Open AccessEstimating Bias of Satellite-Based Precipitation Estimates
Author(s) -
Thomas M. Smith,
Phillip A. Arkin,
John J. Bates,
George J. Huffman
Publication year - 2006
Publication title -
journal of hydrometeorology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.733
H-Index - 123
eISSN - 1525-755X
pISSN - 1525-7541
DOI - 10.1175/jhm524.1
Subject(s) - satellite , environmental science , precipitation , sampling (signal processing) , statistics , estimation , scale (ratio) , sampling bias , climatology , econometrics , meteorology , mathematics , sample size determination , computer science , geography , geology , computer vision , management , cartography , filter (signal processing) , aerospace engineering , engineering , economics
Systematic biases in satellite-based precipitation estimates can be the dominant component of their uncertainty. These biases may not be reduced by averaging, which makes their evaluation particularly important. Described here are several methods of evaluating the biases and their characteristics. Methods are developed and tested using monthly average precipitation estimates from several satellites. Direct estimates of bias are obtained from analysis of satellite–gauge estimates, and they indicate the general bias patterns and magnitudes over land. Direct estimates cannot be computed over the oceans, so indirect-bias estimates based on ensembles of satellite and gauge estimates are also developed. These indirect estimates are consistent with direct estimates in locations where they can be compared, while giving near-global coverage. For both bias estimates computed here, the bias uncertainty is higher than nonsystematic error estimates, caused by random or sampling errors and which have been previously reported by others for satellite estimates. Because of their greater spatial coverage, indirect-bias estimates are preferable for bias adjustment of satellite-based precipitation. The adjustment methods developed reduce the bias associated with each satellite while estimating the remaining bias uncertainty for the satellite. By adjusting all satellites to a consistent base, the bias adjustments also minimize artificial climate-scale variations in analyses that could be caused by the addition or removal of satellite products as their availability changes.