近16年秦巴山区TRMM降水资料的降尺度研究

Statistical Downscaling of TRMM Precipitation Data in Qinling-Daba Mountains Area in Recent 16 Years

为获得秦巴山区高分辨率TRMM数据,采用近16年TRMM 3B43降水资料和MODIS NDVI植被数据,结合地理加权回归模型(GWR),采用统计降尺度方法获得高空间分辨率的降水数据。结果表明:TRMM 3B43资料在秦巴山区具有较好的实用性,其与站点实测数据的相对偏差最大仅为0.13%,相关系数最小也达0.72。整体而言,GWR模型呈现出秦巴山区降水由南到北逐渐减少的趋势,与实测值相关性较好(相关系数为0.71~0.90);尽管较原始TRMM数据相比,GWR降尺度结果整体偏低,但就多年平均而言,后者明显优于前者(相关系数达0.92,通过0.001信度检验)。因此,利用GWR模型开展秦巴山区TRMM 3B43降水资料的降尺度研究具有一定可信度。

In order to obtain high resolution TRMM data from Qinling-Daba Mountains,based on TRMM 3B43 precipitation data and MODIS NDVI vegetation data in the past 16 years,and combined with geographic weighted regression model(GWR）,Qinling-Daba Mountains region was used as the object study in this paper,and the precipitation data with high spatial resolution were obtained by statistical downscaling method.The results are as follows：(1） TRMM 3B43 data has good practicability in Qinling-Daba Mountains.The maximum Bias is 0.13% and the minimum R is 0.72;(2） the GWR model shows that a trend of precipitation decrease from south to north,and a good correlation with the measured values(R is 0.71-0.90）;(3） Compared with the original TRMM data,the GWR downscaled results are lower as a whole,but the latter is better than the former in terms of multi-year average obviously(R is 0.92,P 0.001）.Therefore,using GWR model for studying downscaling TRMM 3 B43 data has some credibility in Qinling-Daba Mountains.