Rain Estimation from Satellites: An Examination of the Griffith-Woodley Technique

Abstract

The Griffith-Woodley Technique (GWT) is an approach to estimating precipitation using infrared observations of clouds from geosynchronous satellites. It is examined in three ways: an analysis of the terms in the GWT equations; a case study of infrared imagery portraying convective development over Florida; and the comparison of a simplified equation set and resultant rain maps to results using the GWT. The objective is to determine the dominant factors in the calculation of GWT rain estimates. Analysis of a single day's convection over Florida produced a number of significant insights into various terms in the GWT rainfall equations. Due to the definition of clouds by a threshold isotherm (?20°C), the majority of clouds on this day did not go through an idealized life cycle before losing their identity through merger, splitting, etc. As a result, 82% of the clouds had a defined life of 1 h (two images) or less: 64% of the defined clouds were assessed no rain because the empirically derived ratio of radar echo area to cloud area was zero for 64% of the sampled clouds. For 76% of the sample, the temperature weighting term was identically 1.0. Terms not directly related to cloud area were essentially uncorrelated with GWT rain volume, but cloud area itself was highly correlated (r=0.93). Discriminating parameters in the GWT rain apportionment algorithm were the temperatures that define the coldest 50% and coldest 10% cloud areas. Further apportionment beyond these two thresholds was found to be unnecessary. Simplifying assumptions were made to the GWT such that the resultant equations were independent of cloud life history. Application of a simple algorithm incorporating these assumptions led to daily rainfall patterns on three days that were, to first order, the same as those calculated from the GWT. Daily totals in the FACE target area were actually closer to the gage determined rain depths than the GWT estimates. Correlations between half-hourly estimates from both techniques and the gage amounts were poor. We conclude that the GWT is unnecessarily complicated for use in estimating daily rainfall. A method in which the relationship between clouds and rain is simple and straightforward can, to first order, duplicate the results of the GWT.