Accuracy Evaluation And Downscaling Of IMERG Precipitation In Arid And Semiarid Regions Of China

Precipitation is a crucial driving factor in the water cycle,and precipitation data is an important parameter in climate model and hydrological model.Precision and high spatial-temporal resolution precipitation data is helpful to carry out researches related to precipitation.This paper chose the Arid and Semiarid Regions of China as the research area.Based on the Merged Precipitation Analysis dataset which developed by National Meteorological Information Center from 2014 to 2016,quantitative accuracy evaluation indexes and monitoring ability evaluation indexes were used to evaluate the accuracy of IMERG V06 precipitation data from different time scales.Using three different statistical downscaling models of multiple stepwise regression,geographically weighted regression and random forest,the downscaled IMERG V06 precipitation data with 1km resolution were obtained by establishing relationships between precipitation and auxiliary factors.Finally,the downscaling models were selected by comparison,which was suitable for Arid and Semiarid Regions of China.The purpose of this study was to explore the adaptability of IMERG V06 precipitation data in Arid and Semiarid Regions of China.This study can also provide data reference and method reference for ecological,hydrological and climatic research in arid and semi-arid regions of China.The results showed that:(1)IMERG V06 data can relatively well reflect the spatial distribution of precipitation in Arid and Semiarid Regions of Chin.Compared with IMERG V05 data,the overestimation and underestimation of IMERG V06 data were improved,but there were still different degrees of overestimation and underestimation.The accuracy of IMERG V06 data in the mid temperate and warm temperate semiarid areas were better than those in the mid temperate and warm temperate arid areas.Compared with the v05 data,the accuracy of IMERG V06 data is significantly improved in the mid temperate and warm temperate semiarid areas,but not in the warm temperate and mid temperate arid areas.(2)The precipitation monitoring ability of IMERG V06 data in the mid temperate and warm temperate semiarid areas were better than those in the mid temperate and warm temperate arid areas.Compared with the v05 data,the precipitation monitoring ability of IMERG V06 data had been improved in arid areas,but the improvement was relatively small.In semiarid area,the improvement of precipitation monitoring ability was not obvious.(3)The accuracy of IMERG V06 data was high from April to September,and need to be improved from October to March of the next year.The accuracy of data was the best in summer,followed by autumn,spring and winter.Compared with V05 data,the accuracy of IMERG V06 data was not significantly improved from October to march of the next year,and there was large uncertainty.The accuracy of other months was improved to varying degrees.In spring,summer and autumn,the accuracy of the data has improved,and the performance was most obvious in summer.However,there was almost no improvement in data accuracy in winter.(4)The monitoring ability of IMERG V06 data in different months has little difference,and the performance was relatively well.Meanwhile,in the Arid and Semiarid Regions of China,there was no significant improvement in the ability of precipitation monitoring in different months.(5)The random forest model and geographically weighted regression model had relatively good downscaled results,which were suitable in Arid and Semiarid Regions of China.