Spatial Downscaling of TRMM Precipitation Data in Karst Mountainous Area
This study obtains high-resolution satellite precipitation data to provide a data base for various research fields in karst regions such as drought and flood disaster assessment and hydrological forecasting.Taking tropical rainfall measuring mission as the data source,using the ordinary least square and geographic weighted regression,this study constructed downscaling model of precipitation with elevation,slope,aspect,longitude,latitude and normalized difference vegetation index for annual downscaling research.And the applicability of OLS and GWR downscaling methods in the karstic mountainous areas of Guizhou province was compared.The results were as follows.1)The accuracy between TRMM data and station observations was good.2)The spatial resolution of the downscaling data was greatly improved to 1 km.The GWR model downscaling annual precipitation was closer to the measured value of meteorological stations than the original TRMM data in most years,and the overestimation phenomenon was ameliorated.Compared with the OLS downscaling annual data,the three indicators of GWR downscaling data performed better.3)At single station,OLS downscaling data tended to have spurious better correlations in cases where elevation and NDVI value changed suddenly.Based on comprehensive evaluation of multiple indicators,the accuracy of GWR downscaling data was generally higher in karst mountainous areas.The subsequent research can get the precipitation data closer with the measured data by dividing vegetation areas and karst areas,adding environmental factor,calibrating downscaling precipitation,and so on.
NETL
NSTL
万方数据
