Comparison of Two Methods for Estimating the Sampling-Related Uncertainty of Satellite Rainfall Averages Based on a Large Radar Data Set

摘要

A number of satellites presently in orbit, including the Tropical Rainfall Measuring Mission (TRMM), are producing global maps of rainfall amounts, sometimes on a daily basis, sometimes on a monthly basis. The rainfall values on these maps have considerable errors in them, partly due to problems with remote sensing techniques for measuring rain, and partly because the satellite doesn't view each spot on the earth continuously. The latter kind of error is referred to as 'sampling error,' because the maps are derived from occasional samples or snapshots taken by the satellite as it orbits the earth. There have been many studies attempting to provide quantitative estimates of how big sampling error might be for each rainfall value at each location on a satellite rainfall map. This paper is a significant contribution to this effort because it uses radar data from a large section of the U.S. (similar to the radar data displayed on weather channels) to make many estimates of what the sampling error in satellite rainfall maps should be for thousands of different cases.