Forecasting Mean Weekly Temperature by Statistical Synoptic Considerations

Author: Ho, Chung Pen

Year: 1948

Degree: Dissertation (Ph.D.)

Advisor: Elliott, Robert Dunshee

Committee Member: Unknown, Unknown

Option: Meteorology; Physics

DOI: 10.7907/hmdc-jx81

Abstract

The scope of this thesis is to explore the relation between the weekly temperature anomaly at a test location and the general temperature anomaly pattern for the preceding week. Since the general pattern of circulation of a given week has a close relationship to that of the preceding week, and because the circulation pattern establishes the temperature anomaly pattern, it seems logical to infer that the temperature anomalies of two consecutive weeks are correlated. Moreover, as far as the properties of air mass are concerned, temperature is the more conservative property.

A test location is chosen and twenty to thirty other stations in an evenly scattered network around the United States are used in representing the temperature anomaly pattern during the preceding week. The weekly temperature distribution curves are then prepared for each station and a division into five ranges or categories is made. Contingency tables are then prepared showing the relation between the temperature anomalies of the test station and those of each network station during the preceding week. The mean number of matches are found and the standard deviation computed. The difference between the mean number of matches and actual number of matches divided by standard deviation gives the normal deviate. There exists a positive correlation if the normal deviate is positive and vice versa.

Using the calculated normal deviate at each of the network stations a contour map of this number is drawn. The result is interpreted on a synoptic basis. If some parts are inconsistent from the viewpoint; of synoptic considerations, then those parts are disregarded as a forecasting basis.

In practical application, the relationships would be established for a large number of test or key stations and thus a weekly forecast temperature anomaly pattern could be made.

Files