THE ESTIMATION OF MONTHLY RAINFALL FROM SATELLITE DATA

Abstract

The global distribution of precipitation is an outstanding example of a pattern whose form cannot be deduced very satisfactorily from conventional observational data. Many investigations of the global hydrological cycle across real periods of time are based on broadly generalized assumptions concerning rainfall patterns, especially over the world's oceans. This paper explores the feasibility of employing observational data from meteorological satellites to yield more acceptable maps of rainfall across periods of 1 mo and upward than is possible using conventional surface measurements, whose distributions are less uniform, and whose derivations are more heterogeneous, than the satellite data coverage. The central problem is related to the fact that satellites cannot measure rainfall directly, and the solution of this problem necessitates the construction of a rainfall coefficient equation to be evaluated from nephanalysis indications of cloud cover. Evaluated coefficients for the months of March, April, May, and June 1966 were plotted against the corresponding rainfall recordings from a selected scatter of surface stations in the Australian region, and a best fit regression line was computed to relate the two sets of values. The regression equation was used subsequently in the compilation of a map of the estimated precipitation field for July 1966 covering Australia and adjacent areas. Finally, the implications, and some potential applications, of the method are discussed, and suggestions are made concerning its possible further development in association with satellite photographs and computer techniques of data processing and data analysis.