A NOTE ON THE DERIVATION OF A SCALE MEASURE FOR PRECIPITATION EVENTS

Abstract

The importance of the relationship between scale and predictability suggests the need to define rational scale measures for various weather phenomena. A simple probability model is used to relate a characteristic space scale measure to point frequencies. In a previous study of the predictability of weather elements (Roberts 1969), the author found some degree of confirmation of a widely held hypothesis regarding the existence of a relationship between the space scale of weather phenomena and the accuracy with which the phenomena are predicted. A recent survey of National Weather Service forecasting performance also showed that the skill of the precipitation forecasts released to the public differed widely from one part of the country to another and between seasons and times of the day (Roberts et al. 1969). Since the basic ability of forecasters should be essentially the same for the entire country, these differences must be attributable to variations in some feature of the precipitation regimes characterizing different parts of the country, possibly involving both time and space scale factors. These findings suggest the need to develop a means of classifying and describing weather phenomena according to characteristic time and space scale measures. In addition to shedding considerable light on the problems of predicting and describing weather phenomena, such information might have a wide variety of applications. Uses for precipitation scale data appear in such diverse fields as weather forecasting, hydrology, weather modification experiments, forest fire danger assessment, and forest fire control, just to mention a few. This paper reports on the problem of defining such a measure for precipitation and examines some practical methods for determining its statistical proper ties.