Forecasting Daily Rainfall From Satellite Data

Abstract

This paper outlines a rainfall forecasting model involving satellite nephanalyses and certain conventional weather data for application in regions where the rain-bearing clouds are advectional rather than convectional. Essentially, the scheme depends upon the selection each day of an appropriate cloud quadrant for the station in question and the evaluation of a ?rainfall prediction index? for that quadrant, embracing its synoptic weather organization, its mean cloud cover, and its included cloud types. The rainfall prediction index is then used to forecast a class of rainfall for the day ahead in terms of ?no rain,? ?light rain,? ?moderate rain,? or ?heavy rain;? being interpreted through tabulations of related rainfall measurements. The essential indices are specified for Valentia, Republic of Ireland, from January through June 1967. Rainfall ?forecasts? are then made for July through December 1967 using these indices, and the results are discussed. Verification of these forecasts indicates a ratio of approximately 3:1 when they are classified on a simple right : wrong basis. Probable sources of error are indicated and discussed. Several approaches for extending and improving the model are, suggested. It is concluded that the method could be profitably applied in weather and hydrological forecasting programs for regions that receive their rainfall from clouds advected from oceanic areas poorly documented by conventional weather observations.