Validation,Application And Improvement Of Satellite Remote Sensing Precipitation In The Globe And Typical Regions

Precipitation is one of the most important components of the global water and energy cycle.Satellite remote sensing is the only way to achieve global precipitation observation.The Tropical Rainfall Measuring Mission(TRMM),launched in 1997,opens the golden era of satellite remote sensing of precipitation.The Global Precipitation Measurement(GPM)mission,launched in 2014,has substituted TRMM and represents the hotspot and forefront of satellite precipitation.This study aims to answer three major questions including(1)how accurate are the latest GPM products?(2)what is the performance of satellite precipitation in hydrological application? and(3)how to improve current algorithms and products? This paper focused on the validation,application and improvement of satellite precipitation which forms a research system with scientific logic.The Level 2 products of spaceborne radar precipitation and the Level 3 products of multisatellite precipitation were validated on multiple temporal and spatial scales.Then,the Level 2 products were applied to analyze the precipitation-topography relationship over the Tibetan Plateau,and the Level 3 products were applied to monitor the rainstorm and flood disasters in southern China.Finally,Level 2 snowfall retrieval algorithm and the Level 3 precipitation phase discrimination product of GPM were improved.Validation is the premise of improvement and application.For the Level 2 products,for the first time,the rainfall and snowfall data of three existing spaceborne precipitation radars(TRMM PR,GPM DPR and Cloud Sat CPR)were inter-compared,and the accuracy of PR and DPR was evaluated by using high-density hourly rainfall gauges in China.For the Level 3 products,TRMM and GPM products were validated at multiple temporal scales in the Ganjiang river basin,typical regions and Mainland China.Furthermore,an error function was proposed to describe the uncertainties of using rain gauge data to evaluate satellite products.Application is the embodiment of the value of validation and application.For the Level 2 products,the total errors and retrieval errors of TRMM PR and GPM DPR in the Tibetan Plateau are estimated and corrected on the basin scale based on ground observations.The precipitation-topography relationship is analyzed and explained from the aspects of the plateau,watersheds,mountains and profiles using the corrected data.For the Level 3 products,under the background of the extensive rainstorm and flood disasters in southern China in 2016,a flood identification method based on accumulated rainfall and underlying surface characteristics were proposed to investigate the performance of four mainstream near-real-time satellite precipitation products in monitoring floods caused by heavy rain storms.Improvement originates from validation and application.For the Level 2 algorithm,GPM DPR and Cloud Sat CPR rainfall and snowfall were fused in high latitudes considering their complementarity.The merged data set was used to train the deep neural network(DNN),with GPM passive microwave data,MODIS infrared data and ECMWF reanalysis data being used as input.In this way,direct retrieval of rainfall and snowfall in high latitudes was realized which was proved better than the algorithm used by GPM at present.For the Level 3 products,eight temperature downscaling methods are adopted to address the problem caused by the coarse resolution of reanalysis data used in precipitation phase discrimination by GPM products.Among them,the methods based on the temperature lapse rate are better than the statistical interpolation methods,and the phase discrimination based on downscaled wet-bulb temperature was better than that of the Level 3 products of GPM.