Comparison of TRMM/PR rain rate with ground based observation

摘要

The precipitation Radar (PR) on the Tropical Rainfall Measuring Mission (TRMM) satellite continues observation. Many ground validation efforts have been made to confirm the accuracy of rain rates estimated by standard PR algorithms from observed radar reflectivity. However, so far there have been no conclusive results giving the absolute difference of PR estimated rain rates from the true rain rates from ground validation efforts. The difficulty of ground validation of rain is due to the large variance of spatial and temporal distributions of rain. However, there is a dense rain gauge and radar observation network operated by Japan Meteorological Agency (JMA) that covers Japan. In this paper, TRMM estimated rain rates are compared to Radar-AMeDAS rain analysis data on a monthly averaged basis, and retrieval errors were estimated by assuming that Radar-AMeDAS is truth data. The estimated averaged retrieval error for five 5 degree by 5 degree areas for 12 months was 25%, and PR and Radar-AMeDAS rain amounts were almost the same for a year. However, the estimate exhibits seasonal dependency. Specifically, PR overestimates rain rates in winter and under estimates them in summer. Sampling errors included in TRMM monthly products were also estimated. The estimated sampling error was 15%. This magnitude of sampling error seems quite reasonable considering the results before TRMM was launched.