m/sec L = 3000 meters class C' u = 4 m/sec L = 2000 meters class D'd u = 5 m/sec L = 1600 meters (day) class D'n u = 5 m/sec L = class EI F' u = 3 m/sec L = 300 meters (night) 800 meters {night) where the stability classes are taken as Pasquill-Gifford conditions as modified by McElroy (1969); (6) dispersion parameters, oy and oz, in the rural source r r expression are determined for points in the urban area using McElroy's modified Pasquill curves at x+x' where xis the distance from the edge of the zone and x' is a virtual distance corresponding to a standard deviation determined from Pasquill curves at xr, the distance from the rural source to the edge of the zone; (7) the dispersion parameter oz for urban sources is evaluated for points in the urban area using McElroy's curves at x+x" where x" is a virtual distance corresponding to an initial standard deviation of 20 meters; 205 (8) the effect of an inversion layer acting upon vertical dispersion is accounted for by using Turner 1 s assumption that cr 2 = 0.8L for distances greater than 2xL where xL is the distance at which cr 2 = 0.47L; (9) rural and urban emission rates are continuous and constant over time, and urban emission rates are constant throughout the urban zone; (10) (11) so 2 is conserved during transport; the urban so 2 source distribution is completely symmetrical and the urban box is always in line with the wind direction (the leading edge of the box is always perpendicular to the wind direction). Assumptions (1), (6), (7), and (8) are taken as providing as accurate a basis as is available for the limit model approach. As- sumption (2) makes it unnecessary to do detailed calculations using actual urban source inventories. As will be discussed, an urban source distribution, Q(x), determined from emission inventories, can be used in the model to provide a more realistic description of urban sources. The effective stack height is taken to be equal to the actual stack height; this assumption overestimates the impact of the rural sources and simplifies the calculations. An average effective stack height of 30 meters in urban St. Louis has been suggested by Williams et al. (1967). The wind speeds in assumption (5) are taken from the definition of Pasquill stability conditions (Turner, 1970); use of the Gaussian model requires a mean wind speed. Unfortunately, few data are available which relate inversion layer heights to stability conditions. The values 206 listed in assumption (5) appear to be reasonable for the St. Louis region (McElroy, 1976). Assumption (9) is required because available data do not specify emission rates as functions of time. If such information were available, then a function Q(x,t) could be formulated for use in the model. One might expect to see diurnal and seasonal variations in emissions; variations on such time scales are easily adaptable to the model. Roberts (1975) has reported so 2 conversion rates for St. Louis to be approximately 3% per hour. conversion rate. Assumption (10) neglects this More importantly, assumption (10) neglects the depo- sition of so 2. Vaughan et al. (1975) reported that for emissions of S0 2 from a St. Louis power plant, as much as 40% of the so 2 was lost over transport distances of 43 km. Deposition of so 2 emitted from a low-level industrial source was found to be as much as 30% over 23 km. Vaughan et al. estimated the deposition velocity of so 2 to range from 3 to 6 cm/sec. Assumption (10) thus overestimates the impact of rural so 2 plumes. At any point in the urban region, rural plumes will have had at least as much or more contact with the surface than urban emissions. As will be shown in Part II, rural plumes are dispersed to the ground before reaching the urban region. Urban emission data by Williams et al. (1967) indicate that assumption (11) is not unreasonable. Williams' data, presented in a grid system for the region, show high emission rates in the center of the area with decreasing rates towards the edges of the grid . The symmetry of emission distribution allows us to rotate the urban box with the wind and simplify the calculations considerably. 207 5. Sensitivity Analysis In any given problem there will generally be a lack of data as well as uncertainties in the data which are available. Therefore, it is important to determine the sensitivity of the calculated results to uncertainties and/or errors in the input parameters. As indicated in Equations (6) and (7), the percentage impact area is dependent upon the value of w , the value of Qr/Qu, the rural transport distance, and the 0 effective stack heights of both rural and urban sources. The percentage impact area also depends upon atmospheric stability and inversion height through their influences upon the values of the dispersion parameters. Unfortunately, data which statistically relate the inversion height to stability class for a given region (although extremely valuable) are generally not available. Due to the presence of logarithmic, exponential, and integral terms within Equations (6) and (7), a sensitivity analysis of the percentage impact area requires numerical evaluation. The percentage impact area was calculated with respect to variations in stability class, w , Qr/Qu' inversion height, transport dis0 tances and effective stack heights. The Baldwin power plant, along with the stability class D'd (day), was chosen to serve as the baseline for the sensitivity analysis; for this case L = 1600 meters, Hr= 75 meters, and H = 30 meters. The results of these calculations are summarized in Figure 2 where the percentage impact area is shown as a function of each independent variable. It is clear from the first diagram in Figure 2 that atmo- spheric stability is a major factor in the relative effects of rural and urban sources upon urban air quality. Worst-case conditions occur under I A' B' I D'd I D'n I STABILITY CLASS C' /, I I,~ 0 I 10 10 0.4 -- t:. D 0 I 0.6 0.8 Or/Ou 20 30 Xr (km) 20 30 H(m) I 40 40 1.0 I I 50 0 2 'f'O 3 - 1200 800 L (m) 'Y 400 200 600 ♦ Hr (m) • 50 0 400 0 I I ---- 800 1600 4 I 1000 co 5 T I co N 0 Sensitivity of percentage impact area, A, to changes in stability class, ratio of rural and urban emission rates (Or/Gu), rural transport distance (xr), urban effective stack height (H), specified rural to urban concentration ratio (~0 ), inversion height (L), and rural effective stack height Hr. The baseline case was the Baldwin power plant under D'(d) stability conditions: Or/Ou= 1. 12, Xr = 47.2 km, H = 30 m, ~o = 1, L = 1600 m, and Hr= 75 m. as in assumption (5 l 0 E'F' 0.2 I I ■ INVERSION PRESENT, L=f (stability), e NO INVERSION, L=CO I 0 10 20 Figure 2. A(%) 30 40 50---------------------------------------------- 209 D1 n (nighttime) stability (with an inversion present) and cause the percentage impact area to more than double over that calculated for unstable conditions. It appears that the presence or absence of an inversion layer has a much larger effect upon the magnitude of A during unstable conditions than during stable conditions. This can be at- tributed to the greater extent of vertical mixing which occurs during unstable periods. The limited amount of vertical mixing which occurs under stable E1 F1 conditions eliminates the effect of an inversion at the assigned height of 300 m. The effects of an inversion at 1600 m during class D1 d periods are also negligible. At night under D1 n conditions, where the inversion height is assumed to be 800 m, it appears that vertical mixing is limited to some extent by the presence of the inversion layer. In the second and third diagrams in Figure 2, the abscissa scale for each variable is written above the variable. As indicated in the center diagram, the value of A increases by a factor of 6 as the ratio of Qr and Qu ranges from 0.2 to 1.2. This is the reverse of the effect of values of ~o upon A; the percentage impact area decreases by a factor of 6 as ~ovaries from 1 to 5. The equal, but opposite effects of Qr/Qu and ~o upon the value of A were expected because of the form of equations (6) and (7). Small changes in the percentage impact area occur as a result of changes in the inversion height as illustrated in the third diagram in Figure 2. Unexpectedly, the impact area due to a 1600 meter inversion is smaller than that which occurs during no inversion. Detailed inspection of the calculation procedure indicates that this anomaly results from Turner's approximation for inversion height 210 reflections. During the transition of cr2 from a varying function to a constant, when inversions exist, the calculated concentration actually decreases faster with distance than when no inversion is present. culations indicate that errors due to this deviation are small. CalThe percentage impact areas for the range of inversion heights from 400 meters to no inversion decreases from 48% to 32%. While this change is significant, it is less than the variations which occur due to changes in stability class, concentration ratio, or source strength ratio. If the transport distances are changed over the range given in Table l for the five rural power plants, surprisingly small changes in the impact area occur. As shown in Figure 2, the percentage impact area increases from 22% to 32% as the source distance increases by a factor of 4 from 11.8 km to the actual distance, 47.2 km. Even though the area expression includes a relatively simple relationship with cry, the percentage impact area is rather insensitive to changes in xr. r It is important to note that for a specified atmospheric stability condition, emission rate, stack height, and value of ~0 , there exists a transport distance, xmax' where the value of A is a maximum (see Figure 3). Under the standard set of conditions for the Baldwin power plant, the transport distance is such that the percentage impact area is quite close to its maximum value for ~o = 1. Once a value of ~o is chosen (from an assessment of desired air quality), a power plant should be located so - that the transport distance is larger than that which produces the maximum value of A for~_::: ~0 • For distances less than xmax' the percentage impact areas will be smaller than the maximum area, but concentrations within the area will be higher than concentrations which are 211 100--~---,------,------------------Percentage Impact Area vs Rural Transport Distance 90 80 S02 Emission Rate (kg/sec) 70 60 Xmax=99km • 2.7 kg /sec ■ 5.4 kg /sec • 10.8 kg/sec A(%)50 t>I • 21.6 kg /sec 40 Xmax=50km 30 10 0 l.,j__Jt:::::::::.--1..::::..---1~~-_j__ 0 Xmax= 17km 200 300 400 500 _.I:::::::t:::::1._J 600 700 800 Xr (km) Fig ur e 3. Determination of the r ural transpo rt distance, xma x , where t he Dercentage impact area, /\, is a maximum. 212 found in an equal-sized area resulting from transport distances larger than '\nax· The impact area expression is relatively insensitive to changes in rural and urban effective stack heights. The areas in Figure 2 calcu- lated for various values of Hr indicate that assuming the actual stack height equal to the effective stack height serves as a close upper bound for all emission heights. As the urban effective stack height increases, the impact area increases slightly, Figure 2. Selecting an average source height for all sources will cause only small errors in the percentage impact area. In view of the limited resources which are generally available to improve air quality, the preceding sensitivity analysis is useful because it focuses attention upon the input variables which have the greatest effects upon the accuracy of the final impact assessment. For example, the sensitivity analysis in this case suggests that time and effort should be directed towards obtaining accurate emission rates. The relative order of importance of all the input variables can be determined by using the curves in Figure 2 to assign a simple functional relationship between the percentage impact area and each input variable for the range of values under consideration. Standard error analysis (Bevington, 1969) can then be used to calculate the magnitude of the rate of change in A due to variations in each input variable. The results for such a procedure are given in Table 2; the assigned functional dependence for each variable and the values of the derivative of A with respect to the variable for a specified value or range of values are given for each input variable. The relative sensitivity of A ( 1Okm<x <50km) - r(Hr= 75 m) 0.27 % km -1 A= (-11.6%)in L + 117% { A= (0.27% km- 1 )xr + 19.5% 1 3 {- 0.024% mA= 26.2% e-Hr/lO m + 9.9% _ - 0.014%m 1 t/Jo L xr Hr 70 m 41 m 3.7 km 143 m 90/ 34 m 12% 55% { 7% (xr = 50 km) 37% (xr = 10 km) 9% /0 15% 3% { 3% (Q/Qu = 1.2) 15% (Qr/Qu = 0.2) (%) ~/m· 100 0. 77 0.03 0.03 m ~A = (aA) am - ~ ~m for M = 1 H A= (0.10%m- 1)H + 29 . 3% 0. 1%m-l (15m~H~45m) 67% (H = 15 m) 10 m { 22% (H = 45 m) * Note that the input variables are listed in decreasing order of sensitivity. (H r = 600 m) L = 1600 m L = 400 m - 0. 029%/m - 0.007%/m l/1 0 = 5 { 1. 3% l/1 0 = 1 32. 4% A= (-32.4%)/t/! 0 - 0.41 % (. 2~Q/Qu~l. 2) 28.7% Q/Qu m A= (28.7%)Qr/Qu - 0.08% (!~) m Functional Relationship: A= f(m) Input * Variable: m COMPARISON OF RELATIVE SENSITIVITY OF THE PERCENTAGE IMPACT AREA TO THE INPUT VARIABLES TABLE 2 N _, w to the input data can be determined by comparing the percentage uncertainties in the variables which produce changes in A of one percentage point. On this basist the data in Table 2 indicate that the relative sensitivity of percentage impact area to the input parameters decreases in the order: Qr/Qu ~ ~o > L > xr >Hr> H. This result suggests that in impact assessment effortst resources for gaining more accurate or additional information about the input parameters should be appropriated according to the relative order of sensitivity of A to the input parameters. Reduction of uncertainties in the values of Qr/Qu involves accurate inventories of pollutant emission rates. Specification of the concentration ratio (or in other cases just the concentration} requires an economic and physiological assessment of the concentration levels where adverse health and property effects become important. The uncertainty in A due to the effect of an inversion layer upon vertical dispersion could be reduced by correlating inversion heights with stability conditions using meteorological data that is currently available. The sensitivity of the percentage impact area to xr can be used to assess the impact associated with the location of new rural sources. Uncertainties in estimating effective stack height appear to produce relatively small changes in A; more accurate determinations of effective stack heights are of limited value in comparison to information needed concerning the other input parameters. 215 6. The Effect of Source Distribution upon the Impact Area In addition to det ermining the sensitivity of the impact area to var iations in the magnitude of i ndividual input parameters, it is important to determine the effect of variations in the form of the distribution of emissionso For example, the emission distribution might be described (8) as a constant, Q(x) as a sine function, =; 0Q~l 2 sin (;x); 2 ( 9) ( l O) or as a ramp function, In each of these forms, Q(x) is the emission rate per unit area, and appears in place of Qu/o 1o2 in Equation (6). The three distribution forms are compared to so 2 emiss i on inventory data by Williams et al. (1967) in Figure 4. The grid emission data have been averaged over rows and columns so that the step-curve i n the figure represents an average distribution as a function of downwind distance (see Appendix B)o We see that the s ine and ramp function s appear to approximate the data better than does the uniform distribution. The effects of these different forms on the relative impact of rural and urban sources are examined in Figure 5 where so 2 concentrations due to urban sources are shown as a function of downwind distance fo r three forms of Q(x). Near the edges of the zone, so 2 concen trations which result from the uniform emission dis t ributi on are high~r than the 216 6 • Io- 5 r - - - - . - - - - , . - - - - , - - - - - r - - - - , - - - - ST. LOUIS EMISSION DISTRIBUTION FUNCTIONS - O(x), measured* - · - Q(x), uniform •••••• Q(x), sine ---- Q(x), romp 4•10- 5 Q grams/ sec-m 2 .• :i, .• :,I 3· 10- 5 :I : I ·-·-+'· • I fl : I ;, /I : I : I ! ,___ __,J : I : I !1 :I :, :, :, l' 18.9 Figure 4. S02 emissions in St. Louis as a function of the distance from the edge of the urban zone. *Measured emission distribution was calculated from data by Williams et al. (1967) and Littman (1975) as given in Appendix B. Figure 5. 0 200 400 (l,JG/M3 ) ( SeJ..t 600 800 1000 0 8 12 16 20 Urban concentration of SO? as a function of downwind distance for the three forms of the urban emission distribution function. DISTANCE FR~M LEADING EDGE (KM) 4 + Q(X) RAMP CLASS• 0 • (!) Q( X) UNIFCJRM ~ Q( X) SINE ATM. STABILITY " N _, 218 levels which result from either the sine or ramp emission distributions. In the center of the zone, this situation is reversed. However, the areas where ~(x,y) .::_ l (the rural source impact is equal to or greater than urban source impact) are almost identical in magnitude for the different distribution formso Levels of so 2 caused by Baldwin emissions during D'd conditions were greater than levels caused by the constant, sine, and ramp urban emission distributions in 32%, 33%, and 33% of the urban area, respectively. While it is clear that the ramp and sine functions each provide a better fit to the emission data, the relative area impact of rural to urban sources can be described equally well by all three representations. For this reason and for simplification we have assumed the emission rate to be uniform throughout the region for further applications of the area source model. It appears that in cases where the relative impact of rural to urban sources is in question, it is more important to accurately determine the total amount of urban emissions than to determine their spatial distribution. 219 7. Summary and Conclusions Gaussian expressions for elevated rural sources and an urban area source were used to develop an upper-limit ratio of rural to urban so 2 concentrations which occur within the urban region. An expression for the area where specified concentration ratios may be found within the urban zone was derived. The dispersion parameters in the resulting expression were evaluated using Pasquill-Gifford dispersion data for rural transport and McElroy's dispersion data for urban transport. A sensitivity analysis indicated that the ratio of the rural to urban source strengths, and the specified ratio of the rural to urban concentrations are the most important terms in determining the size of the percentage impact area. Atmospheric stability also has a dominant influence upon the percentage impact area through its effect upon rates of dispersion and its assumed correlation with inversion height. Vari- ations in inversion height and rural transport distance have significant, although relatively smaller, effects upon the percentage impact area. Evaluation of the effect of rural transport distance upon A showed that a distance exists which produces a maximum percentage impact area for a specified set of conditions; ideally, rural sources should be located beyond the distance where this maximum percentage impact area occurs . The impact area expression was found to be least sensitive to uncertainties in rural and urban effective stack heights. Assuming that the actual rural source stack height is equal to the effective stack height provided an upper bound upon the percentage impact area. Because of the form of the rural to urban concentration ratio, uncertainties in the average wind speed have no effect upon the percentage impact area. 220 We believe that sensitivity analyses in both this case and in other modeling applications can provide a relatively simple means of determining where to direct available resources in order to most efficiently obtain the best data base for the problem under consideration. In the St. Louis limit model it appears more desirable to obtain an accurate emission inventory, to determine an acceptable concentration ratio, and to correlate inversion height data with stability class frequencies than to acquire very exact estimations of effective stack heights or to determine average wind speeds. The actual urban source spatial distribution was compared to three idealized distributions (sine, ramp, and uniform distributions) suitable for use with the model. Both the sine and ramp distributions appear to give more reasonable representations of the emission inventory data. However, the percentage impact area was found to be quite insensitive to the form of the spatial distribution of urban emissions. Consequently, the spatially uniform distribution of urban emissions was adopted for use in subsequent calculations. The model can now be used to assess the overall relative impact of rural and urban sources upon the urban region . As will be described in Part II, such an assessment involves an analysis of the time and area associated with specified values of the concentration ratio which may be found within the urban area. The results of this analysis will be presented in terms of the percentage of area and the percentage of time where rural sources contribute the greater part of urban so 2 concentrations. In addition, we will examine the overall area and time averaged so 2 concentration and rural/urban concentration ratio. 221 APPENDIX A Derivation of Urban Area Source Expression Turner's Gaussian expression for an infinite line source may be written for an infinite thin strip source as: (A-1) where x(x) is the ground-level concentration at x;(x-aj is the source-toreceptor distance, cr 2 is the vertical crosswind standard deviation, Qu is the total urban emission rate, His the average urban effective stack height, u is the mean wind speed, and D1 and D2 are the crosswind and downwind dimensions of the urban zone. For an infinite series of strips between o and x the expression becomes: X(x) (A-2) This equation may be rewritten by setting n = ~-a, dn = - 0da , and H( ) = e n H • 2 2 cr { nD ) • z 2 x(x) = (A-3) 222 Reversing the order of integration yields (A-4) which is the expression for the ground-level concentration in the urban area. 223 APPENDIX B The Distribution of so 2 Emissions in St. Louis Williams et al. (1967) reported the distribution of so 2 emissions in urban St. Louis during three winter months by using a rectangular grid system consisting of 6 rows and 7 columns. In order to obtain an average distribution for comparison with three idealized distribution forms, we assumed that the winter distribution was representative of the yearly emission distribution. Furthermore, since the total urban so 2 emission rate estimated by Williams et al. (1967) was less than that given by Littman (1975), the distribution was assumed to have uniformly increased to agree with the 1975 total emission rate. For simplicity, the rates given in the grids from the easternmost column were added to those in the adjacent column to yield a grid system measuring 6 x 6. Since the rates in the outside column were relatively small, this did not change the overall pattern of emissions. To smooth the distribution, the grid emission rates were summed by row and by column. The total of each row was averaged with the total of the corresponding column beginning with the northernmost row and westernmost column. The final result, shown in Figure 4, is a smoothed distribution of emissions as a function of downwind distance, independent of wind direction. 224 Acknowledgments We wish to thank William Wilson, James McElroy, and Francis Pooler of the U.S. Environmental Protection Agency for their interest. advice, and comments. We also wish to thank Lou Howard of the Massachusetts Institute of Technology for his stimulating discussion. Finally, we thank Fred Littman of Rockwell International for his help in obtaining emission data. under Grant No. R802160-03-2. This work was supported by the U.S.E.P.A. 225 Literature Cited Bevington, Po Ro (1969} Data Reduction and Error Analysis for the Physical Sciences. McGraw-Hill, New York, PPo 56-65. Frenkiel, F. N. (1956} "Atmospheric Pollution and Zoning in an Urban Area, 11 Sci. Monthly 82, 194-203. Lamb, B. K., J. D. Bruchie, and F. H. Shair (1975} "Tracer Studies for Characterizing the Transport and Dispersion of Plumes Emitted at Ground and at Elevated Levels," U. So Environmental Protection Agency EPA Grant No. R802160-03-2, Research Triangle Park, North Carolina. Littman, F. (1975} Personal conmunication. Thousand Oaks, California. Rockwell International, Martin, D. D., P. A. Humphrey, and J. L. Dicke (1967} "Interstate Air Pollution Study, Phase II Project Report, 5, 11 National Center for Air Pollution Control, Cincinnati, Ohio. McElroy, J. L. (1969} "A Comparative Study of Urban and Rural Dispersion," i• .B.E.EJ.. Meteor . .§_, 19-31. McElroy, J. L. (1976} Personal communication. Agency, Las Vegas, Nevada. Environmental Protection Moses, H. (1969} Mathematical Urban Air Pollution Models. National Center for Air Pollution Control, Chicago Dept. of Air Pollution, Argonne National Laboratory ANL/ES-RPY-001, Argonne, Illinois. Neiburger, M. (1971} "Meteorological Aspects of Air Pollution and Simulation Models of Diffusion, Transport, and Reactions of Air Pollution," Project Clean Air Report, Dept. of Meteorology, UCLA, Los Angeles, California. Pasquill, F. (1970} "Prediction of Diffusion over an Urban Area--Current Practice and Future Prospects," from Proceedings of the Symposium on Multiple-source Urban Diffusion Models, A. C. Stern, ed .. National Air Pollution Control Administration, APCO Publication No. AP-86, Environmental Protection Agency, Research Triangle Park, North Carolina, pp. 3.1-3.26. Pasquill, F. (1974} Atmospheric Diffusion, 2nd edition. John Wiley and Sons, London, England, pp. 329-400. Roberts, P. T. (1975} "Gas-to-particle .Conversion: Sulfur Dioxide in a Photochemically Reactive System," Ph.D. Thesis, California Institute of Technology, Pasadena, California 91125, p. 61. 226 Seinfeld, J. H. (1969) "Mathematical Models of Air Quality Control Regions," presented at Symposium on the Development of Air Quality Standards, sponsored by the eight Tuberculosis and Respiratory Disease Associations of Southern California and the Institute of Urban Ecology, University of Southern California, Los Angeles, California. Seinfeld! J. H. (1975) Air Pollution Physical and Chemical Fundamentals. McGraw-Hill, New York, pp. 32-48, 260-342. Stern, A. C., ed. (1968) Air Pollution, Vol. 1, 2nd ed .. Academic Press, New York, pp. 215-223. Stern, A. C., ed. (1970) Proceedings of Symposium .Q.!l Multiple-source Urban Diffusion Models. National Air Pollution Control Administration APCO Publication No. AP-86, Environmental Protection Agency, Research Triangle Park, North Carolina. Symposium .Q.!l Atmospheric Diffusion and Air Pollution of the American Meteorological Society (1974). American Meteorological Society, Boston, Massachusetts. Turner, D. B. (1970) "Workbook of Atmospheric Dispersion Estimates," Environmental Protection Agency No. AP-26, Research Triangle Park, North Carolina. Vaughan, W. M., R. B. Sperling, N. V. Gillani, and R. B. Husar (1975) "Horizontal S02 Mass Flow Rate Measurements in Plumes: A Comparison of Correlation-Spectrometer Data with a Dispersion and Removal Model," presented at the 68th Annual Meeting of the Air Pollution Control Association, Boston, Massachusetts. Williams, J. D., G. Ozolins, J. W. Sadler, and J. R. Farmer (1967) "Interstate Air Pollution Study, Phase II Project Report, ~, National Center for Air Pollution Control, Cincinnati, Ohio. 11 227 (Submitted for publication in Atmospheric Environment) A LIMIT MODEL FOR DETERMINING THE RELATIVE IMPACT OF RURAL AND URBAN EMISSIONS UPON URBAN AIR QUALITY PART II. THE RELATIVE IMPACT OF RURAL POWER PLANTS AND URBAN SOURCES UPON ST. LOUIS SO 2 CONCENTRATIONS by Brian K. Lamb, and Fredrick H. Shair Division of Chemistry and Chemical Engineering California Institute of Technology Pasadena, California 91125 July, 1977 228 Abstract The upper limit model developed in Part I of this series was used in Part II to determine the relative impact of rural and urban sources upon so 2 concentrations which occur within the urban region of St. Louis. Impact was measured in terms of the percentage of urban area where rural emissions governed urban so 2 concentrations by a specified ratio of the concentrations caused by rural and urban sources. The maximum percentage of the urban area where the specified concentration ratio was greater than or equal to 1 was 39%. Incorporation of wind direction and atmospheric stability frequency data into the analysis indicated that this percentage impact area occurs no more than 4% of the time. The combined rural sources have a greater impact than urban sources upon urban so 2 levels during 67% of the time in some portion of the urban region. Furthermore, rural emissions appear to govern urban so 2 levels by a factor of at least 5 during 52% of the time in some portion of the urban area. A total impact variable, defined as a function of fractional impact area, fractional impact time, and specified concentration ratio, provided a means of comparing the impact among five rural power plants upon urban so 2 levels. In terms of this impact function, the source with the highest emission rate had the second largest impact, whereas the second largest source had the second lowest impact. These results were governed by differences in power plant emission rates, rural transport distances, and wind direction frequencies. The annual average so 2 concentration in the urban area was com• puted and compared to available air quality data. Generally, the 229 predicted value serves as an upper bound to the measured data. Rural sources were found to contribute only 25% of the yearly averaged so 2 concentration within the designated urban St. Louis area. 230 1. Introduction In Part I of this series, the concept of a limit model was in- troduced as a means of answering air quality management questions in a multi-source regional context. A particular limit model was developed to assess the relative impact of rural versus urban pollutant sources upon urban air quality (Lamb and Shair, 1977). Assumptions necessary to simplify the calculations were formulated so that the urban impact from rural sources was overestimated while the impact from urban sources was underestimatedo The pollutant concentration ratios thus obtained formed an upper limit on the relative effects of rural versus urban so 2 sources upon the urban region. The purpose of this paper is to demonstrate the versatility of the limit model concept and its simplicity of application to realistic air quality problems. After a brief review of the limit model, attention will be focused upon air quality in the St. Louis area. The model will be used to determine the percentage of time and percentage of urban area where rural power plants dominate urban so 2 levels. This analysis will serve as a means of ranking the rural sources in terms of impact upon urban air quality. Finally, the model will be employed to determine annual average so 2 concentrations in the urban region. 231 2. Review of the Limit Model Expressions The St. Louis urban area is a densely populated zone (pop. 710,000, U.S. Dept. of the Interior, 1970) surrounded by five large power plants which are located between 19 km and 49 km from the edge of the urban region. Following Williams et al. (1967) convention, we define the urban zone as the 18.9 km by 18.9 km square shown in Figure 1. Williams et al. described this area as containing essentially all of the urban so 2 sources. Urban and rural emission rates are given in Table 1 (Littman, 1975) along with important rural source parameters. Although rural sources are located as far as 49 km from urban St. Louis, the data in Table 1 indicate that the combined rural so 2 emissions are three times as large as the total urban emissions. The commonly used Gaussian plume model described by Turner (1970) was employed in Part I to characterize the dispersion of emissions from the rural power plants. Dispersion of urban emissions was described in Part I using a Gaussian expression based on an infinite series of line sources. The expressions associated with rural and urban sources are, respectively, (1) and Xu = (2) Figure 1. @ @ o ~ RAPS SITES 2 5 3 4 5 10 AMEC POWER LANT 0117 PLANT \ PLANT BALDWIN POWER "'· £0WAROSVILL£ y N The urban region of St. Louis and five surrounding rural power plants. KILOMETERS O 1 2 --------- MISSOURI State Highways Stole Boundary U.S. Highways In ter state Routes ILLINOIS 0123 N N w 233 TA13LE 1 EMISSION RATES OF so 2 ASSOCIATED WITH URBAN AND RURAL SOURCES Average Sulfur Dioxide Emission Data for St. Louis (Littman , 19 75) . A. B. Ou Ou Tons ~o 2 day Grams so 2 second 885 9,300 - 35 920 370 9,670 Qr Qr Baldwin-Illinois Power 2. Labadie-Union Electric 3. Sioux-Union Electric 4. Meramec-Union Electric 5. Wood River-Illinois Power 1,026 838 475 280 228 10,800 8,800 5,000 3,000 2,400 Total Rural Sources 2,847 30,000 ,. Urban Sources Point Sources - mostly industrial and utility power plants located within the urban region of interest 2. Non-point Sources Total Urban Sources Rural Sources * ,. RURAL POWER PLANT PARAMETERS Power Plant xr, Distance from urban area (km) Baldwin Labadie Sioux Meramec Wood River 47.2 49.2 22.4 22.4 18.4 13, Bearing angle with respect to urban area (clockwise from north) (deg) 147 259 344 202 8 Hr, Stack height (meters) 75 210 180 60 60 *Low-level industrial sources primarily located near the Wood River Power Plant have not been considered. 234 where xr and Xu correspond to ground-level pollutant concentrations which result from rural and urban emissions with rates Qr and Q. u The nomenclature, assumptions, and conventions associated with Equations (1) and (2) are discussed in Part I. One means of evaluating the relative impact of rural and urban sources upon urban air quality is to determine the ratio of xr and Xu within the urban zone. Dividing Equation (1) by Equation (2) yields ijJ(x,y) = Xr = _l_ Xu ffTT This ratio can be used to determine the percentage of the urban zone where rural sources dominate urban air quality. As indicated in Part I, Equation (3) can be solved to give (4) .!.::2 where y0 (x,ijJ 0 ) is the crosswind distance from the plume centerline to the point where ijJ = ijl0 , evaluated at x, the distance from the upwind edge of the urban zone; ijJ 0 is a specified value of the concentration ratio. When a rural plume is carried directly through the urban zone, the function y0 (x,ij10 ) outlines the area within the urban zone where the rural source has an impact at least ijl 0 times greater than the impact 235 from urban sources. For example, if ~o is taken to equal 1, then xr ~ ~' and y0 (x,1) outlines the portion of the urban zone where at least half of the total observed so 2 concentration is due to a rural source. The percentage of the urban area, A, where~~ ~o can be calculated as (5) In this expression, x0 is the distance downwind of the urban edge to where y0 (x,~0 ) equals zero. As noted in Part I, the value of x0 can be determined by interpolation using a second degree polynomial fit to ~(x,o); all integrals can be evaluated numerically using Simpson's rule. Values of A will henceforth be termed percentage impact areas; it should be understood that in all cases A refers to the percentage of the urban area where~~ ~0 • Tracer studies, as well as measurements of so 2 profiles in power plant plumes, indicate that the rural plumes are narrow with respect to the width of the urban zone (Lamb et al., 1975, Vaughan et al., 1975). Equation (5) implies that near the leading edge of the urban zone, y0 (x,~0 ) is less than one-half the width of the urban area. As will become apparent, this implication holds for the rural sources under consideration. In the remainder of the paper, the St. Louis limit model expressions will first be used to determine the percentage impact area due to each power plant under all stability conditions for a range of specified concentration ratios. The relationship between percentage impact area and frequency of occurrence for the power plants will then 236 be developed utilizing wind direction and stability class frequency data. The results of these calculations will be used to determine an impact function for each power plant in order to compare the effects of the rural sources upon urban so 2 levels. In the final section, the limit model will be used to calculate the annual average so 2 concentration in urban St. Louis and the ratio of the annual average so 2 concentrations caused by rural and urban sources. 237 3. Impact of Rural Sources upon Urban St. Louis The percentage impact areas associated with each power plant were calculated for McElroy (1969) stability classes A1 B1 and C' with ¢0 = 1, 2, and for stability classes D d (daytime), D n (nighttime), and E F 1 with ¢0 = 1, 2, ... 5. 1 1 For classes A B and C 1 1 areas were less than 1% for all ¢0 : 3. 1 , 1 the percentage impact A typical outline of rural source impact area is shown in Figure 2 for the Baldwin power plant under neutral, nighttime conditions. As illustrated in the figure under neutral and stable conditions, the area where¢~ ¢0 can sometimes extend completely through the urban region. In every case, the width of the impact area is less than the width of the urban zone. For a given power plant, the length of the impact area decreases as stability decreases, while the width of the impact area increases as stability decreases. As the value of ¢0 increases, both the length and width of the impact area decrease. The effects of these variations upon the magnitude of the impact area are apparent in Table 2 where the percentage impact areas for the power plants are tabulated. The last column in Table 2 is a list of the combined percentage impact areas from the Sioux and Wood River power plants. Examination of the region in Figure l indicates that northerly winds can carry emissions from the Sioux and Wood River power plants over different portions of the urban zone at the same time. For this reason, we will consider the combined impact of the Sioux and Wood River power plants (designated by SiouxWood River) in the remainder of the discussion. Generally, as atmospheric stability increases, the percentage impact area increases. In the case of¢: 1, A is larger under class 238 PCJNER PLANT • BALOW I N ATM. STABILITY CLASS• □• NIGHTTIME 16 12 ~ WIND 2. 1. 8 8 12 16 URBAN AREA 18.9 KM BY 18.9 KM Figure 2. Description of the portion of the urban area where (~ = 1, 2, 3, and 5) which results from BaTdw9n em9ssions under D'n stability conditions. ~ > ~, 239 TABLE 2 PERCENTAGE OF THE URBAN AREA WHERE~~ ~o' A(%) Stability Class A1 B1 c· D'day D'night E1 F1 A1 B1 c• D'day D'night E1 F1 D'day D'night E' F' D'day D'night E'F' Power Plant SiouxBaldwin Labadie Sioux Meramec Wood River Wood River ~ ~ 1 17 8 5 1 1 6 27 15 12 3 9 3 32 25 15 10 23 8 39 34 28 19 8 11 33 27 16 14 27 11 ~ ~ 2 2 <l 2 <l 1 2 4 3 2 1 1 3 16 12 8 12 5 4 21 13 8 12 5 4 25 13 13 8 6 5 ~ ~ 3 10 7 5 3 8 3 11 6 5 3 3 7 17 7 8 4 4 3 ~ ~ 4 8 5 2 2 6 4 7 4 5 4 2 2 10 4 5 3 3 2 ~ ~ 5 D'day D'night E1 F1 6 5 7 4 3 3 3 3 2 2 2 2 2 2 2 4 4 4 240 D'n conditions than under class E'F' conditions because the impact area extends completely through the urban zone under both stability conditions and the impact area is slightly wfder under neutral, nighttime conditions than under stable, nighttime conditions. For a given stability class, the percentage impact area scales with rural emission rate, in spite of differences in rural transport distances. Under all conditions, the largest impact area is caused by the Baldwin power plant which also has the highest emission rate of S0 2 o As expected, for a given stability class and a particular power plant, the percentage impact area decreases as the value of ~o increases. In the case of the Labadie power plant under class D'n conditions, the value of A decreases elevenfold as ~ovaries from 1 to 5. During unstable conditions, the percentage impact area becomes very small (<1%) as the value of ~o increases beyond 2. In no case under unstable conditions does a rural power plant impact a significant portion of the urban area at so 2 levels greater than three times the so 2 levels caused by urban sources. It appears that the presence of an inversion layer does not significantly affect the impact of the three rural sources closest to the urban region; the percentage impact areas for classes D'd and D'n are almost identical for each power plant. The only difference in the two stability conditions is the height of the inversion layer. Under worst-case atmospheric conditions, for~~ 1, the largest impact area which occurs covers no more than 39% of the urban area. 241 For~> 2, no more than 25% of the urban area is affected, and, for ~ ~ 5, no more than 7% of the urban zone is impacted. Although a rural source can greatly increase urban so 2 concentrations, such increases only occur in relatively small fractions of the total urban area. 242 4. Frequency of Rural Source Impact The preceding analysis described consequences of rural power plant emissions being transported directly over the urban region of St. Louis. However, a realistic comparison of rural and urban sources must account for the time variability of emission rates, wind directions, and atmospheric stability conditions. Although emission rate data are not available as a function of time, the expressions for Xr and Xu can accommodate changes in Qr and Quon time scales longer than a few hours. The effects of variations in meteorology can be determined using wind direction frequency data (diagramed in Figure 3) and stability class frequency data. The data in Figure 3 were taken from wind roses measured 1100 feet above the surface at Lambert Airport, St. Louis, for the ten year period, 1950-1959 (Martin et al., 1967). Although wind roses were available on a seasonal basis, variations were small; the data presented in Figure 3 represent yearly averages. Stability class frequencies, also available seasonally, were averaged to give the following annual values: E'F', 29%. A'B', 6%; C', 10%; D d, 34%; D n, 21%; and 1 1 These data were obtained from Holzworth (1975) for the five year period, 1965-1969. In this model, we assume there is no correlatioo between wind direction and atmospheric stability. A brief inspection of the meteorological data indicates that southerly winds occur approximately 10% of the time and that northwesterly winds occur roughly 8% of the time. Generally, there is little difference between daytime and nighttime frequencies. and nighttime conditions occur 55% of the time. Neutral daytime Stable conditions account for 29% of time, while unstable conditions occur only 16% of the . I I I _ _r--7 80 160 200 WIND ANGLE (DEG) 120 SIOUX 240 280 LABADIE L_.J I I I I 320 360 I I N ~ w Yearly av era ged f r eq ue ncy of wi nd directi on ve r sus wind direct ~o n for th e St. Louis r eg i on (Marti n et al ., 1967). 40 ,_.J L-7 I I I I -7 BALDWIN MERAMEC r- r-, FREQUENCY OF WIND DIRECTION VERSUS WIND DIRECTION Dayt ime Nightime WOOD RIVER 00 2I 4 6 8 10 Figure 3. LL 0:: w 0 :::> w z >u 0 ::-e - - 12 14 16------.----,.--......---,.------,----.--....-----.---,.---,--------.----.---.---,-----,--"""T"""---, 244 time. The data imply that emissions from the Meramec power plant are transported through the urban zone almost twice as often as emissions from any of the other rural sources. The stability frequency data indicate that the stability conditions associated with the largest percentage impact areas occur the greatest frequency of the time. These observations can be examined quantitatively by calculating the frequencies with which plumes from each rural source are transported through the urban zone. In order to simplify this calculation, we will assume that a rural plume is transported either completely within the urban zone or completely outside the urban zone. Within this "hit-or- miss"concept, a plume is defined to be completely within the urban zone if the plume centerline is no more than the distance y0 (x 1,~0 ) (Equation (4)) outside of the urban area. Although this definition slightly overestimates the rural impact, the small plume widths and large transport distances associated with rural emissions are such that very small changes in wind direction cause plumes to move completely in or · out of the urban area. Because y0 goes to infinity at x=o, we will use x1 = 02/50 = 378 mas the leading edge of the zone for purposes of the calculation. The wind angle window associated with a rural plume crossing the urban zone can be determined as follows: (6) where the wind angle window extends from e1 to e2 and 8 is the bearing angle of the source with respect to the urban center (Table 1). 245 The wind angle window is a function of emission rate, transport distance, stability class, and the value of ~0 • For each wind angle window, there is a corresponding percentage impact area, as given in Table 2. Furthermore, each wind angle window is associated with a wind direction frequency which can be determined from the data shown in Figure 3. If the percentages of time related to the wind angle intervals in Figure 3 are assumed to be uniform within each interval, a frequency can be assigned to each wind angle within the interval. Sunming these angle frequencies from e1 to e2 for each source yields the desired wind direction frequency, fw. The quantity fw is the percentage of time during which a rural plume is transported through the urban zone under a particular atmospheric stability condition. If the wind direction frequencies under a given stability condition are summed over all of the rural power plants, the relationship between percentage impact area and frequency of occurrence can be determined for the combined rural sources. As shown in Figure 4, the relationship takes the form of a six-step curve for each stability class. Three of the steps correspond to the impact from Baldwin, Labadie, and Meramec, respectively; two of the steps correspond to the impact caused either by Sioux or Wood River, alone. The sixth step corresponds to the simultaneous impact of Sioux and Wood River. The percentage impact areas are presented in decreasing order. The curves indicate that the wind angle window and, hence, the wind direction frequency, are almost independent of stability class. Some portion of the urban area experiences significant so 2 rural source impact during about 65% of the time under any given stability class. .,. 30 40 0 0 10 I ·-·-1 •. •••••••••••• ·= 20 .L----,1 . . .I • 30 FREQUENCY (%) 40 - - 50 --60 •.·I . --.,·, • ..... • •• •• • • • • • •.. •. ,. . . . . . . 1 --- - - ·- . ·-•- -· I I 70 Percentage impact area (i 2 l, all rural power plants), as a function of frequency of occurrence during specified atmospheric stability conditions. 10 E'F' D' (nightime) L__ --,-tI --- __ _------------.,, . -- . - I . ll:J.7 I -- ·-- 1 c' o' (daytime) 1 A 8 ATMOSPHERIC STABILITY CLASSES I - - - - - .,L - - - - - , _ ___ °'1 : L I • • __ .,____ I . L.:.=.:.=;7 II I . - - - -:iI ••••••••••• ·r... I • • • • • • • ••I • ----- I 1 ------,1L-- ·-·7i Figure 4. <[ ~ 0 - O 20 LL 0:: :::, CD <t z <t 0:: w <t 1\1 PERCENTAGE IMPACT AREA VERSUS FREQUENCY OF OCCURRENCE 50-----...------.----...--------.--.------.-------.---- ~ ;)-;> 247 The step curves describe the percentage of time that a percentage impact area occurs during a specified stability condition. For example, during class C' conditions percentage impact areas less than 27% and greater than 10% occur 30% of the time. However, stability class frequency data indicate that class C1 conditions only exist 10% of the Thus, values of A between 27% and 10% which exist under class C1 time. conditions actually occur only 3% of the time. Similar weighting of wind direction frequencies with respect to stability class frequencies for all stability classes yields the impact frequency, F, over all time. A new step curve relating A to F for the combined rural sources can be diagramed for a series of specified concentration ratios, as shown in Figure 5. In this case, each step in a curve corresponds to a per- centage impact area caused by a particular power plant under a particular stability condition. According to these step curves, rural sources govern air quality 4% of the time in no more than 39% of the urban area. For up to 52% of the time, air quality in some portion of the urban area can be influenced by rural sources at least five times as much as by urban sources. How- ever, urban sources have a greater effect than rural sources during 34% of the time. The local urban sources contribute the greater part of existing so 2 concentrations in about 61% of the area all of the time. The patterns exhibited in Figure 5 illustrate the sharp decrease in percentage impact area associated with increasing concentration ratios. However, the frequencies change less than twenty percentage points as ~ovaries from 1 to 5. This results because the initial widths of rural plumes at the leading edge of the box do not change .___--,, "' ~5 -i.. "'~3 "'~2 "'~1 10.0 20.0 ij0.0 F, FREQUENCY Ct) 30.0 50.0 60.0 70.0 Percentage impact area (all rural power plants) as a function of frequency of occurrence for a range of specified concentration ratios. 0.0 O.OL-----...L..-------'--------'-------'-----._._---_ __._~.__- 10.0 20 .0 Fi_gure 5. <( . ~ LL D :::J a: ro a: :z a: w a: a: 30.0 IJO.O so.a N co +:> 249 drastically as ~o increases . The calculated wind angle windows change only a few degrees as ~o increases. Similar observations can be made on a more detailed scale by considering the impact of the individual power plants. In Figure 6, percentage impact area is given as a function of frequency for each power pl ant and for the Sioux-Wood River complex. In the diagram for Sioux-Wood River, the curves include the impact which occurs when a plume from either Sioux or Wood River crosses the urban zone, as well as the impact which occurs when plumes from both sources cross the region simultaneouslyo The Baldwin and Labadie power plants each dominate urban air quality approximately 10% of the time; the Sioux, Meramec, and Wood River sources each govern urban so 2 levels no more than 18%, 24%, and 15% of the time, respectively; the Sioux-Wood River complex influences urban so 2 levels no more than 22% of the time. Concentration rat i os greater than or equal to 5 are found in less than 10% of the urban area for any power plant, but such ratios occur as much as 20% of the time. Meramec affects urban air quality the largest percentage of the time, but it does so in less than 14% of the urban area. The rural power plants can be ranked by impact upon urban so 2 levels, by considering an impact term, I, which is a function of percentage impact area, impact frequency, and concentrati on rat i o. The magnitude of I will serve as a measure of the total so 2 impact of a rural source; as total impact increases, the value of I increases. We define the impact parameter, I, to equal the volume under the surface formed by taking A versus F versus~. This three-dimensional function 00 F(%} 10 ~?.1(256) 00 1/1~5 (40) 10 F(%) IIJ,?.3(69) I \/f ?.2 (113) I 10 A(%} II I 10 F(%} 20r MERAMEC 00 \/I ?.3 (65) (34) A(%} 1/1?.5 \/I?. 2 (115) 10 20 20 "'- 3 (201 (34) 10 1"1?. 2 (60) 20 00 tJ, ?.5 (54) IO 1'?.3(99) A(%} 20 10 F(%) t/t?.I (374) 20 F(%) 30..- SIOUX-WOOD RIVER , 1/1?:5 00 A(%} 1.,, > 10 Pgrcentage impact area as a_function of _jmpact fr~quency by power plant. T~e area under each curve, IA(~ 0 ), shown in parentheses, is a measure or the impact of a rural source for a particular value of ~o- IA(4 ) = -55(Bdn), 36(Lbd), 46(Sio), 5l(Mer), 26(WR), 72(S-WR ) . o F(%} Figure 6. 0 t?.5(26) \jl?.3 (55) 10 10~ o/ ?.5 (40) 20J- 20 ~ \jl?.I (258) LABADIE 30J- I 40 A(%} I 10 i.Jt?. I (283) BALDWIN A(%). o/ ?.2(147) 30 40 SIOUX 20i:::......... I llJt?.I (253) WOOD RIVER 0 N u, 251 can be visualized by considering the curves of A versus Fin Figure 6 as horizontal cross sections of an irregular pyramid with height measured in terms oft. The peak of the pyramid, at tmax' is taken as the integer value oft where the maximum value of A becomes less than 0.01%. We also define the quantity, IA(t 0 ) to equal the area under a curve of A versus F fort 2:_ t 0 ; IA(t 0 ) is a measure of the impact of a rural source in terms of percentage impact area and impact frequency at a specified concentration ratio. For the purpose of this analysis, the impact function, I, was developed using five discrete values of t 0 ranging from l to 5. A more detailed analysis could be performed using smaller intervals of t0• Values of IA(~ 0 ) were calculated for each power plant over the range 1 ~ ~o ~ 5 by summing the areas under each step in the curves in Figure 6. The value of I was estimated for each power plant by summing the volumes of the cross-sectional layers from t 0 = 1 to ~o = 5 and assuming that the volume under the function from ~o = 5 to ~max was equal to the volume of a regular pyramid with a base of area IA(S) and height equal to (~max-5). The volume of each cross-sectional layer was estimated to equal the product of the height of the layer times the average area of the layer. The average area of a layer was calculated by averaging the areas IA(~ ) 0 and IA(~ +1). 0 Values of IA(~ ) are given 0 in Figure 6, and values of ~max' and I are shown in Figure 7 for each rural power plant. Each horizontal cross section in the pyramids diagramed in Figure 7 has an area equal to IA(~ 0 ) (as shown in Figure 6), a frequency equal to the total impact frequency (from Figure 6), and a I=.850 BALDWIN .087 "'I \ \ 35l Ff 111 =.b61 1't I MERA - .325 34 I=.495 LABADIE .254 Figure 7. . 170 "' I \ ~ 23\ SIOUX I=l.082 I '···-· "' "' I\ .218 .149 WOOD RI VER N <.n N Impact of rural power plants upon urban St . Louis in terms of fractional impact area, A, total fractional impact occurrence Ff, and foncentration ratio, w. The volume of each pyramid, I, is a measure of the total impact of a rural source upon the urban zone . / 27 253 percentage impact area equal to IA(¢0 )/F. Smoothing the data with respect to percentage impact area yields a simplified version of the total impact function. At this point, the values of I can better be brought into perspective by considering fractional areas and fractional times, instead of percentages. Thus, a value of I=l could represent a power plant which dominates one-third of the urban area (i.e.¢.::_ l) one-third of the time with ¢max= 27. It should be noted that large values of ¢max correspond to small values of A which occur near the leading edge of the urban zone. That is, emissions from the rural power plants are bound to dominate the pollutant concentrations at the upwind edge of the urban region when it is assumed that no sources (other than the power plants) exist upwind of the urban region. caused by Baldwin emissions, equals .850. The maximum value of I, The Sioux-Wood River complex yields larger values of I than Baldwin does, although the values of ¢max for the sources are nearly equal. The combined effects of all of the rural sources yield an impact parameter equal to 4.123. In terms of total impact, the power plants are ranked as follows: Sioux-Wood River, Baldwin, Sioux, Meramec, Labadie, and Wood River. This order is considerably different from the order based only upon percentage impact area. The results given in Table 2 in terms of percentage impact area scale directly with rural source strength. The results shown in Figure 7 include the effects of wind direction and stability class frequency. It appears in this case that the meteorology of the region dictate, to a large extent, the impact of a rural source. The second largest so 2 source, Labadie, has the second lowest impact because it is located 254 relatively far from the urban zone within an infrequently occurring wind corridor. The Sioux-Wood River complex has a combined emission rate slightly smaller than the Labadie emission rate. Because of a large wind angle window and the moderately high frequency of winds which blow through the window, the Sioux-Wood River complex has the largest impact of all rural sources. Meramec, which has an emission rate slightly more than half that of Sioux, has a total impact which measures only 5% less than the total impact estimated for Sioux. Meramec is located within the wind angle interval which has the most frequently occurring winds. The Wood River power plant has the lowest emission rate among the rural sources and also has the lowest estimated total impact. Further insight into the relation between the urban impact of rural sources and the meteorology of the St. Louis region can be obtained by diagraming the percentage impact area as a function of wind direction as shown in Figure 8. The values of A, F, e1, and e2 given in the figure have been averaged with respect to stability class frequency. In terms of wind angle windows, Meramec, Sioux, and Wood River have roughly equal sizes, 59°, 59°, and 67°, respectively. These windows are almost double the size of the windows associated with Baldwin and Labadie, 37°, and 35°, respectively. The window associated with the simultaneous impact of Sioux and Wood River is 39° wide. The combined rural sources contribute the greater part of urban so 2 levels during 67% of the time through 61% of the wind angles. . <( 'N. ID LL :::) cc cc co z w a: cc cc 0.0 20.0 L!O. 120 . I Figure 8. 80 I I 60.0 80.0 100.0 160 BFIUJf l N I I I 200 HEHMC 27.t 21.J:0 280 IL.reAOIEI WIND ANGLE I 320 I 360 5laJX l«JOO RI\IEFI ~ l!O Percentage impact area as a function of wind direction for¢: 1. I I 11.t A,AVERRGED ~VER ATMOSPHERIC STABILITY CLASS FREQUENCY F=t OF TIME A, OCCURS 80 N (J1 (J1 256 Thus, it can be seen that due to the geographical distribution of the power plants, rural emissions are transported through the urban region a majority of the time. 257 5. Average so 2 Concentrations in Urban St. Louis The preceding application of the limit model provided a means for determining the relative impact of rural and urban sources upon urban air quality in terms of percentage impact area and percentage impact time. Average so 2 concentrations within urban St. Louis can be calculated by averaging the rural and urban dispersion expressions (Equations (1) and (2)) over the urban area. The resulting average expressions can then be weighted with respect to wind direction and stability class frequencies to give the ratio of the average rural and urban so 2 concentrations and the total of the average so 2 concentrations in the urban zone on a yearly basis. The latter value can be compared to available air quality data for St. Louis . As described in Appendix A, averaging the rural dispersion expression over the urban area yields: (8) Xr = where y1 = Averaging the urban dispersion expression over the urban area gives the average so 2 concentration caused by urban sources : (1-n) He (n) exp[-½He (n) 2 ]dn. (9) 258 The calculation of the average concentration of so 2 caused by a rural source, must take into account the effects of changes in wind direction and stability conditions. Thus, the time and area averaged concentration caused by a rural source is written: k=E'F' <x r > = L Xr(k) • f w(k) • fk (10) k=A'B' where k represents the stability class, fw(k) is the impact frequency, and fk is the stability class frequency. This expression is similar to that from Pooler (1966), except that he included the occurrence of wind speed classes in place of stability classes. The calculation of the average so 2 concentration resulting from urban emissions must account for only changes in stability conditions. Changes in wind direction do not affect the average urban source concentration because emissions are assumed to be uniformly distributed throughout the urban zone. The time and area averaged so 2 concentration caused by urban sources is written: k=E' F' <xu> = L xu(k) • fk (11) k=A'B' We now define <x>t to be the sum of the average rural and urban concentrations, while <\/J> ts defined as the ratio of the average rural and urban concentrations. Calculated values for these terms are listed in Table 3. The power plants individually contribute less than 8% of the area averaged so 2 concentration that results from urban sources. Because the concentrations in the rural plumes are averaged over the entire urban TABLE 3 AVERAGE S0 2 CONCENTRATIONS AND CONCENTRATION RATIOS IN THE ST. LOUIS URBAN REGION <x r > µg SO/m 3 <x u> µg so 2;m 3 <x>t µg SO/m 3 <\jJ> Baldwin 21 259 280 0.081 Labadie 15 259 274 0.058 Sioux 18 259 277 0.069 Meramec 21 259 280 0.081 Wood River 10 259 269 0.039 85 µg/m 3 259 µg/m 3 344 µg/m 3 0.328 Power Plant TOTAL k=E' F' <xr> = Xr(k) • fw(k) . fk I: k=A'B' k=E' F' <x u> = xu(k) • fk I: k=A B 1 1 <x>t = <x r > + <x u> <x r > -<x> u where xr(k) = the average urban so 2 concentration due to a rural power plant under stability class k; xu(k) = the average urban so 2 concentration due to urban sources under stability class k; fw(k) wind frequency associated with a specified power plant under stability class k; fk = stability class frequency. 260 area and because under worst-case conditions, no more than 39% of the urban zone is dominated by a rural source, the power plants, individually, have only a small influence on the average so 2 level observed in the urban zone. The largest so 2 rural source, Baldwin, and the most frequently impacting source, Meramec, have equal average concentration ratios. They are followed in order of decreasing values of<~> by Sioux, Labadie, and Wood River. This order is similar to that based upon the total impact function given in the previous section. Collectively, the rural sources contribute one-third the impact of urban sources upon urban so 2 levels. The combined rural emissions account for 25% of the average urban so 2 concentration. Even though the rural source impact has been calculated as an upper bound, the rural power plants yield an average concentration ratio of only 0.33. As we have observed, this ratio is much less than the values of~ that can occur in portions of the urban region on a short-term basis. While it is difficult to obtain experimental data on the relative impacts of rural and urban sources, data are available regarding ambient so 2 concentrations in St. Louis. As indicated in Table 3, the calculated concentration is 344 µg/m 3. Williams et al. (1967) evaluated data from 20 to 40 sampling stations and found that 24-hour averaged levels ranged from 43 to 276 µg/m 3 during the winter of 1964-65. Shir and Shieh (1974) presented 25-day averages of 2-hour averaged data that ranged from 50 to 250 µg/m 3 at 10 St. Louis monitoring sites during February, 1965. The same authors used a complex diffusion model to predict maximum 2-hour averaged so 2 concentrations that ranged 261 from 335 to 1428 µg/m 3 for conditions existing during February, 1965. Gifford and Hanna (1973) reported an average so 2 concentration equal to 161 µg/m 3 for St. Louis. The annual and 24-hour federal air quality standards for so 2 are 80 µg/m 3 and 365 µg/m 3, respectively. The measured values cited here suggest that the average so 2 level predicted by means of the limit model generally serves as an upper bound on the data. However, concentrations calculated with the limit model are based upon dispersion terms determined from 10-minute averaged concentrations. The effects of ~oncentration sampling times and wind direction measurement times have not been considered in the calculation of the average so 2 concentration. Further comparison between a complex, multi-source diffusion model such as the one presented by Shir and Shieh and the limit model using the same data base would be interesting. A more complete air quality and meteorological data base would be of considerable value in validating the limit model. Determination of correlations of urban so 2 concentrations with stability condition, wind direction, and mixing depth would also be particularly interesting. 262 6. Summary and Conclusions The concept of the limit model may be summarized by considering the step-by-step application of our particular model to the St. Louis region: (1) The geographic context and source distribution suggested appropriate derivations of expressions describing the dispersion of rural and urban emissions. (2) The sensitivity of the percentage impact area expression was assessed with respect to the independent variables. (3) The percentages of urban area where rural sources govern urban so 2 levels by a specified ratio were calculated for all atmospheric stability conditions. (4) Wind direction and stability condition frequency data were used to determine the percentages of time that rural emissions were transported through the urban region. Percentage impact area was thus described as a function of impact frequency. (5) Results of the impact area and time calculations were summarized by defining an impact function as equal to the volume under the surface formed by taking fractional impact area versus fractional impact time versus concentration ratio. The relative magnitudes of the impact functions provided a means of ordering the rural sources with respect to impact upon the urban area. 263 (6) The ratio of the average urban so 2 concentrations caused by rural and urban emissions and the sum of the average concentrations were determined. Wind direction and stability condition frequencies were taken into account in this procedure. Limit models generally will involve steps to describe pollutant transport and dispersion, to define a measure of relative impact (upper or lower-bound) among different sources, and to account for the effects of sourcedistributions and regional meteorology upon the defined impact variable. Models developed along these lines will serve, qualitatively and quantitatively, as indicators of the general relationships which exist among pollutant sources and receptors in a particular geographic context. In the St. Louis region, application of the limit model concept indicated that, in terms of percentage impact area alone, urban impact of rural sources scaled directly with source strength. Under worst- case conditions, the largest so 2 source, Baldwin, governed urban so 2 levels in no more than 39% of the urban zone. Consideration of the effects of variability of wind direction and atmospheric stability indicated that these conditions were found to exist only 4% of the time. The combined sources had a greater influence upon urban so 2 levels than did urban sources during 67%of the time; rural sources had five times the effect of urban sources in a portion of the urban region during 52%of the time. 264 On an individual basis, the rural power plants governed so 2 levels between 8% and 25% of the time. Evaluation of the total impact of each power plant indicated that emission rate, transport distance, and meteorology were all determining factors in calculating the relative impact of a source. Listing the rural power plants in decreasing order of total impact upon the urban area yields: Sioux-Wood River, Baldwin, Sioux, Meramec, Labadie, and Wood River. In spite of producing rural to urban concentration ratios as high as 35 in small fractions of the urban area, the rural power plants are predicted to contribute only 25% of the annual, area averaged so 2 concentration. The ratio of the average concentrations caused by rural and urban emissions was found to be 0.33. Even though the limit model favored the impact of rural sources upon urban air quality, these sources contributed only one-third the average level of so 2 caused by urban sources. 265 APPENDIX A Determination of Average so 2 Concentrations Within the Urban Area The average concentration of so 2 within the urban area resulting from rural emissions can be calculated with the expression: (A-1) which is simply the Gaussian dispersion expression for an elevated point source averaged over the length, D2, and width, o1, of the urban zone. Equation (A-1) can be simplified by setting y = )I terms within each integral: ./'lay and collecting r 2 e-y dydx where y1 = D1 (A-2) The second integral is directly related to the error function and, thus, Equation (A-2) can be written: dx. (A-3) 266 This expression can be evaluated numerically to give the area averaged so 2 concentration caused by rural emissions. An expression describing the average urban concentration of so 2 resulting from urban emissions can be derived from: Xu= Q 1x=D21n=x/D2 2 4vTT uHDuD He(n) exp[-½He(n) ]dndx l 2 x=o n=o (A-4) which is the area source dispersion expression averaged over the length of the urban zone, o2. It is not necessary to average Xu over the width of the box because there are no lateral dispersion terms. In Equation (A-4), n = x-a (where x-a is the source-to-receptor distance) and D2 Equation (A-4) can be nondimensionalized further by setting n* = L to yield D2 _ ./l Qu ln*=l1n=n* Xu= 7rr uHDl n*=o n=o (A-5) The variables in the limits of the integrals can be changed and the order of integration reversed so that: 1 n=l ./'l. Qu Xu = Tn uHD 1 n=o - 1 n*=l 2 H (n) exp[-½H (n) J e e n*=n dn*dn (A-6) 267 which simplifies to the final expression: - ./l. 1 Qu Xu = 7rr uH o 1 1 0 (A-7) 268 Acknowledgments We are delighted to thank William Wilson, James McElroy, and Francis Pooler of the U.S. Environmental Protection Agency for their interest, advice, and comments. We thank Glen Cass of the Environmental Quality Laboratory, California Institute of Technology, for his very helpful suggestions concerning the application of the limit model. We wish to thank Lou Howard of the Massachusetts Institute of Technology for his stimulating discussion. Finally, we thank Fred Littman of Rockwell International for his help in obtaining emission data, and George Holzworth of the U.S. Environmental Protection Agency for providing stability frequency data. This work was supported by the U.S.E.P.A. under Grant No. R802160-03-2. 269 Literature Cited Gifford, F. A., and S. R. Hanna (1973) "Modeling Urban Air Pollution , " Atmos. Environ. I, 131-136. Holzworth, G. (1975) Personal corrmunication. North Carolina 27711. Research Triangle Park, Lamb, B. K., J . D. Bruchie, and F. H. Shair (1975) "Tracer Studies for Characterizing the Transport and Dispersion of Plumes Emitted at Ground and at Elevated Levels," U.S. Environmental Protection Agency EPA Grant No. R802160-03-2, Research Triangle Park, North Carolina. Lamb, B. K., and F. H. Shair (1977) 11 A Limit Model for Determining the Impact of Rural Power Plant Emissions Relative to Urban Emissions upon Urban Air Quality. Part I. Development and Sensitivity Analysis of the Limit Model." Submitted for publication to Atmos. Environ .. Littman, F. (1975) Personal conmunication. Thousand Oaks, California. Rockwell International. Martin, D. D., P. A. Humphrey, and J. L Dicke (1967) "Interstate Air Pollution Study, Phase II Project Report, 5," National Center for Air Pollution Control, Cincinnati, Ohio. McElroy, J. L. (1969) 11 A Comparative Study of Urban and Rural Dispersion," i• ~-Meteor.~, 19-31. Pooler, F., Jr. (1961) 11 A Prediction Model of Mean Urban Pollution for Use with Standard Wind Roses, 11 Int. i • Air Water Poll. i, 199-211. The National Atlas of the United States (1970). the Interior, Washington, D. C.. U.S. Department of Shir, C. c., and L. J. Shieh (1974) 11 A Generalized Urban Air Pollution Model and Its Application to the Study of S02 Distributions in the St. Louis Metropolitan Area, 11 i• ~ - Meteor. ll, 185-204. Turner, D. B. (1970) "Workbook of Atmospheric Dispersion Estimates," Environmental Protection Agency No. AP-26, Research Triangle Park, North Carolina. Vaughan, W. M., R. B. Sperling, N. V. Gillani, and R. B. Husar (1975) ''Horizontal so 2 Mass Flow Rate Measurements in Plumes: A Comparison of Correlation-Spectrometer Data with a Dispersion and Removal Model, 11 presented at the 68th Annual Meeting of the Air Pollution Control Association, Boston, Massachusetts. 270 Williams, J. D., G. Ozolins, J. W. Sadler, and J. R. Farmer (1967) "Interstate Air Pollution Study, Phase II Project Report, 8, 11 National Center for Air Pollution Control, Cincinnati, Ohio. 271 CHAPTER 7o DETERMINATION OF CONCENTRATIONS OF HALOGENATED COMPOUNDS DISSOLVED IN VARIOUS LIQUIDS BY ELECTRON CAPTURE GAS CHROMATOGRAPHY by Brian K. LamB and Fredrick Ho Shair The following reprint appeared in Analytical Chemistry 48, 1976. 272 Determination of Concentrations of Halogenated Compounds Dissolved in Various Liquids by Electron Capture Gas Chromatography Brian K. Lamb and Fredrick H. Shair• Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, Calif. 91125 Determination of halogenated compounds dissolved in liquids by electron capture gas chromatography Is three to four times faster than usual volumetrlc-manometrlc solubility methods, requires no solvent degassing, and Is capable of accuracies of approximately ±6%. Measured solubilities of SF 6 in water over a range of temperatures were In good agreement with literature values. The solubility of SF 6 in a homologous series of n-alcohols was also determined. The error In this case was within predicted limits, although larger than satisfactory . Means of improving the reproducibility are discossed. The technique is capable of measuring dissolved halogenated compounds at extremely low concentrations and is expected to form the basis for the development of a multiple liquid tracer system. Knowledge rega rding the concentra t ion of gases dissolved in liquids is of importa nce to a wide vari ety of technica l interest s. In addition to determining the solubility whi ch will occur under well-specified equ ilibrium con d itions, there is ofte n a need to measure the concentra tion of a gas or liquid t race r which is prese nt in a liquid sample. In some cases, for example , tracer investigations, the concen tration of the trace r in the liquid phase will be orders of magnitude lower th a n that which would exist in t ypi ca l phase equ ilibria experiments. In an ex te nsive rev iew of gas solubility methods, Battino a nd Clever (I) have di scussed th e traditiona l volumetricmanom et.ric methods used t.o d etermin e the solubil ity of gases in liquids. Th e vo lum etri c- manometric meth ods involve sa turating a previously d egassed solvent with the gas of interest und er co nditions where appropriate volumes, pressures , and temperatures may be accurately measured. Equilibrium betwee n th e gas and liquid phase is genera lly ac hi eved by s tirring a mixture of the two, bubbling the gas through the liqui d, or circul ating a film of th e liquid through the gas. Careful d egass ing requires :l to 5 hours followed by equ ilibrium times of :l to 12 hours. These methods ca n yie ld gas solubility da ta thought to be accu rate to better than I%. The purpose of this work was to d evelop a sensitive tec hnique accurate to approx im a tel y ± 5%, which would avoid solvent degassing and which would require much less time for a na lysis than the traditional methods. We re port he re s uch a technique whic h is based upon a n a nalysis by means of electron captu re gas chromatography. Consequently, thi s method can be used to determine the solubility of gases and liquids which have rela tively high e lectron-ca pture cross- sections such as SFi;, Freons, a nd other halogenated compounds. This technique offers a s imple and useful method for gain ing more information a bout the anomalous beha vior of fluoro compounds in solution . In particular, the conC'e ntra tion of such gases as SFi;, CC l:iF, and CH:iCl can easily be measured in various solvents, non polar and polar , including those possessing high vapor pressures at the tempera! ures of interest. The rapid determination of concentrations of ha logena ted compounds in air a nd dissolved in water, pa rti cularly sea water, a ppea rs very importa nt in view of the curre nt de bate concerning stratosph eri c ozone d ecompos ition by Freons and other ha logenated compound s. Ocea ns have been suggested as possible natural sources and sinks for some of these com pound s. Th e technique described here may be useful in dete rmining the oceanic conce ntrations of halogenated compou nd s. EXPERIMENTAL To test the acc uracy of the method , the solubility of SFf; in wate r from 10 to SO °C and in the s traight.chain alcohols from metha nol to heptanol at :lf>.0 °C was dete rmined. The S F,; was obtnined from Ma theson, dis till ed wate r was ta ken fr om the building: s up p ly, a nd th e alcohols, s pectroquality, were purchased from AldriC"h Chemical and Matheson, Colema n and Rell. Temperature was con trolled to ± ll. 2 °C in a large Dewar flas k with a Neslah Ci rcula ting Temperature Controll er. Water satura tion too k place in ;).0-cm:1 polye thy lene disposab le sy rin ~es a ttached to a syringe ca r ouse l threaded on the shaft of a va ri a bl e-speed electri c stirre r. Du rin g sa turation , the sy rin ge:-- a nd ca rousel we re subme rsed in the hat h a nd s pun as rap idly as possible. Initiall y each s.',·ringe was flus hed severa l times with SFi; a nd tht-n fill ed with .l_O cm :i of S F,;. The ca pped syringes were pl aced in the temperaturl! hath for a pproxima te ly l f> min . For the same lengt h of time, a glass screw -top via l contai ning :.W cm;1 of distilled wate r was a lso plan•d in the ha th . After both t he gas a nd wate r we re te mpe ra ture equilibrat ed, 1.0 <'ITI :1 of wa te r from the via l was quic kl y drawn into Pach sy rin ge, a nd the syring e ca pped a nd dampt!d ve rti <'ai l)' to the ca rouse l. Norma ll y, filling: and d ;.unping: ('ould he don e in lt>ss I han GO s. During: the wa te r tes ts, three syring es were pre part-'d a nd spun a t. eac h temperatu re. The pa rtial press urP of SF,, in the syrin j:{e was s imply the diffe r ence between t he atmosphe ri c press ure in th e room measured on a me rcury ba rom eter and the va por press ure of wat e r a t. the saturation temperature. Tlw low solubility of SF1; in the s mall amount of sol ve nt µ resent caust-'d less than a Jl )l i press ure dro p in the syrin ge due lo disso luti on of the gas, while a llowing equ ilibrium lo he reached within ver)' short period s of time. Th e sy rin ges we re s pun 40 min to e nsure sa turation , a lthough preliminary tests indica ted equilibrium was reac heci in :W min. After spinning , eac h syrin ge was a llowed to res t in the bath at least 10 min before being: a na lyzed. Allowing th e syrin ges lo rest :l h be for e a nal ys is did nol cha nge the res ul ts. For the alco hol tes ts, '2().c m:1 sy ringes contai ning: I c m :1 of solve nt. we re used t.o maintain a neg:ligihl e pressu re drop. Six diffe re nt a lcohols we re simultaneo usly saturated with SF,; by s pinning for 90 min . J.:ac h syringe was allowed to rest a l least :W min prior t.o ana lys is. Although t he alco hol res ults we re not as re produ cible as t.he wate r tests, no con~istent correlation between resting time a nd res ults was found, giving no indication of a problem with dispersed gas hubbi es in the solve nt.. Althou~h SFfi is ex tremely insoluble in wate r, the sensitivity of t.he electron ca pture d ete<· tor to SFf. req uired tha t S Fwsa turated water be diluted a t least three orders of mag nitude prior to injection into the chromatog rap h. It is this pro perty that gives the me thod its import.a nt fl ex ibility for analyz ing compounds unde r hot.h equilibrated and extre mely no nequilih ra ted conditions. Dilution was accomplis hed by mea ns of n well -mixed e xponentia l dilu tion system s imil a r tot.hat used hy Drivas et al. {2. 3 ). A fi X 6 X 6 inch Lucite cube was huilt with a mag:netica ll y dri ve n fan posit ioned in side. A slow nitroge n now rate of 120 cm:1/ minute ass ured 273 RES ULTS AND CONCLUSIONS 2 m,n lmin 0 TIME Figure 1. Typical SF 6 chromatogram: ISF 6 j = 9.63 X 10- 15 mol SF 6 /cm 3 . Sample taken during dilution of 50 °C H 2O solubility sample, Xs,, = 2.28 X 10- 5 mol fraction good mixing. If thE.- system were well mixed, the dilution would follow where q is the nitrogen flow rate through the cube, V is the cube volume, and t is the time since flow began; (q/V)t is simply the number of resident air changes N in the cuhe si nce flow began. The measured sample concentration at N is given by C, and the unknown initial concentration in the cube is given by Cu. A graph uf In(' vs. N should yield a straight line with slope of -1. The inlt>rcept of the graph givt>s the initial concentration in the cube. For l~a('h watN test . 1.0 µI of saturatt>d water was transferred to the cul)(' with a :iO-µl S(;J,,: s~·ringr. The cuht> was sealed. and the n,ntents wt'rt' mixt>d for :W min prior to starting the nitrogen !low. Tht> saluralt•d watt'r i11jPdt'll into t!w t'uhe evaporate<l in less than :"1 min. thus lilwrating dissolved SF,;. Flow through the rube was nwasured with a PrPdsion Scientil'ic \,\'et -Tes! Mt>ter. Ont' dilulion run l'onsiste<l of fi\·p l'Uht• concentration measu rements taken ovt>r an hour pE>rio<l . Aflt~r ci t·o mplet e ,rnal)'·sis, thP cuhe was rapidly !lushed with nitrot!ell in preparation for the nex t dilution run . Cube concentrations were measured using a Loenco Model 70 (;as Chromatograph equipped with an 8-port Valeo gas sampling valve fitted with a matched pair of 1-cm:1 sample loops. The cont·e ntric electron captu re det<•ctor contained a 200-mCi tritium foil, which under pulsed voltage operating conditions produced a standing current of :i X 10-i1 A. An 8-rt X 1hi-in. sta inless s tee l column packed with H0- 100 mesh l'orapak Q was fitted to a Carle fi. port switching valve t.o allow so lvent venting. The column eluted O:! in 2!") s, SFn in 70 s, and water and the alcohols sometime afte r !Of> s. A typica l chromatogram is shown in Figure I. The column venting is an important part of this technique since the radioactive foil hecomes rapidly conta minated and performance dete ri orates if solvents are allowed to pass through the detector. The carrier gas was prepurified nitrogen run at I JO ml/min. An injection port on the chromatograph a llowed direct injection of solvent blanks. Samples containing low levels of dissolved fluorochemicals, as in trace r experiments, can a lso he injected directly. The injection port, column, and detector temperatures were respectively 160, no, and 120 °C. The SF,; peak was integrated with a Spectra-Physics Autolah System I digital integrator. To avoid abso rption problems, calibrations were begun hy in jecting 1.0 µI of pure SF.; into the dilution cube and measuring peak area with respect to the number of air changes. A graph of In peak area vs. number of air changes p roduces a slope of -1.00. The initial concentration divided by the peak area intercept yields a calibration facto r, typically 1000 µV-s ec/lo-"• mol SF,;, that is prog-rammed into I.he integrator for the analyses. The detection runge is from fl X 10 - 1:.t to:, X 10- w mo! SF,i, and the range is line-ar from !"l X lll-J :i mo! tu tht> dt•lPt"tion limit. All solubility mt•asuremPnts were made in the linear rn.nge. The system wns modified for thP alcohol tests hy fitting tht• c;c sample valve with a IA-µ! samplP loop, which was ca librated during ,, CC t·alih ra tion with res pe<'l to a l-cm;1 loop. Also, only 0.!"l µ I of saturated solvent was transfe rred. to the cube from the saturation s~·ringe. These chang-es were necessar)' het·ause of the m\H·h greater soluhilit)"· of SF,; in h)·dro<·a rhons. In all cases. the solvents evttporated in the rube in less than 10 min. A least squares computer fit In C vs. N for each dilution gave the initial cube concentration and, hence, the moles of SF6 per moles of water in 1.0 µI or, more simp ly, the mole fraction. Only dilutions which gave a s lope within ±0.0fi of -1.00 were accepted for calculating the results. In only two or three instances were dilutions not acceptable. The mole fraction of SFi; in water plotted vs. temperature is shown in Figure 2. The data of Ashton et a l. (4) , Morrison and ,Johnstone (5), and Friedman (6) are also presented. The work of Ashton and co-workers offers the best refe rence ranl(e of data and also the most accurate resulLs , with an experimental error of 1%. Our results agree favorably with those of the Ashton group, being in perfect agreement a t 50 °C and 7% high at 10 °C . The ave rage standard deviation of our results over the temperature range was 6%. The Ashton coworkers noted that their results were higher than those of Morrison and ,Johnstone and Friedman, with discrepancies being largest at lower temperatures. This difference was credited to more complete satu ration in the Ashton work. Similarly, we note that our results are slightly higher at lower temperatures than any of the referenced papers. It is possible that our method represents the most complete saturation of the reported techniques. It is interesting to realize that our data consisting of 18 separate ana lyses were taken during a per iod of four working days, while a single solubility ana lysis from the Ashton group volumetricmanometric method required 12 hours for saturation to occur. The soluhi!ity of SFi; inn -alcohols from C, to C7 plotted vs. solvent skeleton lenl(th is l(iven in Fil(ure :l. Data for several other solvents ta ken from th e 1(8S so lubili ty dat.a re view hy Wilhelm and Hattino are also l(iven (7). Chain lenl(ths and molecular sizes were calculated from bond lenl(th and hond anl(le <l a t.a tak en from ref. Ii. The so luh ility of SFi; increases in the alcohols with in creasinl( cha in lenl(th , which corresponds to increasing nonpolar character. In addition, the mole fraction of Sr',; levels off with hexanol and heptanol, mirroring th e behavior of heptane and octane . The SF,; molecular size, as indicated in Figure :l, is within a factor of two of that for most. of the solvent molecu les for whi ch measurements have been reported. The maximum uncertainty, calculated in the standard manner as given by Bevington (9), for the measurement of the mole fraction of SFi; in water was ± 12%. The data were easily within that limit, with an average standard deviation of ±6%. The predicted uncertainty for measuring the mole fraction of SF,; in the alcohols was ±22%, while the experimental average deviation was ±20%. With water as so lvent, the major source of experimental error as predicted by the ca lculations was the measurement of the very smal l volumes of saturated wa ter injected into the dilution cube. It accounted for 10 out of the 12% predi cted uncertainty. With alcohols as solvents, the major sources of error were the measurements of the small volumes injected into the cube and the even smaller volumes injected from the cube into the chromatograph. Together, these measurements accounted for 19 out of the 22% predicted uncertainty. Without these sma ll volume uncertainties, the technique might be capable of solubility measurements of about 2 to 3%. Adsorption of SF,; on the Lucite cube or on the stain less steel GC column is not considered a significant sou rce of error. During typical atmospheric SF,; tracer ca librations in this laboratory using the dilution method (10), linear SFs cal ibration curves are routinely measured down to 1 part SF,; in 10 12 parts air. Work in this laboratory on suitable Freon sampling techniques bas shown that some Freons which are adsorbed by Lucite are not absorbed by Tedlar 274 12)( I 10 z 0 9 V 8 ;:: C6 M,:,(>H 0 0 10' 3 l!:" 7 w _, 6 z Q t- u 0 ::;; ~. JeH1a oC S 2 oCCt 4 II <( 5 a:: ... 4 LL ~ wi 0- 4 .!!!., _j 3 0 ::e 00'--'--,Lo--'~2"0, -~ 730 ~ -'----,40'-:--~---,,"o__, T, 'C H - OH Figure 2. Solubility of SF 6 in water : (e) this work; ( □) Ashton et al. (1968); (0) Morrison and Johnstone (1955); (Li.) Friedma n (1954) 6 10- ~ 0 - ~ - - ~- I5 3.0 ~ - - ~- -~ - ~ ~ - ~ 4.5 6.0 7.5 9.0 CHAIN LENGTH OF SOLVENT (A) ( 11 ). In these instances, a Tedlar-lined dilution cuhe would permit utilization of the so lubilit y technique. Hester el al. (12) and Donohoe ( / /) have reported the use of stain less st.ee l columns for the separation of Freons with no adsorption observed. To ac hievp reprodu cihilit v of" th e procedure close to the predirted limit , ±:l•~ .. the unce rtaint:,· associated with sma ll volurnc>s must he elimina ted. l lse of an internal standard pn.>sent in the so(n.. nt in accuralel~· known <·oncentrations and detectable h:,· electron capturp would a llow pPak a rea comparisons lwtwPPll tht:1 standard and th e dissolved gas or liquid . Thi s ,·ornparison would yie ld the soluhi lit:,· of t.he ~as without tlw uncntaint:,· of small ,·ol11ml's. Tlw c;c can lw cal ibrat ed to within ±:l•~, for a particula r n>rnpound, and the sta nd arcl can hP dilutc>d in th,• solven t to within ±2•1.,_ A spec ific stanclard is not prpsentl\' known. hut the necessary propNt ies of s uch a standard ca n he exami ned. It must he so lubl e in th e so lvent. hut unreacti\'e with solvent and solu te. T he sta ndard must he sepa rable from the solute a nd so lvent hy chromatograp hy ancl sensi t ivP at low levels to e lectron captu re detection. Finally, at ve ry low concentrations, it must not a ffec t. th e so lubility oft.he gas in the .solven t. In many w,iys, low molec·ular wc>ight fluorinated alcohols fit this clescript ion. However, t horoul(h testing for the a hove properties woulcl he necessar:,· heforc> a speci fic se lect.ion ('l)u lcl he mad e. Such a standard, when proved suit able, shou ld limit so lubilit y meas urement uncertainties to±:,%. The extens ion of this technique to include other compounds und e r saturated and nonsaturat.ed cond itions may involve three distinct sa mple handling techniques. For solubility stud ies, the saturated sample must generally be diluted in a su itable manner (i.e., dilution chamber) prior to injection in order to avoid saturation of the electron captu re detecto r. For tracer studies or for a few environmental trace a nalyses , the dissolved com pound concentrations may he such that samp les ran he injected directly into t.he gas chrom atog raph. For most e nvironmental tracr a na lyses. t.l1P concentration of the substances is low eno ugh to re qu ire s tripping and conc(' nt.raling of th e com pound prior to inject ion into the chroma tograp h ( /, /.'/). The e lectron capture )(as chroma togra phy met.hod of nwasuring e lectron capturing l(ases d issolved in liquicl s is capable of a<'curate ('orH·ent ra t ion determ inations undf:\r Figure 3. Solubility of SF 6 in various solvents: (e"') thi s work; ( □ ) Wilhelm and Battino (1973) condition s of we ll -spec ifi ed equ ilibri a with ana lysis times three to four times faster tha n usual volu me! ric- ma nometri c methods. Since di sso lved gases represent the most d iffi cult a nalytica l cha llenge, one may expect t hat the technique will he eas il y appli ed lo the measurement of detect.abl e liqui ds dissolved in othe r liquids. With this procedure , gas or li<.p1id solute concentrations may he determined over a n orders-of-magnitud e range under a va ri ety of cond i- t.ions. The combi ned properties of' this technique result in a n ana lytica l procedur,• with exci ting possibi lities for application in a wide spectrum oft heorelical an d real systems. Future wo rk will hP concentrated on t.he development of severa l ap pli cations. LITERATURE CITED (1) R. Battino and H. L. Clever, Chem. Rev .. 66, 395 (1966). (2) P. J. Drivas and F. H. Shair. Atmos . Environ .. 8 , 475 ( 1974). (3) P. J. Drivas . P . G . Simmonds, and F . H. Shair, Environ. Sci. Technol.. 6 , 609 (1972). (4) J. T. Ashlan, A. A. Dawes. K. W . Miller, E. B. Smith, a nd B. J. Stickings, J. Chem. Soc. A. 1793 t 1968). (5) T. J . Morrison and N. B. B. Johnstone. J. Chem. Soc .. 3655 ( 1955). (6) H. L. Friedman. J. Am. Chem. Soc .. 76, 3294 ( 1954). t 7) E. Wilhelm and R. Battino, Chem. Rev , 73, t tt973). (8) R. C. Weast, Ed., " Handbook o f Chemistry and Physics," 50th ed .. The Chemical Rubber Co .. Clevela nd , Ohio, 1970, p F-155 . (9) P . R. Bevington , "Data Reduction and Error Analy sis fo r the Physical Sciences ," McGraw-Hill, New York, 1969, pp 56 - 65 . (10) 8 . K . Lamb, J . D . Bruchie. a nd F. H. Sha ir. " Tracer Studies for Characterizing the Transport and Dispersion of Plumes Emitted at Ground and a t Elevated Levels,' · (Prelim inary Report) ; prepared for the U.S. Environmental Protection Agency, EPA Grant No. R802160-03-2. 1975, pp 45 - 53 . ( 11) K. G . Donohoe . personal communical ion. 1975 . ( 12) N. E. Heste r. E . R. Stephens, and 0 . C . Taylor, Atmos. Environ .. 9, 603 (1975). (13) P. Hanst, L. Spiller. 0 . Watts. J. Spence, and M . Miller. J. Air Po/Jut. Contr. Assoc .. in press. HE<'EI VEll for rev iew Sept e mber 19, l97!i. Accept.eel De ,·e mhe r 4, 197:,. This work was supported in part. hy the l 1.S. KP .A. under ('ontract No. fi8-0:l-0-1:l4 a nd <:rant No. H80'.!lli0 -0:1-:!. Th,· co nt p"1.s do not n,•,·pssaril y n•fl1•1·t t.lw v ipws and polir.v of tlw Environmt'nt.al Prof.Pelion AgPn<'y . 275 Literature Cited (Chapter 1 and Chapter 2) Angell , J. K., C. Ro Dickson, and W. H. Hoecker, Jr. (1976) "Tetroon Traj ectories in the Los Angeles Basin Defining the Source of Air Reaching the San Bernardino-Riverside Area i n Late Afternoon, 11 y_. .8£.E]_. Meteor . .l.§_, 197-204. Angell, J. K., D. H. Pack, G. C. Holzworth, and C. R. Dickson (1966) "Tetroon Trajectories in an Urban Atmosphere," il_. M£.l. Meteor . .§._, 565-572. Angell, J. K., D. H. Pack, L. Machta, C. R. Dickson, and W. H. Hoecker, Jr. (1972) "Three-D·imensional Air Trajectories Determined from Tetroon Flights i n the Planetary Boundary Layer of the Los Angeles Basin, 11 i • 6EE]_. Meteor. l l9 451-471 o Ayer, H. S. (1961) "On the Di ssipation of Drainage Wind Systems in Va ll eys in Morning Hours," i• w_:i_. Meteor. 1.§_, 560-563. Barad, M. L. (1959) "Analysis of Diffusion Studies at O'Ne ill/' Advances .iD_ Geophys . .§_, 389-398. Batchelor, G. K. (1949) "Diffusion in a Field of Homogeneous Turbulence. I Eulerian Analysis," Australian il_. Sci. Res. _g_, 437-450. Batchelor, G. K. (19 59) 11 Note on the Diffusion from Sources in a Turbu l ent Boundary Layer," unpublished. See : Batchelor, G. K. (1964) "Diffusion from Sources in a Turbulent Boundary Layer, 11 Arch. Mech. Stosowanej. l, 661-670. -- -Bauer, E. (1974) "Dispersion of Tracers in the Atmosphere and Ocean: Survey and Comparison of Experimental Data," I· Geophys. Res. 79, 789-795. Berg, T. C. , and B. Sivertsen (1977) "An Electronic Monitor for Measuring Atmospheric Turbulence," Norwegian Institute for Ai r Research, Lillestr¢m, Norwayo Bevington, P. R. (1969) Data Reduction and Error Analysis for the Physical Sciences. McGraw-Hill, New York, pp. 56-65. Bi erly, E.W . , and E. W. Hewson (1962) "Some Restrictive Meteorological Conditions to be Consi dered i n the Design of Stacks/' I· !ill.Pl• Meteor. l, 383-390. Bowman, M. C., and c. R. Nony (1977) "Tracer Analysis of Estradiol in Animal Chow by Electron Capture Gas Chromatography," Chromat. Sci. li, 160-163. r. 276 Briggs, G. A. (1969) Plume Rise. USAEC Critical Review Series TID25075, National Technical Information Service, Springfield, Virginia 22151. Bri ggs, G. A. (_1972) "Discussion on Chimney Plumes in Neutral and Stable Surroundings," Atmos. Environ . .§_, 507-510. Cadle, S. H., D. P. Chock, J.M. Heuss, and P. R. Monson (1976) 11 Results of the General Motors Sulfate Dispersion Experiment," General Motors Research Laboratories GMR-2107, EV-26a, Warren, Michigan 48090. Calder, K. L. (1977) "Multiple Source Plume Models of Urban Air Pollution - Their General Structure," Atmos~ Environ. ll, 403-414. Carpenter, S. B., T. L. Montgomery, J.M. Leavitt, W. C. Colbaugh, and F. W. Thomas (1971) "Principal Plume Dispersion Models , " J. Air Poll. Contr. Assoc. n_, 491-495. - - -Case, J. R., and F. Nyman (1962) "Some Chemi cal Reactions of Sulphur Hexafluoride," Nature 193, 473. Chamberlain, A. T., and J. S. Marlow (1977) "Solvent Systems and Applications of the Liquid Chromatograph Electron Capture Detector," i• Chromat. Sci. li, 29-31. Chock, D. P. (1976) "The General Motors Sulfate Dispersion Experiment: Assessment of the EPA HIWAY Model," General Motors Research Laboratories, GMR-2126, EV-27, Warren, Michigan 48090. Christophorou, L. G., D. L. Mccorkle, J. G. Carter (1971) "Cross Sections for Electron Attachment Processes Peaking at Subthermal Energies," .Q_. Chem. Phys. 54, 253-260. Cicerone, R. J., R. S. Stolarski, and S. Walters (1974) 11 Stratospheric Ozone Destructi on by Man-Made Chl orofl uoromethanes, Science 185, 1165-1167. 11 Cl emons, C. A., and A. P. Altshuller (1966) "Responses of ElectronCapture Detector to Halogenated Substances," Anal. Chem. 38, 133-136. Cramer, H. E. (1957) "A Practical Method for Estimating the Dispersal of Atmospheric Contaminants~" in Proceedings of the First National Conference Q1l MP lied Meteorology. •American Meteoro 1ogi cal ·Society, Boston, Massachusetts, pp. C-33-C-55. Crutzen, P. J. (1974) "Estimates of Possible Future Ozone Reductions from Continued Use of Fluoro-Chloro-Methanes (CF2Cl2, CFCl3), 11 Geophys. Res~ Letters l, 205-208. 277 David, D. J. (1974) Gas Chromatographic Detectors. Sons, New York, pp. 76-113. John Wiley and Davidson, B. (1963) "Some Turbulence and Wind Variability Observations in the Lee of Mountain Ridges,"~. .B.Q.Ql. Meteor .. £, 463-472. De Bartoli, M., and E. Peechio (1976) "Measurements of Some Halogenated Compounds in Air over Europe," Atmos. Environ . .lQ, 921-923. Defant, F. (1951) "Local Winds," in Compendium of Meteorology. American Meteorological Society, Boston, Massachusetts, pp. 662-667. Devaux, P., and G. Guiochon (1969) "Variations of the Response of the Electron Capture Detector with Carrier Gas Flow Rate, 11 J. Chromat. Sci. ]_, 561-564. Dickson, C.R. (1976) Personal communication. Chief, Air Resources Laboratories Field Research Office, National Oceanic and Atmospheric Administration, Idaho Falls, Idaho 83401. Dietz, R. N., and E. A. Cote (1973) "Tracing Atmospheric Pollutants by Gas Chromatographic Determination of Sulfur Hexafluoride, 11 Environ. Sci. Tech.]_, 338-342. Donohoe, K. G. (1975) Personal communication. Laboratory, Moffett Field, California. NASA Ames Research Doucet, J., R. Gilbert, P. Sauvageau, and C. Sandorfy (1975) 11 Photoelectron and Far-ultraviolet Spectra of CF3Br, CF2BrCl, and CF2Br2, 11 ~- Chem. Phys. 62, 366-369. Drivas, P. J. (1974) 11 Investigation of Atmospheric Dispersion Problems by Means of a Tracer Technique, 11 Ph.D. Thesis, California Institute of Technology, Pasadena, California 91125. Drivas, P. J., and F. H. Shair (1974) 11 Dispersion of an Instantaneous Cross-wind Line Source of Tracer Released from an Urban Highway, Atmos. Environ.§_, 475-485. 11 Drivas, P. J., and F. H. Shair (1974) A Tracer Study of Pollutant Transport and Dispersion in the Los Angeles Area, 11 Atmos. Environ. §_, 1155-1163. 11 Drivas, P. J., and F. H. Shair (1975) The Chemistry, Dispersion, and Transport of Air Pollutants Emitted from Fossil Fuel Power Plants in California; Transport and Dispersion of Plumes Associated with Coastal Meteorology," California State Air Resources Board Contract No. ARB 3-915, Sacramento, California 95814. 11 278 Edinger, J. G. (1959) "Changes in the Depth of the Marine Layer over the Los Angel es Basin, 11 i• Meteor. 1.§_, 219-226. Edinger, J. G. (1969) "Meteorology," from Combustion Generated Air Pollution. University of California, Berkeley, California. E. I. du Pont de Nemours and Company ( 1971) "Toxi co 1ogy of DuPont FE130l Fire Extinguishing Agent," S-35A, Wilmington, Delaware 19898. Ellison, T. H. (1959) "Meteorology," Sci. Prog . .iZ_, 495-506. Fabrick, A., R. Sklarew, and J. Wilson (1977) "Point Source Model Evaluation and Development Study," California Air Resources Board and the California Energy Resources Conservation and Development Conmission Contract No. A-058-87, Sacramento, California 95814. Fenimore, D. C., P. R. Loy, and A. Zl atki s ( 1971) "High Temperature Tritium Source for Electron Capture Detectors," Anal. Chem. 43, 1972-1975. -- -- Fosberg, M.A., D. G. Fox, E. A. Howard, and J. D. Cohen (1976) "Nonturbulent Dispersion Processes in Complex Terrain," Atmos. Environ . .lQ_, l 053-1055. Fosberg, M. A., and M. J. Schroeder (1966) "Marine Air Penetration in Central California," i• ~ . Meteor. i, 573-589. Frenzel, C. W. (1962) "Diurnal Wind Variations in Central California," i• ~- Meteor. l, 405-512. Gifford, F. A., Jr. (1960) "Atmospheric Dispersion," Nuclear Safety l, 56-62. Gifford, F. A., Jr. (1960) "Atmospheric Dispersion Calculations Using the Generalized Gaussian Plume Model, 11 Nuclear Safety _g_, 56-59. Gifford, F. A., Jr. (1961) "Use of Routine Meteorological Observations for Estimating Atmospheric Dispersion, 11 Nuclear Safety _g_, 47-51. Gifford, F. A., Jr. (1968) "An Outline of Theories of Diffusion in Lower Layers of the Atmosphere," in Meteorology and Atomic Energy 1968, D. H. Slade, ed .. TID-24190, u. S. Atomic Energy Conmission, Oak Ridge, Tennessee, pp. 101-103. Gifford, F. A., Jr., and S. R. Hanna (1971) "Urban Air Pollution Modeling," in Proceedings of the 2nd International Clean Air Congress. Academic Press, New York, pp. 1146-1151. 279 Gifford, f. A., Jr., and S. R. Hanna (1973) "Modeling Urban Air Pollution," Atmos. Environ.]_, 131-136. Giroux, H. D., L. E. Hauser, L. H. Teuscher, and P. Testerman (1974) "Power Plant Plume Tracing in the Southern California Marine Layer," from Proceedings of the Symposium on Atmospheric Diffusion and Air Pollution. American Meteorologica1Society, Boston, Massachusetts 02108, pp. 238-245. Halasz, S. P., and O. Glemsar (1971) Sulfur ..in_ Organic and Inorganic Chemistry Vol. l, A. Senning, ed .. Marcel Dekker, Inc., New York, p. 217. Hanna, S. R. (1971) 11 A Simple Method of Calculating Dispersion from Urban Sources," i• Air Poll. Contr. Assoc . .£!_, 774-777. Hay, J. S., and F. Pasquill (1959) "Diffusion from a Continuous Source in Relation to the Spectrum and Scale of Turbulence," Advances .iD.. Geophys . .§_, 345-364. Hester, N. E., E. R. Stephens, and 0. C. Taylor (1974) "Fluorocarbons in the Los Angeles Basin," i• Air Poll. Contr. Assoc. 24, 591-595. Hester, N. E., E. R. Stephens, and O. C. Taylor (1975) "Fluorocarbon Air Pollutants II, 11 Atmos. Environ. 2._, 603-606. Hewson, E.W. (1945) "The Meteorological Control of Atmospheric Pollution by Heavy Industry," Quart. i• .B_. Meteor. Soc. ll., 266. Hickam, W. M., and R. E. Fox (1956) "Electron Attachment in Sulfur Hexafluoride Using Monoenergeti c Electrons, 11 i• Chem. ~ - ~' 642-647. Hinds, W. T. (1970) "Diffusion Over Coastal Mountains of Southern California," Atmos. Environ . .i, 107-124. Holland, J. Z. (1953) "A Meteorological Survey of the Oak Ridge Area," Report OR0-99, U.S. Atomic Energy Corrmission, Oak Ridge, Tennessee. Hovind, E. L., T. C. Spangler, and A. J. Anderson (1974) 11 The Influence of Rough Mountainous Terrain upon Plume Dispersion from an Elevated Source," from Proceedings of the Symposium on Atmospheric Diffusion and Air Pollution. American Meteorological Society, Boston, Massachusetts, pp. 214-217. Islitzer, N. F. (1961) "Short-Range Atmospheric Dispersion Measurements from an Elevated Source," i• Meteor . .l§_, 443-450. 280 Islitzer, N. Fo, and Ro K. Dumbauld (1963) "Atmospheric Diffusiondeposition Studies Over Flat Terrain,'' · Int. ·J . Air Water Poll . ]_, 99_9.--1022. - Johnson, W. B., Ro C. Sklarew, and D. B. Turner (1976) "Urban Air Quality Simulation Modeling," in Air Pollution Vol. I, 3rd ed., A. C. Stern, ed .. Academic Press, New York, pp. 503-562. Juret, R. S., Jr., ands. P. Cram (1974) "Gas Chromatography," Ana 1. Chem. 46, 101 R-124R. Kao, S. K. , H. N. Lee, and K. I. Surdy (1974) "A Preliminary Analysis of the Effect of Mountain-Valley Terrains on Turbulence and Diffusion," from Proceedings of the Symposium on Atmospheric Diffusion and Air Pollution. American Meteorological Society, Boston, Massachusetts, pp. 59-63. Kapila, S., and W. A. Aue (1976) "The Effect of Pressure on the Response of a D. C. Electron Capture Detector," i• Chromat . 118, 233-235. Koch, R. C., and G. E. Fisher (1973) "Evaluation of the Multiple Source Gaussian Plume Diffusion Model," U.--- S. Environmental Protection Agency Contract No. 68-02-0281; available from NTIS PB-249-062. Koch, R. Co, and S. D. Thayer (1971) "Validation and Sensitivity Analysis of the Gaussian Plume Multiple Source Urban Diffusion Model," U. S. Environmental Protection Agency Contract No. CPA 70-94; availabl e from NTIS PB-206-951 Koch, R. C., and S. D. Thayer (1975) "Evaluation of the Multiple Source Gauss i an Plume Diffusion Model - Phase II," U. S. Environmental Protection Agency EPA-650/4-75-018-b; available from NTIS PB-249-729. Krey, P. W., R. J. Lagomarsino, L. E. Toonkel, and M. Schonberg (1976) "Atmospheric Residence of SF5, 11 from Health and Safety Laboratory Environmental Quarterly, E. P. Hardy, Jr., ed .. ERDA, New York , pp. 50-57. Lamb, B. K., J. D. Bruchie, and F. H. Shair (1975) "Tracer Studies for Characterizing the Transport and Dispersion of Plumes Emitted at Ground and Elevated Levels," U.S. Environmental Protection Agency Grant No. R802160-03-2, Research Triangle Park, North Carolina 27111. Lamb, B. K., A. Lorenzen, and F. H. Shair (1977) 11 Tracer Study of Power Plant Emission Transport and Dispersion From the Oxnard/Ventura Plain," California Air Resources Board Contract No. ARB-5-306, Sacramento, California 95814. 281 Lamb, .B. K., and F. H. Shair (1977) "Atmospheric Tracer Studies to Characterize the Transport and Dispersion of Pollutants in the California Delta Region," California Air Resources Board Contract No. ARB-A5-065-87, Sacramento, California 95814. Lawrence, J. F., and J. J. Ryan (1977) "Comparison of Electron Capture and Electrolytic Conductivity Detection for the Gas-liquid Chromatographic Analysis of Heptafluorobutyrl Derivatives of Some Agricultural Chemicals." i• Chromat. 130, 97-102. Lester9 D., and L.A. Greenberg (1950) "The Toxicity of Sulfur Hexafluoride,'' Arch. Indust. ~- Occup. Med. ~, 348-349. Lillian, D., and H.B. Singh (1974) "Absolute Detennination of Atmospheric Halocarbons by Gas Phase Coulometry," Anal. Chem. 46, 1060-1063. -- -- Lorenzen, A. (1974) "Climate of the Sacramento Valley Air Basin," California Air Resources Board, Sacramento, California 95814. Lovelock, J. E. (1963) "Electron Absorptive Detectors and Technique for Use in Quantitative and Qualitative Analysis by Gas Chromatography," Ana 1. Chem. 1§_, 474-481. Lovelock, J. E. (1975) "Solute Switching and Detection by Synchronous Demodulation in Gas Chromatography," i• Chromat. 112, 29-36. Lovelock, J.E., and S. R. Lipsky (1960) "Electron Affinity SpectroscopyA New Method for the Identification of Functional Groups in Chemical Compounds Separated by Gas Chromatography," J. Amer. Chem. Soc. 82, 431-433. - - - - - -- Lovill, J.E., and A. Miller (1968) "The Vertical Distribution of Ozone Over the San Francisco Bay Area," i• Geophys. Res. ]l_, 5073-5079. MacCready, P. B., Jr., L. B. Baboolal, and P. B. S. Lissaman (1974) "Diffusion and Turbulence Aloft over Complex Terrain," from Proceedings of the Symposium on Atmospheric Diffusion and Air Pollution. American Meteorological Society, Boston, Massachusetts 02108, pp. 218-225. Maggs, R. J., P. L. Joynes, A. J. Davies, and J. E. Lovelock (1971) "The Electron Capture Detector-A New Mode of Operation ,1' Anal. Chem. 43, 1966-1971. Martin, D. D., P.A. Humphrey, and J. L. Dicke (1967) n1nterstate Air Pollution Study, Phase II Project Report Vol. 5," National Center for Air Pollution Control, Cincinnati, Ohio. 282 Martin, R. R., M. Zutter, and N. R. Anthonisen (1972) "Pulmonary Gas Exchange in Dogs Breathing SF5 at 4 Atm., 11 ;!_. 6e.El· Physiol . .TI_, 86-92. Matheson Company, Inc. (1966) The Matheson Gas Data Book. pp.53-56, pp. 455-457. ----- New Jersey, McElroy, J. L. (1969) "A Comparative Study of Urban and Rural Dispersion," ;!_. ~ Meteor. f, 19-31. McNeil, E. E., R. Otson, W. F. Miles, and F. J.M. Rajabalee (1977) "Determination of Chlorinated Pesticides in Potable Water," J. Chromat. 132, 277-286. Miller, Ao (1968) "~Jind Profiles in West Coast Temperature Inversions," Report No. 4, San Jose State College, Department of Meteorology, San Jose, California. Miller, A. (1973) "Pollutant Distributions in West Coast Inversions," from Proceedings of the Second Joint Conference on Sensing of Environmental Pollutants. Instrument Society of America, Pittsburg, Pennsylvania, pp. 287-293. Miller, A., and C. D. Ahrens (1969) "Ozone Within and Below the West Coast Temperature Inversion," Report No. 6, San Jose State College, Department of Meteorology, San Jose, California. Molina, M. J., and F. S. Rowland (1974) "Stratospheric Sink for Chl orofl uoromethanes: Chlorine Atom-Catalyzed Destruction of Ozone, 11 Nature 249, 810-812. Morgan, T., and R. D. Bornstein (1977) "Inversion Climatology at San Jose, California," Monthly Weather Review 105, 653-656. Neckers, D. C. (1967) Mechanistic Organic Photochemistry. Publishing Corporation, New York, p. 266. Reinhold Neiberger, M. (1955) "Tracer Tests of the Accuracy of Trajectories Computed from the Observed Winds in the Los Ange 1es Area, 11 Air Po 11 uti on Foundation, Los Angeles, California. Noutary, C. J. (1968) "Mass Spectrometric Study of the Photoionization of Some Fluorocarbons and Tri fl uoromethyl Hali des, 11 ;!_. Res. Nat. Bur. Std. 72A, 479-485. Odom, R. W., D. L. Smith, and J. H. Futrell (1975) "A Study of Electron Attachment to SF 6 and Auto-detachment and Stabilization of SF6-," ;!_. Phys.!!_.§_, 1349-1366. 283 Orgil 1, M. M., R. N. Lee, L. L. Wendel 1, R. W. Nickola (1974) "Transport and Diffusion Measurements with an Airborne Gas Chromatograph," from Proceedin s of the Symposium on Atmospheric Diffusion and Air Pollution. erican Meteorological Society, Boston, Massachusetts, pp. 91-94. Paciorek, M. (1977) Personal communication. of Technology, Pasadena, California 91125. California Institute Palmer, D. (1977) "Tracer Characterization of Exhaust Emissions Entry into Automobiles on Freeways," Masters Thesis, California Institute of Technology, Pasadena, California 91125. Pasquill, F. (1961) "The Estimation of the Dispersion of Windborne Material," Meteorology Mag. 90, 33-49. Pasquill, F. (1974) Atmospheric Diffusion. New York. John Wiley and Sons, Pasquill, F. (1976) "Atmospheric Dispersion Parameters in Gaussian Plume Modeling - Part II: Possible Requirements for Change in Turner Workbook Values," U.S. Environmental Protection Agency EPA-600/4-76030b, Research Triangle Park, North Carolina 27711. Pasquill, F., and F. B. Smith (1971) "The Physical and Meteorological Basis for the Estimation of the Dispersion of Windborne Material, 11 from Proceedings of the 2nd International Clean Air Congress. Academic Press, New York, pp. 1067-1072. Pellizzari, E. D. (1974) "Electron Capture Detection in Gas Chromatography," Chromat. Rev. in press. Proceedings of Symposium on Multiple-source Urban Diffusion Models, A. C. Stern, ed., 1970. National Air Pollution Control Administration, APCO Publication No. AP-86, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina. Ragland, K. W. (1976) "Worst-case Ambient Air Concentrations from Point Sources Using the Gaussian Plume Model," Atmos. Environ . .1.Q, 371-374. Ragland, K. W., ands. Dennis (1975) "Point Source Atmospheric Diffusion Model with Variable Wind and Diffusivity Profiles, 11 Atmos. Environ. 2., 175-189. Raynor, G. S., P. Michael, R. M. Brown, and S. Sethu Raman (1974) 11 A Research Program on Atmospheric Diffusion from an Oceanic Site," from Proceedings of the Symposium £!l Atmospheric Diffusion and Air Pollution. American Meteorological Society, Boston, Massachusetts, pp. 289-295. 284 Reynolds, s. D., P. M. Roth, and J. H. Sei nfeld (1973) "Mathematical Modeling of Photochemical Air Pollution-I," Atmos . Environ.]_, 1033-1061. Richardson, L. F. (1926) "Atmospheric Diffusion Shown on a Distanceneighbor Graph," Proc . .8Ql_. Soc. London AllO, 709-737. Roberts, H. L. (1968) Inorganic Sulfur Chemistry, G. Nickless, ed .. Elsevier Publishing Company, Amsterdam, p. 420. Sandberg, J. S., W. J. Walker, and R. H. Thuiller (1970) "Fluorescent Tracer Studies of Pollutant Transport in the San Francisco Bay Area, i• Air Poll. Contr. Assoc. 20, 593-598. 11 Saltzman, B. E., A. I. Coleman, and C. A. Clemons (1966) "Halogenated Compounds as Gaseous Meteorological Tracers: Stability and Ultrasensitive Analysis by Gas Chromatography," Anal. Chem . 38, 753-758. Schmidt, W. (1925) 11 Der Massenaustausch in Freier Luft and Verwandte Erscheinunge," Probleme der Kosmischen Physik J_. Scolnick, M. (1969) "The Concentric Cylinder Electron Capture Detector. Standing Current Studies in the D. C. and Pulsed Voltage Modes, i• Chromat. Sci.]_, 300-304. 11 Seinfeld, J. H. (1975) Air Pollution: McGraw-Hill, New York. Physical and Chemical Fundamentals. Shir, C. C. , and L. J. Shieh (1974) 11 A Generalized Urban Air Pollution Model and Its Application to the Study of S02 Distributions in the St. Louis Metropolitan Area," i• ~ - Meteor . .ll, 185-204. · Shoemake, G. R., J.E. Lovelock, and A. Zlatkis (1963) "The Effect of Temperature and Carrier Gas on the Loss Rate of Tritium from Radioactive Foils," i• Chromat . .11., 314-320. Sides, G.D., T. O. Tiernan, and R. J. Hanrahan (1976) "Measurement of Thermal Electron Dissociative Attachment Rate Cons t ants for Halogen Gases Using a Flowing Afterglow Technique , " i- Chem.~- .2i, 1966-1975. Siegel , M. W., and W. L. Fite (1976) "Terminal Ions in Weak Atmospheric Pressure Plasmas . Application of Atmospheric Pressure Ionization to Trace Impurity Analysis in Gases," i• Phys. Chem. 80, 2871-2881. Siegel, M. W. , and M. c. McKeown (1976) 11 Ions and Electrons in the Electron Capture Detector, 11 J. Chromat. 122, 397-413 . 285 Simmonds, P. G., D. C. Fenimore, B. C. Pettit, J. E. Lovelock, and A. Zlatkis (1967) "Design of a Nickel-63 Electron Absorption Detector and Analytical Significance of High Temperature Operation," Anal. Chem. 39, 1428-1433. -- -Simmonds, P. G., A. J. Lovelock, and J.E. Lovelock (1976) "Continuous and Ultrasensitive Apparatus for the Measurement of Air-Borne Tracer Substances, 11 ~- Chromat. 126, 3-9. Singer, I. A., and M. E. Smith (1953) "Relation of Gustiness to other Meteorological Parameters,"~- Meteor . .J..Q_, 121-126. Singer, I. A., and M. E. Smith (1966) "Atmospheric Dispersion at Brookhaven National Laboratory, 11 ~- Air Water Poll . .J..Q_, 125-135. Sivertsen, B. (1975) "Land-Sea Breeze Studies in Telemark 1974/75," Norwegian Institute for Air Research, Lillestr¢m, Norway. Slade, D. H., ed. (1968) Meteorology and Atomic Energy 1968. Atomic Energy Commission TID-24190, Oak Ridge, Tennessee. U.S. Smalley, C. (1957) 11 A Study of Air Flow Patterns in the San Francisco Bay Area, 11 Information Bulletin 6-15-70, Technical Services Division, Bay Area Air Pollution Control District, San Francisco. California. Smith, T. B., and S. M. Howard (1972) "Methodology for Treating Diffusivity, 11 Meteorology Research, Inc., FR-1030, Altadena, California. Spence, D., and G. J. Schulz (1973) "Temperature Dependence of Electron Attachment at Low Energies for Polyatomic Molecules,"~- Chem. Phys. 58, 1800-1803. Start, G. E., C.R. Dickson, and N. R. Ricks (1974) "Effluent Dilutions over Mountainous Terrain and Within Mountain Canyons, from Proceedings of the Symposium on Atmospheric Diffusion and Air Pollution. American Meteorology Society, Boston, Massachusetts, pp. 226-232. 11 Strom, G. H. ( 1976) "Transport and Diffusion of Stack Effluents, in Air Pollution Vol. I, 3rd ed., A. C. Stern, ed .. Academic Press, New York, pp. 401-501.11 Stukel, J. J., R. L. Solomon, and J. L. Hudson (1975) A Model for the Dispersion of Particulate or Gaseous Pollutants From a Network of Streets and Highways," Atmos. Environ. 2._, 990-999. 11 Sutton, O. G. (1932) "A Theory of Eddy Diffusion in the Atmosphe'r e, Proc. Roy. Soc. London Al 35, 143-165. 11 286 Taylor, G. I. (1921) "Diffusion by Continuous Movements, 11 Proc. London Math.. Soc. _g_, 196-202. Taylor, M. P. (1962) "Hazards of Radioactive Foils in Electron Capture Detectors," i• Chromat. 2., 29-32. Thomsen, E. L., and J. E. Lovelock (1976) "A Continuous and Intermediate Method for the Detection of SF6 and Other Tracer Gases by Electron Capture in Atmospheric Diffusion Experiments," Atmos. Environ. l.Q_, 917-920. Thynne, J. C. J., P. W. Harland, and R. MacDonald (1973) Autodetachment of Electrons in Sulphur Tetrafluoride and Sulphur Hexafluoride, 11 Dynamic Mass .Spectrometry£, 171-176. 11 Trochimowicz, H. J.(1975) Personal communication. Chief, Inhalation Toxicology Section, Haskell Laboratory for Toxicology and Industrial Medicine, E. I. du Pont de Nemours and Company, Wilmington, Delaware. Turner, D. B. (1961) "Relationships Between 24-Hour Mean Air Quality Measurements and Meteorological Factors in Nashville, Tennessee, 11 i• Air Poll. Contr. Assoc . ..!.l, 483-489. Turner, D. B. (1970) "Workbook of Atmospheric Dispersion Estimates," U.S. Environmental Protection Agency, AP-26, Research Triangle Park, North Carolina. Unger, Co D. (1974) "The Climate of the San Joaquin Valley Air Basin, 11 California Air Resources Board, Sacramento, California 95814. Van de Wi el , H. J., and P. Tonmassen ( 1972) 11 Effect of Oxygen on Electron Capture Detection, 11 i• Chromat. 11, 1-7. Vaughan, W. M., R. B. Sperling, N. V. Gillani, and R. B. Husar (1975) "Horizontal so 2 Mass Flow Rate Measurements in Plumes: A Comparison of Correlation-Spectrometer Data with a Dispersion and Removal Model, 11 paper presented at the 68th Annual Meeting of the Air Pollution Control Association, Boston, Massachusetts. 11 Vukovich, F. M., J. W. Dunn III, and B. W. Crissman (1976) A Theoretical Study of the St. Louis Heat Island: The Wind and Temperature Distribution," i• Ae£:!_. Meteor . .l_i, 417-440. Wanta, R. c., and W. P. Lowry (1976) "The Meteorological Setting for Dispersal of Air Pollutants," in Air Pollution Vol.J_, 3rd ed., A. C. Stern, ed .. Academic Press, New York, pp. 327-400. 287 Wentworth, W. E. (1971) Recent Advances .iD. Gas Chromatography, I. I. Domsky, and J. A. Perry, ed .. Marcel Dekker, New York, pp. 105-124. Wentworth, W. E., and R. S. Becker (1962) "Potential Method for the Determination of Electron Affinities of Molecules: Application to Some Aromatic Hydrocarbons. 11 i• Am. Chem. Soc. 84, 4263-4266. Wentworth, W. E., R. S. Becker, and R. Tung (1967) Thermal Electron Attachment to Some Aliphatic and Aromatic Chloro, Bromo, and Iodo Derivatives," i• ~ . Chem. I!_, 1652-1665 . 11 Wentworth, W. E., and E. Chen (1967) i• Gas Chromat . .§._, 170-179. 11 Electron Attachment Mechanisms," Wentworth, \iJ. E., E. Chen, and J.E. Lovelock (1966) "The Pulsesampling Technique for the Study of Electron-Attachment Phenomena," i• Phys. Chem. 70, 445-458. Wentworth, W. E., E. Chen, and J.C. Steelhammer (1968) "Determination of Electron Affinities of Radicals and Bond Dissociation Energies by Electron-Attachment Studies at Thermal Energies--Electron Affinity of Acetate Radical , 11 i • Phys. Chem. 11.., 2671-2675. Wentworth, W. E., R. George, and H. Keith (1969) "Dissociative Thermal Electron Attachment to Some Aliphatic Chloro, Bromo, Iodo Compounds," i• Chem. Phys. fil_, 1791-1801. Wentworth, W. E., and J. C. Steelhammer (1968) Radiation Chemistry 11.., H. S. Hart, ed .. Advances in Chemistry Series, American Chemical Society Publications, Washington, D. C., pp. 75-100. Wentworth, W. E., A. Tishbee, C. F. Batten, and A. Zlatkis (1975) 11 Photo-e 1ectron Capture Detector, 11 i• Chromat. 112, 229-246. Willmott, F. W., and R. J. Dolphin (1974) "A Novel Combination of Liquid Chromatography and Electron Capture Detection in the Analysis of Pesticides," i• Chromat. Sci. ]1_, 695-700. Wood, L. D. H., F. Ruff, A. C. Bryan, and J. Milie-Emili (1969) "Influence of Air Density on the Distribution of Ventilation,t' Physiologist ]1_, 398. Yamartino, R. J., Jr. (1977) 11 A New Method for Computing Pollutant Concentrations in the Presence of Limited Vertical Mixing,"~- Air Poll. Contr. Assoc. ll_, 467-468. Yauger, W. L., L. M. Addison, and R. K. Stevens (1966) J. Assoc. 0ffic. Anal. Chemists 49, 1053. 288 Zimmerman, J. R., and R. s. Thompson (1975) "User's Guide for HIWAY, A Highway Air Pollution Model," U. S. Environmental Protection Agency EPA-650/4-74-008, Research Triangle Park, North Carolina 277110 289 Abstracts of Propositions l. Heats of Reaction for Oscillating Chemical Systems A method is proposed for comparing calculated and experimental heats of reaction for the oscillating Belousov reaction. 2. Determination of the Solubility of Gases in Liquids at High Pressures Solubilities of gases in liquids at pressures above atmospheric can be determined simply and rapidly using a special high pressure saturation vessel in conjunction with electron capture gas chromatography. The design and use of such a device is proposed and mole fraction estimates for SF6 , CF4 , and NF3 in water under high partial pressures are calculated. 3. Determination of Trace Amounts of Atmospheric Sulfur Compounds A method for the determination of trace amounts of atmospheric sulfur is proposed. The procedure involves quantitative conversion of sulfur compounds to sulfur hexafluoride and subsequent analysis by electron capture gas chromatography. 4. Experimental Characterization of Accidental Spills of Liquified Natural Gas A tracer technique utilizing CH F. as a liquified gaseous tracer and S F6 and CBrF as cold gaseous trac~rs is proposed for the analysis of the vaporizatfon and dispersion of a pool of liquified natural gas. Infrared photography is suggested as one means of monitoring the behavior of the cold cloud of gas. 5. The Flux of Naturally Emitted Hydrocarbons into High Pressure Systems, over th.e lhtted States The flux of terpenes from rural vetetation is estimated and the resulting concentration of these reactive hydrocarbons is predicted in the high pressure air masses which periodically cross the United States. The cakulati'on points out the lack of data which currently exists concerning the role of naturally occurring hydrocarbons in photochemical smog. Heats of Reaction for Oscillating Chemical Systems Bray 1 in 1921 reported measuring the oscillating concentration of iodine and the oscillating production of oxygen in the iodate catalyzed reduction of hydrogen peroxide. Belousov 2 in 1959 similarly noted oscillating behavior of Ce(IV)/Ce(III) concentrations in the cerium catalyzed oxidation of malonic acid by bromate ion. An even more startling discovery by Zhabotinskii 3 was the report that spatial patterns developed when the so-called Belousov reaction was left unstirred. Currently there are many in- vestigators studying both the oscillating and pattern forming phenomena. This report deals only with the temporal oscillating Belousov reaction, although the presentation here will have implications in the iodate-iodine system and possibly the pattern forming Belousov system. As might be expected, both temporal oscillating systems exhibit fluctuations in other species concentrations and in physical properties. Liebhafsky 415 has reported measuring oscillations of iodide ion concentration, pH, and temperature as well as iodine concentration, and oxygen production for the Bray reaction. .. .. 6 7 Field, Koros, and Noyes ' have studied the oscillations of bromide ion concentration and the changing ratio of Ce(IV)/Ce(III). Temperature fluctuations have been reported by Busse; 8 no pH measurements have, as yet, been reported. 291 Although much work has been done on the Belousov reaction, there have been no reported studies on the heat of reaction for the system. A very complex, detailed chemical mechanism has been proposed by Field, Kores, and Noyes (FKN). 7 , 9 It appears that heats of reaction for the various major steps in the FKN mechanism can be calculated and then compared with an experimentally determined heat of reaction. This comparison may be accomplished for two cases; the overall experimental heat of reaction can be determined and the heat of reaction for one particular cycle of the reaction can be determined. It is expected that the second case will prove the most interesting and challenging. In this paper, I propose a suitable method for its examination. The FKN Mechanism for the Belousov Reaction The following description of the proposed mechanism has been given by Field, Koros, and Noyes. There are two major processes involved which operate almost independently of one another. One proceeds by the reactions of singlet species, while the second requires the formation and destruction of free radicals. The dominance of Process A over B or vice-versa depends upon the instantaneous bromide ion concentration. Process A dominates above a certain critical bromide ion concentration and Process B dominates below this critical level. The system oscillates / 292 because bromide ion is consumed by Proc e ss A and produced by Process B. These reactions are given by FKN as PROCESS A + 2H+ + HBrO + HOBr 2 + H+ + 2HOBr Br 2 + (Br + HOBr + H + Br 2 + H2O) x 3 + (Br z + CHz(COOH)z + BrCH(COOH)z + Br + H) Br (R3) + BrO 3 + HBrO (R2) (Rl) X 3 (RB) PROCESS B + Bro;+ HBrO 2 + H 2BrOz + H2O + + (Ce(III) + BrOz + H + (RS) Ce(IV) + HBrO 2) x 2 (R6) 2Ce(III) +Bro;+ HBrO2 + 3H+ + 2Ce(IV) + HzO + 2HBrO2 2HBrO 2 + HOBr + BrO - 3 + + H . (G) (R4) Initially, the system contains bromate ion, bromide ion, cerium (III), malonic acid) and sulfuric acid in different concentrations. In Process A, bromide ion reduces the bro- mate through a series of two electron reductions . The rate is determined by (R3). When the bromide ion concentration drops below a critical value, Process B begins to produce the radical BrOZ from bromate and bromous acid. This causes the oxidation of 293 Ce(III) and has the net effect given by (G) of generating bromous acid autocatalytically. Bromous acid is consumed by the second order step (R4) to maintain it at a low steadystate concentration. The stoichiometry is given by Z(G) + (R4) followed by the net effect of (Rl) and (R8) to yield the overall reaction of Process B: Bro;+ 4Ce(III) + CH (COOH) 2 +SH++ 4Ce(IV) 2 + BrCH(COOH) + 3H 2O. 2 (B) The Ce(IV) produced by Process Bis reduced by (R9) and (RIO): 6Ce(IV) + CH (COOH) + 2H o + 6Ce(III) + HCOOH + 2CO 2 + 6H+ 2 2 2 (R9) + 4Ce(IV) + BrCH(COOH)z + 2H2O + 4Ce(III) + HCOOH + 2CO2 +SH + Br (RIO) Bromide ion is produced by these reactions and the system swings back to Process A for a repeat of the cycle. With cerium not appearing, the overall Belousov reaction is given by - 3BrO 3 + 3H + + SCH 2 (COOH) 2 + 3BrCH(COOH) 2 + ZHCOOH + 4CO 2 + SH O (C) 2 which is derived from the sum of ½(Process A), S/Z(Process B), and (R9 + RlO) . 294 Heats of Reaction The heats of reaction for each of the major processes and the overall reaction can be calculated from the heats of formation of the individual species. Most of these values can be found in the International Critical Tables. 10 Problems in determining the heat of formation of the radical Bro 2 are avoided by considering only Process Band not the individual steps. Bromous acid is a postulated species; its heat of reaction is obtained from analogy with aqueous c10 . The heat of formation of Ce(IV) was calculated from 2 the heat of formation of Ce(III} and the standard potential of the Ce(IV)/Ce(III) half cell. The calculated heats of reaction are given in Table 1. Table 1 HEATS OF REACTION FROM THE FKN BELOUSOV MECHANISM Process 6H (kcal) A B (R9 + Rl0) C - 31 - 25 -268 -346 If each of these reactions occurs during the course of one cycle of the system, it is apparent that heat will almost exclusively be released during (R9 + Rl0). Any 295 temperature rise should correspond to the reduction of Ce(IV) to Ce(III). This reduction can be visually traced if an indicator, ferroin, is added to the system. One sees the system change sharply from a red low oxidation state to a blue high oxidation state and back to the red. Students in the E 5 lab 11 during studies on the physical properties of the reaction have observed sharp temperature changes corresponding to the sharp red-blue-red transition. Figure 1 depicts the nature of these observations. Figure 1 TEMPERATURE VERSUS TIME FOR THE BELOUSOV REACTION 29.00 _J _____ _j _____ _ 28.80 T 28.60 (°C) 28.40 _____ _r----- ---(reduced) (oxidized) 28.20 28.00 260 280 300 320 TIME (sec) It is difficult from available data to determine if the temperature fluctuation corresponds to the oxidation or the reduction of the indicator and cerium. The temperature changes slowly decrease in size, although the period of each cycle remains constant throughout most of the reaction. 296 Oscillations begin approximately 2 minutes after mixing and cease from 30 to 60 minutes later. For the purpose of a rough comparison between the calculated and experimental heats of reaction for one fluctuation, it is useful to assume the major reactants are completely consumed monotonically by the time oscillations cease. If the length of a period is taken as 20 seconds, and the total time is 1800 seconds, the amount of reactant consumed is equal to the initial concentration multiplied by 20/1800. From Process C for the consumption of bromate ion and malonic acid, the calculated heats of reaction for one oscillation are -16 cal and -43 cal, respectively. Experimentally, for a system with mass 200g, CP equal to unity, and a particular temperature fluctuation of 0.2°c, the heat of reaction for one oscillation equals -40 cal. The assumptions or Process C is obviously incorrect for the consumption of bromate, but seem remarkably good for malonic acid. A more correct and complete analysis of observed and calculated heats of reaction can be undertaken in two ways. Sophisticated, simultaneous measurements of temperature and either color or electric potential would provide the exact concentration, temperature relationships for at least the redox couple. Such measurements would be expected to show a 297 temperature rise corresponding to the reduction of cerium. Fast response thermocouples can be used to trace temperature changes, while standard potentiometric or spectrophotometric methods could be used to follow the redox process. The experimental set-up in either case must facilitate an adiabatic system. This kind of study does not provide direct measurements of other species involved, for example bromate ion, malonic acid, or the postulated bromous acid. While it may be possible to determine the concentrations of the first two components, it is very unlikely that bromous acid can be determined. It would be useful to look at the system from the mechanistic view. If the kinetics of each step were characterized, the reaction could be defined by a system of differential equations whose solutions would provide concentration versus time data for each species. Field, Keros, and Noyes have reported this study for the simplified case of a 5-step model. The numerical solution of the model correctly predicts the observed oscillations of bromide ion and cerium ion. Additional work with this model in the case of a specific experimental situation may yield the consumption and production rates of the components which, in turn will yield a prediction of the heat released as a function of time. The model and to a good approximation the mechanism could then be compared with experimental temperature I 298 fluctuation measurements. Comparing the experimental and calculated exothermic properties of the Belousov system on a small time scale should provide a check on the major processes of the mechanism. Because of the success of the FKN mechanism and model in explaining other observed properties, it is expected that the heat of reaction comparison will provide a positive check on the system. However, in the study of the iodate- iodine system, a thorough experimental analysis of the reaction should act as an important guide in the formulation of a mechanism. Using a similar analysis on the pattern- forming Belousov reaction may or may not produce clues to its complete understanding. • 299 References 1. W.C. Bray,~- Amer. Chem. Soc., i}_ 1262 (1921). 2. Given in 7, B.P. Belousov, Sb. Ref. Radiats. Med. Moscow . , 1958 145 (1959) . -- 3. A.N. Zaikin and A.M. Zhabotinskii, Nature (London), 225 535 (1970). 4. J.H. Woodson and H.A. Liebhafsky, Nature (London), 224 690 (1969). 5. I. Matsuzaki, J.H. Woodson, and H.A . Liebhafsky, Bull. Chem. Soc. Japan, i}_ 3317 (1970). 6. R.M. Noyes, R.J. Field, and E. Keros, J. Amer. Chem. Soc., 94 1394 (1972). - 7. R.J. Field, E. Keros, and R.M. Noyes, J . Amer. Chem. Soc., 94 8649 (1972). - 8. H.G. Busse, Nature (London), 233 137 (1971). 9. R.J. Field and R.M. Noyes, J. Chem. Phys., 60 1877 (1974). - -- 10. E.E . Washburn, Ed.-in - Chief, International Critical Tables, (National Research Council, New York, 1929), V. 11. Data obtained from John Ernest, Teaching Assistant, E 5 Laboratory Research, California Institute of Technology, 1975. Determination of the Solubility of Gases In Liquids at High Pressures The method of measuring concentrations of gases disso l ved in liquids using electron capture gas chromatography may be attractively applied to a variety of solution situations. Measurement of SF 6 concentrations in water may afford a means of tracing underground water systems, characterizing the depths of large reservoirs, or following the paths of deep ocean currents. Comparative measurements between SF 6 concentrations in saturated water injected into underground oil fields and in oil samples taken from the field after water injection may yield approximations of the oil reservoir size. Industrial processes sometimes require fluoro- chemical solubility information. For example, the design of pressurized refrigeration systems must account for the solubilities of refrigerants, frequently fluorochemicals, in lubricants within the system. 1 The common element of each of these applications is the presence of a gas dissolved in a liquid at pressures above atmospheric. Accompanying these and other high pressure gas and liquid processes is a strong interest in the theory of high pressure solution phenomena. Orentlicher and Prausnitz 2 have applied a pressure-corrected Henry's law expression similar to one first derived by Krichevsky and Kasarnovsky to several nonpolar systems. However, there is little information relating the pressure-corrected Henry's law to 301 nonpolar-polar interactions or, more specifically, to hydrogen bonding solvent systems. In view of this fact and be c ause of the interest in high pressure applications, I propose a simple method for applying the electron capture solubility technique to high pressure solubility problems. This report pr oposes the design and use of a high pressure saturation vessel and, further, uses the KrichevskyKasarnovsky equation to estimate data for nonpolar gases dissolved in a polar solvent. These estimates provide a test of suitabili t y of the method to high pressure problems. Predicted solubilities can be easily measured with this technique. Proposed Experimental , Procedure A gas can be quickly dissolved in a liquid at atmospheric pressure in a small syringe. High pressures can be generated in the syringe by simply forcing the plunger against the gas and locking the plunger in position. For accurate measurement of the gas solubility at high pressures, a syringe must be modified as shown in Figure 1 to provide pressure and volume measurements and to assure safe handling. A sample valve and 5.0 µl sample loop, also shown in Figure 1, when back-pressured using a small volume pr e ssure unit, 302 PRESSURE GAUGE HIGH PRESSURE SYRINGE METERING VALVE BACK-PRESSURE PLUNGER 6-PORT SAMPLE VALVE 5.0 µ1 SAMPLE LOOP DILUTION CUBE Figure 1. HIGH PRESSURE SYRINGE AND INJECTION SYSTEM 303 provides a means of transferring a quantitative sample from the saturation syringe to the dilution cube and chromatograph system. The saturation syringe volume, 60 cm 3 , will provide gas partial pressures to approximately 60 atmospheres. Larger syringes can be used for higher pressures as long as the seal 1s maintained and the contents can be safely compressed. It is important to note that a force of 1500 pounds would be needed to move a plunger with a square inch area against 100 atmospheres pressure. Safety requires that the syringe body and plunger be machined from stainless steel. A neoprene 0-ring on the plunger head will provide a good seal. The plunger handle can be machined from a micrometer handle to allow volume calibrations and to allow the plunger head to slide in and out without rotation to maintain the seal. The syringe can be volume calibrated by weighing samples with known densities. Pressures are measured through the plunger with a Matheson stainless steel test gauge, 0 to 1000 p.s.i.g., accurate to ±0.25% of full scale. A small diameter metering valve is used to seal the syringe tip and to load and transfer materials. At the beginning of an experiment, the syringe is loaded with 1.0 cm 3 of temperature equilibrated solvent and 59 cm 3 of temperature equilibrated gas. Next, the contents are slowly compressed to the desired pressure in the temperature batl1. Suitable force can be applied if 304 necessary using a wrench to turn the plunger handle. The syringe is mechanically agitated in the bath to secure rapid saturation of the solvent. Following equilibrium, the syringe is removed from the bath and quickly connected vertically to a gas chromatograph sample valve mounted on the nitrogen inlet line to the dilution cube. Solvent can be drawn through the sample loop by opening the metering valve, allowing pressures to equilibrate among the saturation syringe, sample loop, and the small volume pressure plunger, and then slightly backing the small pressure plunger off until liquid solvent fills the sample loop. At that point, the sample valve is turned, flushing the solvent into the cube. The cube is sealed, the contents are well-mixed, and analysis proceeds according to the manner of the electron capture solubility method. An error analysis indicates that the pressure drop due to the dissolution of the gas will be negligible, while the uncertainty associated with the microliter sample loops will contribute the major portion of the uncertainty of the solubility, approximately ±5%. The estimated error is based on the values obtained from experimental results at atmospheric pressure with the assumption that the sample loops are calibrated as suggested in the attached research report. 305 Predicting High Pressure Solubilities Prausnitz, in Molecular Thermodynamics of Fluid-Phase Equilibria, discusses the calculations of gas fugacities saying, "To increase our knowledge, we require on the one hand new results from theoretical molecular physics and on the other, more accurate experimental data on equilibrium properties of dense mixtures, especially for t~ose containing one or more polar components." Prausnitz's suggestion to study solubilities in polar systems at high pressures can be easily carried out using the method proposed. The gases SF , CF 4 , and NF 3 are 6 detectable by electron capture and have low solubilities and, as yet, thermodynamically anomalous behavior in water. Little work, if any, has been done relating pressure to the properties of these gases. However, prior to any experi- mental work, it would be advantageous and interesting to estimate the solubilities of these fluoromolecules in water at high pressure and to examine their predicted behavior with respect to known behavior at 1 atmosphere. The estimated solubilities, as will be shown, can be easily obtained following the derivation of the KrichevskyKasarnovsky equation given by Prausnitz, while making several assumptions concerning the nonideality of the gases and their behavior in a polar solvent. 306 Henry's law is properly given as (1) where f 2 is the fugacity of the dissolved gas, ~ 2 is the gas fugacity coefficient, y 2 and x are the gas mole frac2 tions in the gas and liquid phases respectively, Pis the pressure of the system, and H2 1 is Henry's constant for gas 2 dissolved in solvent 1. ' Henry's constant is not dependent upon composition, but is effected by temperature and, to a smaller extent, pressure. At high pressures, the effect is appreciable and is given by the rigorous equation (2) where 2 is the partial molar volume of the gas in the liquid phase. Thermodynamically, Henry's constant is u (at constant temperature). (3) Substitution of (3) into (2) gives (4) 307 -00 where u is the partial molar volume of the gas at infinite 2 dilution in the liquid phase. Integrating (4) and assuming the fugacity of the gas is proportional to x at constant 2 temperature and pressure gives a pressure-corrected Henry's law fz ln- = lnH 2 l Xz + ' (S) RT where Henry's constant is evaluated at some reference pressure, As x 2+o, the total pressure becomes Pv, the vapor pressure of the solvent. Therefore, Pr can be taken as Pv. If the solution temperature is well below the critical -oo temperature of the solvent, u 2 can be assumed to be independent of pressure. Then (5) becomes (6) which is the Krichevsky-Kasarnovsky equation. f2 For P much p greater than Pv, plotting lnXz versus RT should yield a -00 straight line with slope u . and intercept H 1 . 2 2 ' mental data can be handled in this manner. Experi- 308 The equation can be rearranged to a form useful for estimating high pressure solubilities: (7) --00 u p I-1;2 lexpRT ' For this result, the fugacity is given by (8) where y 2 can be assumed to be approximately unity. fugacity coefficient ~ 2 The can be calculated from the rigorous equation (9) where Q; 2)T,V,n1 can be evaluated from an appropriate equation of state. The virial equation of state in pressure explicit form p =RT+ RTB + RTC -3-· u ~ u (10) is based on solid theoretical foundations, unlike many empirical relations, and can be used as an approximation to high pressure reality. For this case, it will be truncated after the second virial coefficient for simplicity. 309 Dymond and Smith 4 list experimental B values for SF 6 and CF 4 , while for NF 3 , B can be calculated from van der Waals constants as (11) where band a are calculated from the critical properties of the gas. 5 ' 6 The fugacity coefficient then can be reduced to 2B expu (12) z where u is the molar volume of the vapor phase and z is the compressibility factor. The molar volume can be calculated from the virial equation which gives ( 13) 4BP In the case where RT < -1, the molar volume becomes complex and the truncated virial equation is not applicable at the temperature and pressure of interest. Since Bis usually negative, this can be an important factor at high pressures. Henry's constant is taken from atmospheric pressure experimental data for each gas. 7 , 8 I suggest that u 2 can be -00 assumed to be approximately ub, the molar volume of the gas at its boiling point. Miller and Hildebrand 9 have pointed out that contrary to nonpolar solvent behavior gases 310 dissolved in water are not associated with large increases in partial molar volumes over boiling point molar volumes. Table 1, from data compiled by Miller and Hildebrand illustrates this point. Table 1 MOLAL VOLUMES (cc) OF THE GASES AT THEIR BOILING POINTS COMPARED WITH THEIR PARTIAL MOLAL VOLUMES IN WATER AND IN BENZENE, 25°C Uz at tb Uz in H20 Uz in C6H6 The equation for x 2 Hz Nz CH 4 CzH6 28 <28 35 35 33 58 34 37 57 55 51 73 can now be given as p ZB -z expu (14) and the mole fraction can be estimated for a range of temperatures and pressures. Experimental and estimated solubilities of SF 6 , CF 4 , and NF 3 are shown in Figure 2, plotted as lnx 2 versus lnT. Predicted points are given for pressures 1, 20, and 60 atmospheres over the temperature range 27° to 80°C. In the case of SF 6 no estimates are made above 20 atmospheres because the calculated molar volumes become complex and the 311 :o Figure 2 9 8 Sol ubil ity of Fl uo r omolccu ! c :; .i n l'iatcr 6 4- 2 0 ·a tm. I. . t-· ·---· Mole - - -- r - - - - - - - ~ - - - Fra~ ~ion ~- • - - - - -·-- -- '' !• - t_c-t - i iGi---a i i • 60 •(m -f 1~ .. ' -- ·-· --·-··--· .- ·- - -..; ..-- . . ·-- · -··· ·-- .. --·-·- ·--. • . . . !,I .... . ··-·- -·--·, -.·-.---- _-_-__- _-,___ _ r , _,_ ...J - ·-·- ··- ·. • . . - ·- ·-- ·-' •• • --· • . • -- - --'---- 8 . . . . . . . : ·· ··· ·····--- ---· •- --- - ----- -------- ------ - - - - - -- - -- -------- 5.68 5.72 ln T ( °K) 5. 76 5.80 5.84 5.88 312 truncated virial equation is no longer applicable. For the pressures and temperatures exam i ned here , the solubility r ange is well within the capabilities of electron capture analysis. The amounts of gases dissolved also ind i cate that the pressure drop due to dissolution of a gas will be negligible as predicted. At every pressure , NF 3 is much more soluble than the other gases. The NF 3 molecule possesses a very small dipole moment, µ~O.Zn, 10 which may be the reason for the higher mole fractions. However, it re- mains a rather insoluble gas, like SF 6 and CF 4 , and can be considered to lie on the borderline of nonpolarity. The elec t ron capture solubility method modified in the way proposed here offers a chance to widen its appl ication possibilities and, at the same time, to provide some of the experimental answers to Prausnitz's request . Determination of Trace Amounts of Atmospheric Sulfur Compounds Increased consumption of sulfur-containing fuels have brought about a variety of methods to determine the sulfur pollutants emitted in the combustion process. Sulfur dioxide can be determined by colorimetry 1 , iodometry, polarography, gas chromatography 2 , and x-ray emission spectrometry. 3 Sulfates, in aerosols and particulates, can be measured directly by electron microscopy 4 and by conversion to so 2 , and subsequent analysis by coulometry and flame photometry. 5 The sensitivities for these methods vary from more than 1 ppm down to approximately 1 ppb. This range generally corresponds to levels found in most urban areas. On a world-wide basis, Fisher, Lodge, and co-workers have found trace amounts of so 2 and sulfates in both the tropical and polar regions of the earth. 4 , 6 , 7 They employed a West-Gaeke colorimetric method for so 2 and electron microscopy for sulfate particulates. However, the maximum levels found were only parts per billion at the detection limit, making analysis of lower levels impossible. Situ- ations calling for sub - ppb analyses exist in many non-urban areas, but there is no method currently available to determine ultra-trace levels of these compounds. In this report, I propose such a technique that involves conversion of sulfur dioxide and sulfates to sulfur hexafluoride which can be detected in the parts-per-trillion range. 8 315 The electron capture detector is sensitive to SF from 6 6 3 1 ppm to 10- ppm and is linear from 10- ppm to the detection limit. Quantitative conversion of a sulfur compound to SF 6 may provide an analysis of the sulfur down to the detection limit of SF 6 . A method, similar in principle, has been successfully used to determine beryllium at low levels. 9 In that case, sample filters containing Be were digested and the Be was converted to a trifluoroacetylacetate complex suitable for electron capture detection. Puchelt and co-workers have reported converting metal sulfides in meteorites to SF 6 by the reaction (1) with 96% to 98% yields. 10 The reaction is rapid and com- plete; the products can be easily separated using a combination of liquid nitrogen cold traps and a gas chromatograph Molecular Sieve column. In the Puchelt work, the products were analyzed with a mass spectrometer to determine stable sulfur isotope ratios. Cares reports a method for oxidizing SO 2 in solutions 3 obtained by bubbler samples with H202 to obtain Baso 4 . Thode and co-workers 11 ha~e used a technique for converting sulfates to silver sulfide. One of the sulfides analyzed by Puchelt and co-workers in the SF 6 mass spectrometer method was silver sulfide. It may be possible then, .'316 with suitable analytical care, to convert atmospheric samples to sulfides and then to SF 6 to give low level determinations. Sulfur dioxide can be collected by bubbling air through • • a so 1 ution o f O. 1 M so d"ium tetrac h loromercurate 12 wh.ic h is the first step in the West-Gaeke colorimetric method. Pate and co-workers 13 report the solution to be 100% efficient in collecting so 2 . Alternately, the Pate group determined filters impregnated with a 5% potassium bicarbonate solution to be more that 95% efficient in collecting so 2 . Huygen 14 reported filters impregnated with a solution of 20% KOH and either 10% glycerol or triethanolamine were 95% efficient so 2 retention for several He recovered the absorbed so 2 from the and were stable with respect to weeks when kept dry. filters by washing in a beaker filled with slightly acidic solution. Sulfate aerosols were collected on glass-fiber Gelman circular filters with retention efficiencies of 99.7% in a method outlined ·by Scaringelli and Rehme. 5 Cadle and coworkers 4 collected sulfate particulates on glass microscope slides using a four-stage impactor run at flow rates between 20 and 40 1/min for several hours. Sampling times for the various methods were on the order of one to two hours. Sulfur dioxide absorbed in bubbler solutions or washed from impregnated filters can be oxidized to sulfate ion as described by Cares. The pH of the solution is adjusted to 317 the phenolphthalein end point with 1 M NaOH or 1 M HCl. Drops of 30% H2o 2 oxidize sulfur dioxide to sulfate ion wh i ch can be precipitated with 2.5% Bac32 in 0.1 N HCl. The resulting Baso 4 is filtered, rinsed with water, and air dried. Sulfates obtained by filtration or impaction methods are also prec i pitated, similarly, to Baso 4 . The Baso 4 samples can be reduced in a reduction flask with reflux condenser containing a solution HI, HCl, and H3Po 2 , as given by Thode, et. al. The resulting H2s is swept out with dry nitrogen, washed with distilled water, and absorbed with a solution of cadmium acetate , acetic aci4 and distilled water. The product, CdS, Ag 2s by adding 0 .1 N AgNO 3 . is converted to After coagulation on heating, silver sulfide is filtered through glass wool and washed with concentrated ammonium hydroxide. The solid can be dried in an oven at 120°c. In preparing SF 6 from Ag 2s, the chief experimental difficulty reported by Puchelt and co-workers was handling the extremely reactive fluorinating reagent. Bromine tri- fluoride is a heavy liquid that fumes in moist air. It is toxic and reacts violently with water and organic matter. The Puchelt workers recommend dispensing the reagent over a metal tray filled with dry sand in a well ventilated fume hood. One to two ml of the reagent is dispensed to the bottom 318 of a cold reaction vessel previously flushed with dry, precooled nitrogen and cooled with dry ice. The silver sulfide samples are wrapped .in packets of thin aluminum foil and dropped onto the frozen reagent. Each reaction vessel is placed in a furnace overnight at 200°C with the sealing flange cooled with tap water. A modified reaction and isolation system is shown in Figure 1. Changes to the Puchelt system include the addition of a manifold of five reaction vessels A, a nitrogen flow system for cleaning the apparatus, and a sampling system for coupling to the gas chromatograph. The product from one of the reaction vessels is allowed to stand in contact with sodium hydroxide pellets contained in Monel metal scrubbing traps, B. Vessels A and Bare cooled with dry ice; traps C and D, and sample tube E are cooled with liquid nitrogen. Sodium hydroxide removes the acid constituents HF, HBr, and Br 2 . Then the SF 6 is purified by trap to trap sublimation and finally frozen out in the sample tube E. Total SF 6 produced is measured using an electron capture gas chromatograph fitted with an 8' X 1/8" stainless steel column of 80-100 Mesh SA Molecular Sieve at 120°C. The chromatographic system can be calibrated by the exponential dilution method. Sample tube E acts as a sample loop connected to a gas chromatograph sample valve. After the SF 6 has been sealed in E, the valves between E and the gc 319 Figure 1 SF 6 CONVERSION AND PURIFICATION SYSTEM MONEL METAL ·--·---·- -1 B E C D A 320 sample valve are opened, and SF 6 in Eis flushed into the chromatograph by turning the gc sample valve. Additional samples can be analyzed in the same manner after the isolation system has been cleaned by vacuum and nitrogen flow. Successful utilization of the method will require extreme care in handling samples and reagents, and in preparing the necessary filters, glassware, and tools. Standard solu- tions of soluble sulfates can be used to study recovery efficiencies and, thus, to locate contamination and loss points in the procedure. sampling procedures for It is also possible to test so 2 and its conversion to sulfates by preparing standard so 2 atmospheres in the manner used by Huygen for testing filters. The gc portion of the method is not subject to interferences beyond compounds that elute with SF . In such 6 cases, the compounds can usually be separated by changing operating conditions, i.e. temperature and flow rate. Materials named as interferences in the wet-chemistry conversion procedures were interferences of analysis and not interferences due to competition with the sulfur molecules in chemical reactions. Errors in the sampling and conversion steps are difficult to predict, but can be determined in loss and recovery studies. Electron capture detection of SF 6 is reproducible to within 5% or less. It may become important 321 to prepare a standard SF 6 -N 2 cylinder in order to periodically check the detection system for drift in sensitivity. This technique will be limited more from sampling and conversion procedures than in final analysis of the SF 6 produced. Analysis of very low levels, therefore, must be approached with a great deal of thoroughness and care in the design and practice of the actual chemical methods. Even so, similar analyses of beryllium suggest one may expect successful results in the application of this procedure. It is not proposed as a replacement of faster, more economical methods, but rather as an important path for probing beyond current limits of detection. As a base laboratory practice, the technique can be teamed with far-ranging, remote field studies on the source and sink relationships of atmospheric sulfur compounds. 322 References 1. J.H. Seinfeld, Air Pollution: Physical and Chemical Fundamentals, (McGraw-Hill, Inc., New York,1975), p. 212. 2. A.P. Altshuller, Anal. Chem.,!!._ lR (1969). 3. J.W. Cares, Am. Ind. !::!Y_g_. Assoc. 4. R.D. Cadle, W.H. Fischer, E.R. Frank, and J.P. Lodge, Jr., I· Atm. Sci., ll 100 (1968). 5. F.P. Scaringelli and K.A. Rehme, Anal. Chem.,!!._ 707 (1969). 6. J.P. Lodge, Jr., and J.B. Pate, Science, 153 408 (1966). 7. W.H. Fischer, J.P. Lodge, Jr., A.F. Wartburg, and J.B. Pate, Env. Sci. Tech., I 464 (1968). 8. P.J. Drivas and F.H. Shair, Atm. Env., 8 475 (1974). 9. W.D. Ross and R.E. Sievers, Env. Sci. Tech., 6 155 (1972). 10. H. Puchelt, B.R. Sabels, and T.C. Hoering, Geochim. Cosmochim. Acta.,~ 625 (1971). 11. H.G. Thode, J. Monster, and H.B. Dunford, Geochim. Cosmochim . Acta.,~ 159 (1961). 12. N.M. Trieff, H.C. Wohlers, J.A. O'Malley, and H. Newstein, Air Poll. Contr. Assoc.~-,~ 329 (1968). 13. J.B. Pate, J.P. Lodge, Jr., and M.P. Neary, Anal. Chim. Acta.,~ 341 (1962). 14. C. Huygen, Anal. Chim. Acta., 28 349 (1963). I·,~ 386 (1968). 323 Experi mental Characterization of Accidental Spills of Liquified Natural Gas As a result of the imbalance between growing national energy needs and dwindling energy supplies, large increases in the amount of natural gas imported into the United States are anticipated in the immediate future. In order to incorporate this increased volume of gas into the national energy system, several large receiving terminals have been proposed for construction. For economic reasons, the construction sites for these facilities are generally in or very near urban centers. The enormous volumes of liquified natural gas (LNG) (up to 165,000 m3 per tanker and 88,000 m3 per storage tank) and the hazards associated with the transportation and storage of this fuel have caused intense technical and political debate over the locations of proposed storage complexes . Unfortunately, little experimental data or theoretical agreement exists concerning the probability and consequences of an accidental spill of LNG. In particular, there is sharp disagreement over the impact of a large spill of LNG caused by the collision of an LNG tanker with another large ship offshore of an LNG terminal. It is visualized, in a catastrophic scenario, that a collision would produce a cold pool of LNG which would vaporize and disperse towards shore as a dense, flammable cloud. The possibility exists that the cloud would reach an urban area before dispersing below its lower flammability limit (LFL = 5% natural gas in air); the multitude of ignition sources in an urban area virtually guarantees ignition of such a cloud . 324 Much of the controversy over the safety of LNG centers around estimations of the amount of dispersion which takes place in a cold cloud of gas traveling over the ocean. Estimates of the impact of an LNG spill range from no ignition and, hence, no deaths, to certain ignition, and 70,000 deaths; these predictions were for identical conditions by Science Applications, Inc. (1975) and Socio-Economic Systems, Inc. (1977), respectively, A measure of the disagreement which exists among different impact analyses is shown in Table l. These widely different estimations are the result of differences in key assumptions in models of the vaporization and dispersion of LNG. For example, Fay (1973) assumes that a sheet of ice will form beneath the spreading pool of cold liquid, thus decreasing the heat transfer rate and increasing the size of the pool. Other sources, following recent experimental observations, assume that no ice forms. The model by Hoult (1975) for the Federal Power Commission (FPC) predicts that the vaporized cloud achieves positive buoyancy shortly downwind of the spill. Models by the Coast Guard (1977), Fay(l973), and Burgess (1972) for the Bureau of Mines assume that the cloud is negatively or neutrally buoyant and much less susceptible to mixing. Estimates by Science Applications, Inc. (1975) are based upon the atmospheric diffusion equation; other models use the Gaussian plume expressions and employ dispersion parameters from Pasquill-Gifford as given by Turner (1970). Otterman (1975) reviewed a number of experimental studies of the spreading and vaporization of spilled LNG. Feldbauer et al. (1972) 325 TABLE l COMPARISON OF LNG VAPOR CLOUD DISPERS ION MODELS, COMPARISON OF DISTANCE TO END OF HAZARD ZONE AND UNDERLYING ASSUMPTIONS {STABLE WEATHER CONDITIONS -5 MPH WIND)* Spill Size Cubic Meters Coast Guard {Fay) 100 1.27 mi. 2.5 mi . MIT Bu. Mines ( Burgess) FPC {Hoult) 2.4 mi. 0.114 mi. SAi 1,000 3.46 mi 10.0 mi . 7.6 mi. 0.437 mi. 25,000 14.4 mi . 57.0 mi. 38.0 mi. 1.89 mi. approx. 1.20 mi . 100,000 26.2 mi . 127 .0 mi. 76.0 mi . 3.72 mi. approx. 1.27 mi . 5% 0 .1% 0 .1% 5% 5% Effect of Increased Wind Speed LFL decreases LFL decreases LFL decreases LFL decreases LFL increases Effect of Increased Atmospheric Stability LFL increases LFL increases LFL increases Sensitivity of Cloud Size to Spill Size moderate very large large Assumptions : Cloud LFL Concentra ti on* Cloud Turbulence minor effect on distance to LFL small very sma ll high *Cloud l.FL concen tr ation<. fo, Coas t Guard, FPC, and SAi assume flamm ,1 hlc element~ with perimeter co ncentr ,11ion Kr, 1dient s of 5% LF l.. The MIT a nd Bu . Min es models ass ume that measurement of den sit y a t a give n point over time mu s t be below 0. 1% to assure that no discrete cloud remai ns which has a great enough co ncen tr a ti o n to ignite. *From Socio-Economi c Systems, Inc. (1977) 326 conducted spill tests on pools of water and found that while ice was formed during spills in confined tanks of water, no ice was observed during spills on open water. These workers also reported that the vapor cloud spread radially and did not achieve positive buoyancy. Data from work by Burgess et al. (1972) and Boyle and Kneebone (1973) yielded an average evaporation rate of LNG equal to 0.2 kg/m 2-sec. Burgess et al. also found that turbulence at the interface of the LNG and water prevented ice formation. May et al. (1973) conducted spills of as much as 10 m3 of LNG at sea and found that the vapor flow rate was proportional to wind speed as well as spill size. They reported that the dispersion of the vapor cloud was less than that predicted with the Gaussian plume model. Currently, studies are being conducted at the China Lake Naval Weapons Center, China Lake, California, under the sponsorship of the U.S. Coast Guard (L.A. Times, 1977). These studies are aimed at characterizing the ignition properties of an LNG vapor cloud; the largest spill conducted thus far involved 5.7 m3 of LNG. Tests are also being performed with liquified propane gas (LPG) since this material is transported as a cold liquid, too (b.p. -41°C). Recent results of experimental spills of LPG indicate that it will burn rapidly, but will not detonate in the open. The immediate requirement for additional sources of natural gas, the vast differences observed in the theoretical treatment of a liquified gas spill, and the meager amounts of experimental data forcefully emphasize the need for extensive studies of the vaporization 327 and dispersion of LNG. In this paper, a safe, experimental method is proposed for the determination of the evaporation and dispersion rates of cold liquified gases under measured meteorological and topographical conditions. Specifically, a cold, liquified fluorocarbon, such as CHF 3, is suggested as an inert, nontoxic, and nonflammable tracer of a liquified gas spill; in conjunction, atmospheric tracers, such as SF 6 and CBrF 3, are proposed to simulate the dispersion of cold gas clouds. Because of the unreactive nature of halogenated compounds, the dangers associated with experimentation involving LNG or LPG are reduced to those associated with the handling of cryogenic materialso The use of inert, nonflammable tracers will allow realistic tests to be conducted in regions of interest without endangering nearby populated areas. It is expected that the safety aspect of the proposed method will permit the quantity and variety of experiments needed in this field to be performed by more organizations than just those equipped to do combustion and detonation research. The application of tracer techniques to the study of an LNG spill introduces a great deal of flexibility into the design of LNG experiments. For example, the simultaneous releases of a liquified tracer and a cold gaseous tracer will yield information associated with the relative effects of the vaporization and dispersion processes upon downwind concentrations. As in atmospheric tracer studies (Lamb, 1977), a large number of air samples can be collected and then rapidly analyzed by electron capture gas chromatography to 328 provide fast feedback of detailed experimental datao Experience has shown that rapid data reduction in the field permits knowledgeable improvisations in the design of a field study in order to optimize the usefulness of experimental results. Methane, the major constituent of LNG, has a liquid density equal to 0.47 g/cm 3 at -160°C. One of the few fluorocarbons with a similar denisty (0.67 g/cm 3) is fluoroform (CHF 3, Freon 23) which freezes at -155°C and boils at -82°C. Like other freons, fluoroform is inert, nontoxic, and nonflammable (du Pont, 1969). Clemons and Altshuller (1966) tested the response of the electron capture detector to fluoroform using a Baymal column for separation. They found that CHF3 was approximately 5 orders of magnitude less sensitive to electron capture detection than SF 6. Since the concentrations of greatest concern are those in the range of flammability, 5% to 15%, this lack of sensitivity will not limit the usefulness of fluoroform as a tracer of an LNG spill . It is worthwhile to describe a small-scale experiment which might be performed in the initial stages of an experimental program. The following test could be conducted on land to simulate a spill from a pipeline or storage tank; it could also be performed offshore of a proposed terminal location to simulate a spill from an LNG tanker. It is expected that techniques and pertinent measurements will be similar for both water and land studies. This particular experiment uses SF 6 as a gas at ambient temperatures, CBrF 3 as a gas at low temperatures, and CHF 3 as a cold 329 liquid. All releases are conducted 6 km offshore during a period of steady onshore air flow. The SF 6 tracer is released initially as a quasi-instantaneous puff; a short time later, the CBrF 3 tracer is similarly released, although at much lower temperatures. Finally, the cold, liquid CHF 3 tracer is pumped rapidly onto the surface of the water. During the course of these releases, air samples are collected at various distances downwind of the release. Because of the low temperatures associated with the CBrF 3 and CHF 3 clouds, samples collected at short distances downwind must be obtained automatically with suitable equipment. Evacuated stainless steel cylinders each fitted with a controlled orafice may be of use under these conditions. Further downwind where temperatures are near ambient, samples are collected in plastic syringes. Vertical dispersion data can be obtained by placing samplers at different heights in towers deployed along the shore. Data by May et al. (1973) indicate that vertical dispersion is limited over short downwind distances. All samples are analyzed with electron capture gas chromatography immediately following the test. In order to correlate dispersion data with meteorological data, a continuously recording anemometer is located at the release site. A series of recording thermometers located along the transport path provide data for relating the buoyancy of the plume to the dispersion. An attractive means of monitoring both the temperature and the dispersion of the cold gas clouds is infrared photography. An infrared photograph may show the outline and the intensity of the 330 cold vapor cloud. The thermal response of infrared film can be calibrated with air temperature measurements. Although advanced infrared technology is available as a result of geological and environmental mapping programs, infrared film can be purchased inexpensively and used in a 35 mm camera without special equipment. The use of infrared photography in LNG experiments as proposed herein has evidently not been reported previously. The application of infrared techniques to tracer experiments could produce a tested method for use as a monitoring procedure in the event of an actual LNG incident. In summary, the application of SF 6 and CBrF 3 as gaseous tracers and CHF 3 as a liquified gaseous tracer circumvents the fire hazards associated with LNG experiments and provides a large amount of freedom in the type of experiments conducted. Further flexibility is provided by tracer study techniques which yield large amounts of detailed data in a short period of time. Infrared photography of the cold vapor clouds may serve as a quantitative method for measuring temperature and dispersion. 332 The Flux of Naturally Emitted Hydrocarbons into High Pressure Systems over the United States The recognition of unacceptably high ozone levels as a regional phenomenon in the midwestern and eastern United States has raised questions about the effectiveness of urban hydrocarbon control strategies. Ripperton et alo (1976) conducted field studies which showed that as high pressure systems move across the Uo S., uniformly high ozone concentrations are found over areas as large as 155,000 km 2. These field study results indicated that ozone levels increased as the air masses were transported from rural areas into urban areas. Maximum ozone levels in one such high pressure system increased from approximately 50 ppb in Montana to approximately 80 ppb in Iowa and 100 ppb in Pennsylvaniao Husar et al o (1976) used visibility data from the U. So Weather Bureau observation network to develop visibility contour maps for periods during the summer of 1975. They identified six large stagnant air masses with an average residence time of 8 days over various multi-state regions which became increasingly hazy during transport. Husar et al. found that these hazy masses were accompanied by high midday ozone levels (as measured in St. Louis). They concluded their work by suggesting that major questions associated with these hazy air masses pertained to their anthropogenic and natural gaseous precursors, to the environmental conditions in which the haze develops, and to the fate of these hazy air masses. The purpose of this work is to calculate the flux of naturally emitted hydrocarbons, mainly terpenes, into high pressure systems as they travel from Idaho and Montana across the Great Plains into the 333 populated areas east of the Mississippi River. Determination of the composition of air crossing into urban areas is necessary for specification of boundary conditions in air quality models. It may occur that for regional air quality models, boundary conditions will be given as a function of the trajectory of an arriving air mass. Because of the high reactivity of terpenes to hydroxyl radicals (Darnall et al., 1976), the flux of these cyclic olefins in rural areas may be an important factor in the photochemistry of the air in high pressure systems. Went (1960) first suggested that terpenes emitted from plants and trees could be important in the formation of atmospheric aerosolso He reported that Haagen-Smit measured the release of organic matter from sagebrush and found it to be approximately 50 kg/day-km 2o Noting that coniferous forests produced similar amounts, Went assumed this was true of all vegetation and arrived at a worldwide emission rate of 1J5 · 108 metric tons/year. Rasmussen and Went (1965) tested for the presence of terpenes in a number of rural areas; they found that the concentrations of total volatile carbons reached levels on the order of 15 ppb. They identified isoprene, a-pinene, S-pinene, and myrcene as rurally produced hydrocarbons. Their results indicated that concentrations decreased by a factor of ten from summer (20 ppb) to winter (2 ppb)o Rasmussen and Went also reported that levels decreased during a rainy period , but increased during harvest periods and in the autumn with the loss of leaves from trees. Based upon these measurements, Rasmussen and Went estimated that the worldwide production rate equaled 4.38·10 8 metric tons/year. 334 More recently, Rasmussen (1972) reported measurements of the emission rates of a-pinene, 6-pinene, limonene, and isoprene from different types of trees. The rate of emission varied with plant species, leaf temperature, maturity of the foliage, and integrity of the cell structureo Emission of isoprene was also dependent upon light; no emission occurred in the dark. emitted in the dark. The other terpenes were Emission rates of a-pinene from coniferous trees ranged from 0.4 ppb/min/gm at l7°C to 2.0 ppb/min/gm at 30-32°C inside al liter bell jar. Emission rates of isoprene from hardwood trees varied from 0.04 ppb/min/cm 2 to 2.4 ppb/min/cm 2 with changes in temperature and light intensity. According to Rasmussen, Ripperton et al. (1967) assumed terpenes were the major natural sink for tropospheric ozone; this assumption increased estimated emission rates by a factor of ten. Graedel and Allara (1976) performed a similar analysis and obtained similar results. The estimated emission rates of terpenes from each of the preceding analyses are summarized in Table 1. In view of the sparse data and broad assumptions, the spread in the estimated values is not unexpected. Note that the emission rate of reactive hydrocarbons estimated for the Los Angeles Basin (Reynolds and Seinfeld, 1975) is approximately equal to the values given by Ripperton et al. (1967) and Graedel and Allara (1976), and are no more than a factor of 16 greater than estimates by Rasmussen and Went (1965). High pressure systems regularly sweep across the United States from west to east. A qualitative inspection of the daily weather maps published by the U.S. Weather Bureau shows that from April to October 335 TABLE l ESTIMATES OF TERPENE EMISSION RATES Source World-wide Emission Rate metrtc tons/ year l.75·10 8 Daily Emission Rate/ lhit Area kq/km 2-day Remarks 17.5 50.0 4.38·108 16.0 100 days/yr measured from sagebrush 270 days/yr estimated from ambient levels 4 .38 • l o9* Ripperton et al . ( 1967) 8 Rasmussen 1. 75· 10 (1972) 160. 2-10 times previous estimates 9.7 180 days/yr measured emission rates Graedel and 2.87·10 9* A11 ara (1976) 160. Went ( 1960) Rasmussen and Went (1965) Total Reactive Hydrocarbons in the L..A. Basin Reynolds & 3. 69 · 10 5tons/yr 158 kg/km 2-day 365 days/yr Seinfeld (1975) area= 80 km by 80 km *Estimate based upon the assumption that terpenes are the major natural atmospheric sink for tropospheric ozone. 336 during 1975 and 1976, as many as 40 well-defined air masses were transported from the Pacific Northwest across the U. s .. A typical example occurred between September 11, 1975, and September 15, 1975, as shown in Figures 1-5. A high pressure system pushed into Montana on September 11, spread over the Great Plains on September 12, and slowly edged to New York by September 15. The average residence time of systems observed during the summers of 1975 and 1976 was 5 days. In several instances, similar systems entered the Uo S. from the Gulf of Mexico and then traveled north to the Atlantic Coasta A review of land use from the National Atlas of the Ua S. (1970) indicates that high pressure systems generally are carried over coniferous forests on the east slopes of the Rocky Mountains, the sagebrush deserts and grasslands of the western Great Plains, and the wheat and corn fields of the Central States. The calculation of rural hydrocarbon flux was originally envisioned in terms of a model which describes the flux as a function of the trajectory of an air mass, the ground cover, and the time of year. The high pressure system might be modeled as a well-mixed chamber with a source term which varies as a function of the location of the chamber and time of year. The effects of precipitation and agricultural activities could be included. If sufficiently detailed data were available, source terms could be broken down by specific terpeneso However, as Rasmussen (1972) .noted, there is currently insufficient data to allow different terpene emission rates to be assigned to different types of vegetation. It is necessary therefore, ~j s .. i l -' t . 10' _ M ,\P ,·. t ,\) Ht I{ L .'., T !oo=-s Sct . , . . ~I~~-: . . · ,ooo 4 " _, (: ~-,~;. .. .. / • ,,,, , ,• ' ' \ •0 . -• ~., . , , • I ~ ~o ••,;• •' 1,-,J, // ' / ~·•0·,; , I .. , ,, , -,., ,,..,~ ~", ':;.•✓ ~~ z· o• •, c '~ , / ',"! ' i)'~it~~~'" ''f ;"{; ,_ /l)J,i ~ ~~--,,"' ),.,'- ,,, .,.. ......,; ,> _ .,, ~,, \ ~"~y o, _.,, ...,,.•.;- '>- G, .,ea<>-.-~- -.,,,...,/ c."<t-1" ,., \to. •' .J <;; '"'"''"'" •= ' \ (. "l""lc," W -- - - __ J \.. - - ~ J,,o✓ = S '-" '.. :!.- ' ..- ,.,✓ ~ ' - " - I ' Figure l if r • Sept I, \. - - - , J'T'I.~ </fj,~ / ( ...... / = 1 I ,,:C', I " , , ,.;e1~"\':i ""-- ' • I' ,.._o't;ol A r·:i ' 7'l; '1< .. ",.,wr ' -,,, ~ _.V I <>,;."' \ ,...'1••, s•••• ,'ff'/_,-~ : 'l,_, ' - - \ _,/ / .t ' . , , , . •• ' :-o'!.o \ rt;\ .. ,,, 1t ·•' •"1.,:/ ._ -~ ~•~- -· • o·i... ~' ,,. •• --✓/L,J,~. 0 ~ _,-/ ,/ "" ,)•~- ·"' \ \ (j / A ~ • - _i_D" /....I. ' - \_ _ , J ' ~ -- \ I I I ' I II !(' b - ___.;:1~;,-·~, ~ \ o \ " em er 11, 1975 ~;"-~ ·~ rl,-.,G, ~~;;,,r; , .. ,- , . ~ · \ * ~ -~~ 0 ' ,_;.a;••,'-~,,._'>- • .;;~ ,-t~.~....@:..!' 1.1..i:,..• '" '<<,¥ C/J]•~' • ~...,'IP;,) "°' ,.. ~ -"'~@ [ _,.. . .;-f',';;1 .,,.,~·~ t > / ... - ---·"·( •• ,,. ••t::' -s,p>?t , '!\\• ) li , , I ---~-\,- - - +i -- 'c. '!· • /'9)..i '""'O \ ~ , , , , •.,'l~ , 1 ••-a • •~~,, e,~ "':;,.;- - ,>· .,,-,, \ , '1,,¢..-.r , ~.1<!' ' .., ·" - f ,~ 1'S1' ~-.., ,,,... 10 •,-.-~-'. ~ -"~'""" •. __:, ""Yr, c\ 10,.; n 1012 . • - - -''-"'];>I , 4S -t2:i'11 '""' I ,t ,'t!o.~~q-,•'/ I'' >J'.'~ ... 7 ~;'\:/~)'~•'I ,,o •,.,'\ ~ ; "1'., J _' , • -- - ,<-,,"' °'V' " • : ; . 0 , . _ i V~ ' , -'.' 0 ~• ·' lp.~1 , ,:;,~),o~•-"{\,1-Cg;,,-.~ " ' , " .~ --... .~l ,,l_,, ·-~'·'•, -,~.-· ,., ·,J •$ ,~-• ',. "~,··· .p :l'Z 'f~... . I ' ' ' \, ~--.... .. ' ' i•,l ~·~ 4' '\ • I ,/:,.>1.!b• .L---<Y~',- -~{-"~; ,,.,-,"• ·> -,.k , ,2 •""' -C , •• ,, • ), ·'r' • , ,• .. ';'/'.•>1• ,"1 ,' :Mf..,~' • ' •· ··~ .,~ g" ,·' •,9{: • Yi:,, A, .t,,. - ,"/4,~;~ ;;:., ~>-;t~'.,,.. ''" /f I ,o•>t "'1-.( -"" ,, "'t., " ·" ~ · "'•<.;,~i.;>',, ,.~<r;t~-~~• , ,,. ..""' ,;i,·~ ,R, '-eP. ,w•'' 6': . , : - , " " ~f\.~,1 "l,,~ ' " s5,;'o' • ' '-"~-•" 'Yo' - --- 1008- - - - ..,,,• .,, -"!;._•78 , , '-• 't, - __, '., ~ ¥ ? . . • • ~ """,;•\' 1 rrO '0°9B _, ' I ~ • / '> ,I.;'• •"'''"' ·t: '.'0 6 • t·"!,~ U."ii?~ " ~ ~ 3 •z;.I'\~"!>.£ .,~~ ~r7'1'k~ -~· , 'c.'- 1 •t,; ,,,~ ~-~ ) - ~ t-, ,••, . _ _ ,, 4 ,.," ~ -· •· •._,o,1 1'.\>,1:l , ,, );' ,_ > ~, _,..._ 'cp. •;;}i ,, 5 ,' 't' · -, "-"Ii· .,,, ' -r-_ .. , ..... ,..,1 ,..,. '."'-' i'' • •i -. "'' • '- - '-"@ 0 '• I ' ' ( cl,.. - - . _ _ 4'] 2 ¾ ~f.. , ,. ' I ' ., _,-,. . O c\:j~ , • , ~ • -- • t.ooe'· 1012 '.'7. • 'a \, ,,.,.•• · '--·""'. 1004 -::~""'·~ ·. r~'.. '- ilF'" ~\~~:~~~s~~:••• /,, ~r~~~~ _.";:r. ,3:-,~~, .••• ~Ji(t ~ ',;-•>='<'• ~ ,~~• , " ~ f.,,3ii.;:..r.,.; --r i.•t1., ~✓ ~ ~ . , ✓ ,,. I - ' ... - ,, -c".,, ,,.~• ..••·, >..a<>, .-. - '- .,;;;...·"'•}3.- .,,,•',.,, ::.,,, s't,bs) ~ -- ~C: , ••._,,, ,,1,,,~ ...1,• • :;s,n(o,r.,~ "~", ~ ~ ••--~l~{ ·~.1.,:;!!:?. ..-,,'< ., .. , , l•.• ,• "' • " ~~ ,:'.') / , -~ - \........ - ~~-~ -r~,~-.)f~;~ ~t-:~'C)' Af ·' ,·, ,,,-~ /''" -/4'-___ -' ,~ ~ •-~" ~}~:". •rie,ft,.;,;:f,.J/Ji:S.:~~(/i;J{.:: ~~ £ -"~ ~ ~ o ~ ~~ ✓•'•""" ~ I -i .. (', .. , 1- - • .,_,, " '~'~ l,"'>" 51m .l - -. c ~•,•· ,~"\ - - - - ''=<>"" .,. ([,· ' ·L~"'o'! ' \ " ]' \~:~ - ;T':'"'~1~!c___:,'~":'i.y"'' _, · I ' i) ' ..-·, 1/E--~i'- ·,~-~ ..;. '-,. / ' , ,r,1 _I' , -!£~ •, ~ ~ Hv - - - Hr - - ' "" ..I 'oo• LOW '- 0 I _ - \ ~-- • -·· -··· . ·--~· --·- . ,.!}, •"• 100 ~ 1 2 ,?:_"-r ' '" '" ''!?c',• .,._~ ; ; ~< '" [, . (il ""<' \ 100 - ,. ____ - --,:-·0 1000 "' - ~ ......_ ' \\ \ 'I · - - "' I ~ . ,32•1 \i •• 'I ") ,~'" - , a,• I ~, l ' .,"''' ... -- - - , ~~ • )2 • '"· "11 ' ;o_,,o "·' 3i 1's I ( J ' :I ' )/ ' - '!) - ~ ~ • •" < -- •· ; •>f.,,·•'. ;,."'; ., ,••.,_.. •.· / ·>. ,,_ \ ~ -..:is .... , ) ' ' o ,;oo, - , 1, , .,. , .. 4 i' ' j, ' I ,(,.,'•l'••· - - I\," ., .., 1 • , '' .-,::':ii'Q_,..o\- ,~'P" , 4s,9,o- ~ \ z.o J s1 0 ."',• 1 . <s .. ; .~\2 ' ~;o••~ - - - , ·,C ' 1 -;~-- - 4G@,,;., ,, - \ .,. - / ' " -,~ . " ~ I , ,j > ~1-o- " !'I' 54 h•,,,f •.. ~i ""')' 8«· _ 70125 ~:~ - _ 4a,o ,, 'i?,~- r----.. ~-;;~"'/2 ,' _,31 . .,_,,,,.. 55 ✓ "r 15 •IN~ • ~:• / ' - .. s , , , ,.. . ... -1(j' · .•" •" f(,t, ,: "" / - - ;- - , • ,. ' ~~ " 9 ·,tHfR I 18 .;< \ ' , ! <o . .. •fr;;,._ ') '' ~·•civ• \ ·~· I · I /; ~\ """• " ,j,. \ ' '' '"'.' SJ; •/ r,· \ 0 ~ " " ' . . . '-.__,/ ;; ~, 1 ~;,(' _ I... ,- {L~~ -,;., ,,.\ /1/'-',, ~ /(: j ::~. ' / ,._<:5 II I I I , 'I I ,@.t,1 ' _,Jl' / ·.,. ~ ,~ - - ··-- - _ "tO',• 28- - - •-102~ --- _ 1 - ---- _ iJo..w16-t.°8~ -= - 32• 1 I w w -...J ,. ,,;, ', .' ~ .. , ,I 1· • I >~1 I ~ , I fQ3 I , •-1 aro <:., ,0 - -,8 1 , ' , 2 __ 1 I I , , 0 ' I S7 ~ o,....,, '" ~< L 1½ .. • 49f ,'3t5..fY>. <;u • . ~ 4. r- \ tO lV 10 7$ 03, . ~ e,, 80(+2 ✓ 1 <. " ) I \ _ •' , , .-o • . I _3 \-- _.H _;~ I'.') ' 1 I .. V, ~· ? --'-. ,a Go. - c , , 3 \ I 0 -t,'i.'- '.1..~ ' , " 1 y~ 0 - _, I 0 , "'o~, ""1 ,...5, ;2,b'=- 1 ,.i ~ 43 I - ~ yj~ -- • - • ; ~~ 4 ~ -, ~ , ,,. L ~...,.. • J AS ' f: ' I<-'- / . E==j.::."'==-c--=-::::;;::~~----- Er::=.f:-·=.\.- =s:·---"-=--~ ..... . ..... " .,. \ 201 \ . ,.t. 14• 70-;'"""°c . •, ...... .. " \ 1012 ' -c '- ." . .. , :'\ I / ., ...,-~ >': .\sio A ·: :-::.>,,,_ % ,.;-;; ~t.~ v~'\.:. 0 ,.!l.•~,, --;\4,._ ,~r .....le. • .~. lf\~-,u , ~"'!,~\ ~,'.,'':!?. .,.,,~:?, 4 ' .\"\\ .. ,-., •~ ,,tt; ,-r; , r;:l:) "f;'.7>.._ '1.::<,<1' •• ' ,,.. ,_; . i,., ~ {~ - -,o ,<,\ ~ ,~~ ~ -..o '"'1, '<,':;.f'"> ',i) ~<, _.__ r.,. - '~\J /J.~oo--.::,,:_'b "'o"o,~ ~ , ~ 1:, ,,...s 'I'~/ 7?... 11 ( <: ' 0 \ <,~ s ':J..'1..C, \ • \ ~':> ' \ \ ' \ \ \ ~~ -l e • / 4) r. ..,,~ ( .. \ .1o'° , -:,__ ""li ,0'2~/ _;,.,, ""\":> ~ - ·.,~ ..,.u-.<, e J ,.--,~~ ,.'lv , •'~".,~.,o/ ""\6,- , _::> • ~""\~\f.o'o/ ') • • September 12, 1975 -··-- • .,,,...,• ,oJ!l J Gt~ .,..7'.'. ,.? ,~ _ - ~,=.@"'~ -.o ' ~.,_.,,,_ ~ -,i~."":J c.,c / ~ 11~::e:>s.,.-'· ',. 0~?) ~~ ...., °'),8~ _, ,-...._,...,.,,, - - , • ... ' Figure 2. 7 ~ ~"" ' \ ,· ..., '. .,,o1-- -,,, \S~ ""\ Q) \ ....J ....... ~~9 1 ''\'\, ,fl,./ , ;:_';. \ "y<>-,, ~ :.--~ 6) ,-ol, ,;'?:- f ~~. _, ~~ :;;_'"°':/ ";; ,...::> ,~i'✓ -'o ,"ft;, --1~f-'.., ~'\~ ,.., "l _o'4 "::> ~'/fs.-,0' _, '~~A/A·' l - - - , , fu _ 73 ,,_, +~ 74@~10/ 1'2. )\ 4 73 Q +O '7 "''"'"~ ~ r - - ,., ,r t . . ~ ~(. b':o /,,o~ .., ~ o....P~ ,...'?, 0 ~ ·~ ~ ~ ~ : t ; ~"- ,.,;,, ,J>,)"p." .,,...•~-" -;. .. ..~,- .,,,;~· ,-c:-..,J\,.eo'!-'·,~.-i~~ " t~J.._,~,,'✓ "'.r,.~- • •• • - :, ~ \ cf',_, ""~ " / 101, , 10 J2 l ·' ~'\'. ·. ""="~~<>-< l • -~ .,f;a~,,_.,r ~f.,•i/J;~,r;,~_:/ -~-,_;t/• J, heoi I ':f:i...;'o"?~-~~/' / . .,•• {,, "';};..., \o o •<1J1 A' \A>!>fi<.. ; ~ ' i ...,\A( a 1 .- ~fo~~ I <, ,fJ> ~1' ~~:0· • -., J), \ f>, <~ ' ; l ' ~ ~~ ~ f:P ~~ o "''\ ~-.. \/).~ ' (,, ~ ~ / / -~ ;' •J _ _"'1...',cfl OI I '!,'1 Q'2, ' 4~ ; r; 1• • .,.,·; 7 -z-;-.-a,.'\o e. o"' , 1.> -, • ... • • , "Ye •. _ -· ~.. . ·,, / >. ~ - , 43 ' '- ,..~..,,,I':""{, "'I ', .,~~~-~( ~' / • ..... ~,"f'/"fo~ ' ,'IA ~ , ; .~of'> 1, J9,.~_,,.,:? , '-""" ~ / ~,--t!t '.. 70 4--\ 7 ✓ ""'~ ~ ... "";-.. ::,.;...~ ~~ ,;,,-f '111+r , · -il»~~1 ,.. ,.~ 1ri-& -r-~ ' \ '2.,~.... , 1008 LOW \;:,°' 01 • ) J~t.' , p,O:t, ~ ~ / 1/8 ) i 4' ; # ~ •1-~~ , ~ ...."b -~- ,. ~1,P96 / ., :1g:~ • 11 / ~ ~j ~ fl -.,: ~' .,,__\-r..').'i.;~~.,o~o'of.>~" /J;.o~•~/~ ,c.~ , ;\> \ A<, •ti•• j "~\.,-,r;..,t 1°~ '<; .- f>o'~/ "/.o ., ~ ":,\ ,~ '-"'-"' ~,,C/' :. . 3•~ ~?,4",.." 7.A">/"'1,.,;,,,"\, 41~ ,'2°fi,?,.,~':, 1- - . . r 2V..' sf~~J'"' ~ \ . \ •.•,~ 1 ~: \::...or ;:• ,q'a. .• • - r:P'"" - ~ ,\,q,,l-s .¥:>~~~y,," S ·, J ~--~(.~"' P., oo£' .\ "' dffi c\ _ \' ' +-a J+) rz •z_ s.ff;;,l,'f.-'1- \-~•~,~~;>4~ ·""' ,·l"..11 ,.,;;;"°--'\.1/ ~"'-.j" ' ,t.l 4-~""~• .:.2.-, ~ - "'!.'I ¾--v:; 'J'. ,•,~ + e l ~ , .-1.'.'-t'- - Cg) ~ s.l 445 .,.1..;;; l 15o '2S4. 3~Tt. co / ~ • ~st- 4<./,\~ 5.l< - P"- 37ye - - .. • <; I - __. :;l..._,-, / y\ '¾,\''i.e. ,f,9'1~.,,.?.l.',,,, i~t~',; ' ;~'!!>,' " ' "" At,~..._-: A~'fi - - , ' \' 2.c.s , '.':> ~"-'e 3..._:;,,~ : ~ , _.,,~ , - ~ 0'1.~r r 1 • \••• 't<I" <- - , / ' 2~" '-- - - - - - - - , 31 '21> 3/ ....___...-l '<-~ 32.., ~ ... ., - - -- -:,_l,(~.,_ , :"7.if~o,tH~ e " / 1 -..J ' ~ 1 '2'S~ 1 c,.O -'j;,. \ ,:,; h~ •"',~f..,i,. •~ &>c:-' ~ }.~-"'\__,., ' ~j , " ·,,_ , ~--~~ ,. • "' ,-.:,0 , . /,, ,~ ... 1~ ~ '"'\ ~ "fa ~ -,,.,., \ ~ ~'"h.- : , HIGH ,,_,, - ---~.+,r L- - -,z34 +.3.r_ .,,--..-<2">\;~ 3'2 --0 4o ~ o2.,- I • 1 ,' ,, '< - ~ O ,,, I I I ';' ,2--11!>" •\ 1008 ~-. '-..__So\ ~ - !>?' " i•ii~~\~5'f•i·"T,s<!~~~ f n~ .......__ ~ -vt-t .........__,,,, /·.:,, · ... :·,'· - _ ziJ "'"J;..o- ;;~",c. ; _.. • • O':)e • I •) ,\ .' \~~"'~ \ ....'e--'~, / 1016 : 1012 ii , ~ .7:''· '.I ·I ""' \ , • o\o + h J~"-~ °1 1 ~o 1 ~~-..... ' ,,. OS,~ -1.. ..... •I '-1 ;;;_\ 15~ · ,•,.;; r· -r,, 73 .,,(o4o 4 o..;:_·~ · ;,-. 4-\- o~':.> • s~ ,., ;• 0 91 I !..'--1. ,, . . ..,... ~ <.:. " '( y ~ '.JoP ~-<> , _.:;) 3 7 255 3,a_ :zs<- 3 9 0+2r ~, ,;.e, 0 -/ 5 .,o..:;_, 7@ f<>o - - - 'T' - 1 I • •t~},..-,.-.,, ;39~ 7_ - rel\ i 454 1~ 1 ·, '•...I•~ /{;_:i\" (:s,?~• ' • IOQ4 :., / I e \ ss _,,,- - - .,. __ ' 1-3 'T' t~ <> r, 13 ~ ~~o \ S 7? , ,,1':\ C:, 3 ·7 ,. '• l ".3L ..,; ~ 1~ • ' " •·· !I 1 l I ' ·. ·~, '3')°'):_;3,,7~ ~\\ ~- - 1 ~1-.J r ~~ 1 1~,J . r .I • Ji .., . ,I \i ~ 'Zi 'l j~ ' ' / 2 I ? .., r • , ~•,: ,.3 • • _ ~~;~L - - - ~!ra]i ,,_.,,,,.;.: \.: -' f~-r{':}} 1 ~~-:1 - 3 .-... - ~ J~{~s '- I' ~o ~ •• <') JI:__\ \ ' 70 , J '. ·-'' ·: :Jc. c,~, ---3 - Ii\ tot 1 , ' • 4.J.~ , ~. w ~ • 0 - 3'§' S. , ,· 4 2,- 194, ~ 43 3 9 14 1 4 2 ~ :1 / { ~ , _~> @•2✓~ 57 /, lo '. , / ./ ~ 5 ,-.. .1..._ ~ : , 54s ri .._I:,,~ i,. ~ .._ ; ' ;~\tJ .. ' s , / / 1 / 8 I 1~f!1> ; ,f;;~ l ' - : - , , •• • , ' S ,{ ¾ / .I I ~ .' ,;-,,, ! 1, /,.. . ? 38 - , 1<,M (OJ 3/ ' ./' ,,' , ',. •/ I f ,r, I -_- - - - I I I -................__/ , I 012 I I f '· \1, 1020 Re:,~ !'? - , "'l ~ ~c"3':: , S:?~"-~J, , - - ._ , SURFACE WEAlHER MAP AN D STAT ION WF./ITHER . AT 7 .00 A . M E 5 . T I ' ,_ ".s '·· ¾0·"-fl I "-VIS, \ '®;r~ / $~~ ' < ~----- "'J · ri I " '-,. I I ''/ A '. ,10 1:· \ \ 1 l~, '· I , ! co w w , I / • .,. I/ ' ~,if:f,:: - ' ,·.....J," _ ·l h- :I;~'/-@ 'o, ~ - ·,, " ' '.: J -~ . : _·,, i,,t~. • ,, / •• JG ' '< , Is.,. . '( •I , / ., ''" ,' I a "- ~ 7.b -<~ 1' fi , o...,®1";, C\ "3z\)1: o -_1\,\ ,. - " j <'' ,' ' ,z'o-u'O'Q\ l ' ',,, \ •I~ tooa ' \ -- ' ~<..,r,., ' • ,s ~1016 34~+1" t_OI . /4( 1008 1012 )Olli'.•; '' __ _ , . . . . . . _ _ . . , ~• -,,. " , "- lQQ8 •,J , . . I ~ ' IQJ2 ' ' •I 1008 ·, ~;. ·' ~'""~- ~ tr~~, v<,,'l,P,.' . , . / ' • \ . .fi. / , ,_, o ~• , ) ' ,+,') ~&~";,\ V .,. C, > • I' ' !f),( 8-:: • ,, ,,, , (.· SJ;., ,, :, ,· ~J·r ,. -·• ' \!' '' • ',; ...~C>I ✓ / i, " -1 '- ~1~0/l~ 1004 ~ •✓- "- :..a• : • v> ,. . ·,/ , ~ • .. ,, ,, , ~ , , , o I 1o c1,o , CJf ) S ' •, , · '. - _f.oa ' , o. ,, o,. , , "O -•?.'- \ 11':l ' I , ' OS '' '\ '" .. - W "!,I> o ,I ~ 0 1•"' I w = ' ' ,- ' "' Y>-,J :-.,"',~~:/ •• __ - • , "Qfo"'!,Vb ' .,.J , • ~• ~I,.,~ ,,• • S • " ...•• \ ,., ' I },--"' "l.,r/Af , /: . ~ ~ . , . - , """"''~'/ ~ o ,_, ~ ,., , 3'2.@_,.;:. ~ o - 7" ~l\l <~ . 0 ,,,,.,, ,,,,_, , . . ' zs3 9 · 71 , , _.,.1, ___ ' - J '1- ~@·~ ~ --, ;, " :2~~~ ~✓-~2~, -.... '.;?ao · .. I 3 / \ o..:;_ s., 9,8,>o_ - . - , ""\ - , I •,• , - •o~ _· FACE h l'?J', 1,' ~ 3"N_ \ 2,ou ", - . . •o•·• c 1 ¾lJl~Ot,- , . • '!• 3• ~-,, O> w -61 , ~- ) '°~@+7' 7",zc37 £ ~' @>7/ , ,,,, "b :,;;_ 'j,,,}' f ~ ' ,-,; _.?"'So ',ii,,.,', 1 "-' a o-"¢... b,~, • I ' O • "" p~'-.o.-o- ,. ~--" '«,> :l, , 1 - ~'l.~~~' , · .t"'l.'l.'l.J tp_.:, c, -., '"> -4-\ '"• -. f ~v , - .. - 1 • oor ~ " 0- • _ ~ " , ' , I • • I / 1 ' I '"•• ' I ,... ..... • --~· ..:.> ,; ' .i. .. P21 ~ 1 ,oo, • \ \ ' a,C>])ao-z. • o, ~~ 1 A'> -~ 0, , ' F", gu re 3• -~ •• I' ,..,-~ •~;•'- "':/ ;~ ,. ~ ",'i, 9 7J>'<>; _..... I I "" A """""' ~ d \ ~"\ re r.. ,. I. dm -,....,~A~.-.;:.~1 ~,;•:,,"~ ~ '-.· .,, J. • "' A, . - ~lQ,\~d,> do tSA:....#1 lo 1° ,_,.,., <.4'-s . --~..---~ ~ , -. • ">~ .:,~ .. ~~ 1>"'~ • • -- - · / :/ ~ •• /-;,,--i. .r. ' ""='~I <, ~ - •-"',~, :3:.,,,~/ • <,~: ,j:;,,.,, ~.,(~<oJ--.. __'< ,~J ~-'~ ,~, I /., • , " ... •• • .-, .•• -· l 975 •• ""--'!. \ ,,E> .,, ,, , ·" ,oO"C " -~ ,>.b,. ,E> /4\0 .?f/i.1 ,. " ':.i .-.1i~~'"'\r ~ ®<.,,.·,~ ,~-f..,-, ..,_, .,p,, ~<S,", .-;_';.@~ ·" ( · September 13 ., fl> 'I."' ., ;. , ·~/fu/ o• ~\1 •~M>,i/ > .- -.J,l';;':Jd,r""'i~v-<;;z.j;- · •. r,P ~ 7., fl.."{'✓' ~~~ y -4-··, ~ ~ . "' • ~.t l l ;'<'."l',s., ~1i: .. ,/; C:, · -- · ~°""'' %•00/0' ~~ ~~!.-. .- ~, "•'';° ~ ' ¼ ~ ~./ ~~--t'o'✓"·" z ,,= ,. {o - • ~'c:,7 -,. ,ol , · -.r., ~ J.• '"-''·'· , .~1~f'(-,i',<\} , .. -~ » ~ ~~ty_Ji ,. ~~ ' ,' - : I~_, '£/. __..,;::....,..2;.J~~-=- , A~o...;,~ , s~" _ll,,, -~ - -~~~~ ~ ..3r ,.,,~.-.: . :.#::' ~ .·:·.1;C#:A-" ,,. ~ -\-....o_;::.,., - ' · - _,.,.,,., •• , . --'- 'i ',; L<,, -,.C' So~IAJJA~~ ~n\~~ ,.,, f'-~•3 __--:-'s: -....°"-- .'::. . .... -·- · · -- '. ... . l. _ _, ,--,,,,.. r , ~ • ''"\," d'o -10\ - ~ · ·. • -~ - a 5<"-_,e.~ l f : ' ? ~ rl l><> ••'~ - , c--4;__,~1,a,- ; /4 °T\ 54 lo ·r - -~-'>fi',2_ .,, ' ._,,, ,.,,,_, u• ; ""-1-' _ " 70\073 "7 57~ l '>, a ', lb :I& O.,,_ '7> " •• , .. _,.,, 00 3"> 1 . II ·:: ··' .'.·? · ·. .. ...______ ."' v .,_ I o , .,, .... -7 ;:f --e (_ ''c/s ~ '< '\ oa-, , ..,,, < \ I 0 f1,s, ,~ ' .,✓. s,.,';'-, <t :,_, \ l.oj \, •\✓-,,, 1'~, \ \ .~ , ';.<., 1 '-t;~ ~-~"" '// ' P>' <9' , "v,s, , 4 ~ · ' -,A!P"<" ' '{al ,,.,. ' ,. ',, ·'"-"« ' !', '\~; ..<bf,,, \ - ~· ",' ~'fl;) ~ : , ...~,"{>~," ;;,- ]'d>f~, '· \ , \s1~~ ', .,, 4e ~''" "°T. l:,=I ,,,'ltJ .J.~•~G~'F.J ,;_,,,,,.,"f., "'~;,., ,} "'. l, • ,.."f,,,'1oJ,,, ,, , , ,~-, ~ " o' oc • =°'-"'""n,·~ •$ J%}., " " 1 -' • , ), \ ~®'', ~ -o,§' . '"I' ~~• ,i ~ I -,<' ,., = , t,2 ,>,,' •=" ' , ' _., -A{_•>ri,•p, •C •, , .. WEATH fR MAP '~,'J:6'% ww,rn . , B I ,r!~ • 5ir. fJ-1?_ ,~ l ' -,w.•2.0 ,;, :,_~,, 4' .i!,o@ · <o '- ",'., / ' - .. J ' ,, ·•cl,;y'. .'"';oj)j;,,,..,,:.. - ::_~'ls, ..ic;:,'.':t~£~,1?'/'J':,l(,~ .·'·,. ..ox , ( :~+oa8~ ,;· " ~ '°'----. ;, ~ ;.,,, , ..~ " • • ,.,, •' I ~<@1g-, . / .sQ~,3 4-3.}t; / ' ¼®-~ ,::,.,,-z~,-.J'~?~/4~~ r •·~~,.,,., \· ~ < · "·' a , -, ..... I ' ~ - . . , ' ~:P ,-<:':.- / 1 . ~'-<;;L :i < gz.38 . . ~✓\. '- '8:,__"'I.,"' ~ ~,( /1 .:..,J ;.i ,,, $>.:~. ". ,,, . ' ',' ,, '" - - - "I '°''' ,.,,. : ~ ' • ,"~:. ""' ,· Y-= ., -~ -"" ~ ,;,, '-, _""= ' , ~"°-""'' ~ • q--\~ ', ' ,.,.. = , . , ~ \ . • • 1!." ~ "< ,.J~o '> /",,'~ ' ~ J',, • •. /,, , ~ -is~I .l:4/'< ./ ~, \ ,, c,"(; .,,, ,vfo'lf",J}:..~':'-,'!>,J;,~~_,, ~p;,. ~ I .~HIGW '\ -,,,p,., <.➔e.~ 101,if.,<,. ,,_,,,;:, o/-'o• ,,, •," . ' . - .. - : , '1 ~,.W,, ~o••> " ,> C,,_ , J' • ,-"1,•I i:"'l> • ,;--'c • :;, ' ' ~ < • ' ' I , - - -, - ., .- ·· .<) \ ' • _: ,\" \1,, \ ~ w w 340 ,- s.. Q) .0 ~ +> 0. Q) V) "'0 ., N 0 0 "'0 341 Ln r--- 0'\ r- Ln r- tD 0~ - - - - - ~~ .___::... $" C N 0 I - I I 1 342 to assume that each of the areas in the path of a high pressure system contributes equal amounts of terpenes. For the purpose of this calculation, the value given by Ripperton et al. (1967) is used since it represents an upper bound on the estimates. Consider , for example, a high pressure system entering the U.S. in the summer and crossing from Montana to Iowa i n two days. The area covered by this system is assumed to equal 200,000 km 2 which corresponds to a rough estimate of the average size of air masses observed during the summers of 1975 and 1976. The total flux of terpenes is then very easily calculated for the two day residence time : 160 kg/km 2-day x 2-1 05 km 2 x 2 days= 3.2•10 7 kg. (1) Data presented by the National Weather Bureau indicate that the average maximum mixing height over the Midwest ranges from 1000 m to 2000 m. For an assumed summertime mixing height of 2000 m, the well - mixed concentration of t erpenes equals 30 ppb (v/v, assuming all terpen es are emitted as a- pinene ). This level is similar to those measu red by Ra smussen and Went (1965). Given the range of emiss i on rates in Table l , the second day concentration could range from 2 ppb to 30 ppb. In a model of terpen e behavior in high pressure systems, a sink term wou ld necessarily be included . Smog- chamber experiments by O' Brien et al. (1975a) suggest that in the presence of other photochemical smog precursors, te rpenes will react rapidly to form liquid organic aerosols conta in ing dicarboxylic acids and some polymerized material . Ambient aerosols f rom both urban and rural locations in Ca lifornia conta ined significant amounts of these products according 343 to a second paper by O'Brien et al . (1975b). The flux of terpenes from crops and forests in and around the San Joaquin Valley may be significant in the degradatio n of visibility and air quality in this regiono Ripperton et al. (1973) found that o3 synthesis was greatly increased in irradiated smog chamber experiments which included a-pinene over those which contained only water vapor and N0 2. Graedel and Allara (1976) briefly described a photochemical model which included reactions involving a-pinene. Calculations indicated that a system containing estimated rural concentrations of a-pinene, similar to that predicted to exist in the high pressure systems, produced a maximum rural ozone concentration equal to 32 ppb. They did not report on any investigations of the sensiti vity of this model to variations in the initial concentration of a-pinene. The preceding studies generally indicate that terpenes emitted into the atmosphere can readily react in the ozone photochemical system to form significant amounts of aerosols and to parti cipate in the synthesis and destruction of o3. The calculated flux of terpenes into the high pressure systems which cross the Great Plains appear to be large enough to ha ve a significant impact upon aerosol formation and ozone concentrati ons in air upwind of urban areas . Since the composition of air arriv ing in urban areas is of importance in understandin g and predic ting the subsequent urban plume chemistry, it appears that the estimated rural terpe ne flux is great enough to call for more experimental data concernin g the emission rates and reaction mechanisms over rural areas. 344 Literat ure Cited Darn all, K. R. , A. C. Lloyd , A. M. Weiner, and J. N. Pitts, Jr. (1976) "Reactivi ty Scale for Atmospheri c Hydrocarbons Based on Reaction with Hydroxyl Radical," Envi r on . Sci. Tech . .1.Q., 692 -696 . Graedel, T. E., and D. L. Allara (1976 ) "The Ki neti c Ozone Photochemistry of Natural and Perturbed Nonurban Tropospheres," i n Pro ceed ings of the International Conference Q!!_ Phot ochemical Oxidant Pollution and Its Control Vol. 1. U. s. Environmental Protection Agency EPA-600/3- 77-00la , Resea r ch Tri angle Park, North Carolina, pp. 467-473. Husar, R. B., D. E. Pa tterson , C. C. Pa l ey , and N. V. Gill ani (1976) "Ozone in Hazy Ai r Masses ," in Proceedings of the International Conference on Photochemical Oxidant Pollution and Its Control Vol. 1. U.S. Environmental Protection Agency EPA-600/3-77-00la, Resea r ch Triangle Park, North Caro lina, pp. 275-282. National Atlas of the United States (1970). Interior , W as hington , D. C. U. S. Department of the O' Brien , R. J ., J. H. Crabtree, J . R. Holmes, M. C. Hoggan, and A. H. Bocki an (1 97 5b) "Formati on of Photochemical Aer osol from Hydrocarbons : Atmosp he ric Analysis, 11 Environ. Sci. Tech. 2_, 577-585. O'Brien, R. J ., J . R. Holmes , and A. H. Bocki an (1 975) 11 Formation of Photochemical Aerosol f r om Hydrocarbon s: Chemical Reactivity and Products , " Env i ron . Sci . Tech . 2_, 568-576. Rasmussen , R. A. (1972) "What Do the Hydrocarbons from Trees Contribute to Air Pollution , "~- Air Poll . Contr. Assoc. g, 537-543. Rasmussen, R. A., and F. W. Wen t (1965) "Volatile Organic Material of Plant Or igi n i n t he Atmosphere," Proc . Nat. Acad. Sci. g , 215- 220. Reynolds , S. D., and J. H. Seinfeld (1975) "Interim Evaluation of Strateg ies for Meeting Ambient Air Quality Stan dard for Photochemica l Oxi dant/' Environ. Sci. Tech_. 2_, 433-447 . Ripperton , L.A., and D. Li ll ian (1973) "The Effect of Water Vapor on Ozone Synthesis in the Phot o-oxidati on of Alpha-Pinene , 11 in Chemi stry of Air Pollution . MSS Information Corporat ion , New York, pp. 37- 49 . Ripperton, L.A ., J. J. B. Worth, F. M. Vukovich , and C. E. Decker (1976) "Resear ch Triangle In st itute Studies of High Ozone Concentrations in Nonurban Areas ," in Proceedings of the International Confere nce on Photoc hemical Oxidant Pollution and Its Control Vol . l. U. S. Environmental Protection Agency EPA- 600/3-77-0ola;-Research Triangle Pa rk, North Carol i na , pp . 413-424 . 345 Ripperton , L.A ., 0. Whi t e, and H. E. Jeffries (1967) "Gas Phase Ozone-Pinene Reactions,'' Div . of Water , Air , and Waste Chemistry, 147th National Meeting Amer ican Chemical Soci ety, Ch icago, Ill inois , pp . 54-5~ U. s. Weather Bureau (1975 , 1976) "Dail y Weat her Maps, 11 Weekly Series , U. S. Department of Commerce. Went, F. W. (1960) "Organic Matte r in the Atmosphere and i ts Possible Relation to Petroleum Formation , 11 Proc . Nat . Acad . Sci. 46, 212-221.