An investigation of the error budget of tropical rainfall accumulation derived from merged passive microwave and infrared satellite measurements



An ongoing change in the theoretical framework from deterministic to probabilistic satellite rainfall estimations emerges from applications that require an error associated with rain estimates. The error budget for accumulated rainfall consists of several terms; these terms are related to sampling, algorithmic and calibration errors. From a number of studies, various errors were derived which have improved our understanding of the different terms in this error budget. In this paper, a methodological effort leading to the evaluation of a Tropical Amount of Precipitation with an Estimation of ERrors (TAPEER) is presented. It involves first merging passive microwave instantaneous rain rates from the BRAIN algorithm together with infrared imagery to build rain accumulations, and then evaluating the different terms of the error budget using two techniques. A dedicated error model is used to evaluate sampling errors and a forward error propagation approach is used for the estimation of algorithmic and calibration errors. One of the main findings in this study is the large contribution of the sampling errors and the algorithmic errors of BRAIN on medium rain rates (2–10 mm h−1) in the total error budget. This methodology leads to the formulation of a satellite rainfall product called TAPEER-BRAIN. This product will be one of the operational rainfall products for the Megha-Tropiques mission and will provide 1°/1-day accumulated rainfall estimations and associated error over the whole tropical belt. Copyright © 2012 Royal Meteorological Society