The Problem And Primary Exploration Of Exorbitant Precipitation Retrieval In GPM Product

Precipitation is closely related with our daily life.It has been a hot issue on how to observe and describe it more properly.The GPM(Global Precipitation Measurement)project works as a succession of TRMM(Tropical Rainfall Measuring Mission),which is intended to observe more comprehensive characteristic of precipitation.The loaded Dual Frequency Precipitation Radar adopts dual frequency retrieval algorithm for radar echo,which can capture light rain rates theoretically and improve observational performance.This paper is based on ground station observation,analyzing and comparing 70 stations rain rate in minute level from 1st December 2015 to 30th November 2016 and GPM DPR precipitation retrieval products,putting forward abnormality in rainfall retrieval and analyzing it from respect of algorithm.Then we use data after quality control to further discuss the adaption of satellite data in Jiangsu area.The basical conclusions are as follow:(1)The best matching time for NS,MS and HS scan products and ground rain gauge observation comparsion is 6th minute after passing time.Due to wide swath of NS products,it is relatively with advantage of sample sizes:the probability of detection all year round reaches as high as 60%,the false alarm ration only 15%,the root mean error probably 2mm/hr,the correlation coefficient above 0.5,the relative bias is-20%(the satellite estimation is lower than ground observation).Meanwhile,the NS precipitation products retrieve the best compared with other seasons,the scattering difference varies from station to station in space.(2)From comparsion of ground validation and satellite data,we find big errors in heavy rainfall process in normal scan(DPR NS)?matched scan(DPR MS)and retrieval abnormality and analyze one of causes.In reality,the abnormality pixel in MS and NS products has strong impact in statistics of rain rate retrieval.Using high sensitivity scan(DPR HS)module can pick out these abnormality,improving effectively in rainfall products accuracy.(3)GPM satellite rainfall retrieval algorithm of R-Dm is strongly sensitive to nonlinear retrieval parameter ?.In strong rainfall cases,that the estimated errors are easily amplified by heavy precipitation particles from NS and MS products is inadequate restrictions for retrieval parameter ? in algorithm.