基于ATAK的MSWEP数据空间降尺度及对降水融合的影响研究

Spatial downscaling of MSWEP dataset based on ATAK and its influence on rainfall merging

全球性降水数据为获取大范围降水空间分布提供了新途径,但其空间分辨率不高一直是制约其应用于流域或区域尺度上的重要因素之一,因此研究全球性降水数据的空间降尺度方法具有重要的理论和实用价值。本文采用从区域到区域的Kriging(Area to Area Kriging,ATAK)和反距离权重(Inverse Distance Weighted,IDW)两种方法,不考虑地面雨量资料及影响雨量的有关辅助信息,在汉江流域将全球性降水数据MSWEP的空间分辨率由0.1°×0.1°提高至0.02°×0.02°。结果发现ATAK降尺度得到的月雨量场虽然在统计精度上与IDW无明显差异,但提高了对月降水量局部空间变异特征的描述能力,在一定程度上克服了IDW的平滑效应。进一步以ATAK、IDW降尺度处理后的MSWEP数据以及不作空间降尺度处理的原始MSWEP数据为背景场,采用GWR方法分别与雨量站网降水数据融合,发现3种情况下得到的月降水融合数据在空间基本格局上相同,精度统计结果也较为接近,但雨量场的空间连续性及细节特征仍有一定差异。在地表雨量站网密度较高的情况下,背景场差异对MSWEP和站点降水融合结果的影响不能完全消除,甚至在局部可能放大。因此,对于MSWEP等全球性降水数据与站网降水资料的融合而言,选择适当的空间降尺度方法是必要的。本文的结论和认识为全球性降水数据的空间降尺度和雨量场精细化估计提供了重要参考。

Global precipitation datasets provide a new way to obtain large-scale precipitation spatial distribution,but the low spatial resolution has always been one of the main gaps restricting their applications at the basin or regional scale.Therefore,spatial downscaling method for global precipitation datasets is of significant value both in theory and practice.Using two statistical methods including Area to Area Kriging(ATAK)and Inverse Distance Weighted(IDW),Multi-Source Weighted-Ensemble Precipitation(MSWEP)gridded rainfall is spatially downscaled from 0.1°×0.1°to 0.02°×0.02°in the Hanjiang River Basin,without considering the ground rainfall data and any auxiliary information.The results show that although the monthly rainfall fields obtained by ATAK downscaling is not rather different from IDW in terms of statistical accuracy indices,however,they improve the ability to describe the local variation of monthly rainfall and overcome the smoothing effect of IDW to a certain extent.Furthermore,through Geographically Weighted Regression(GWR),the disaggregated MSWEP data by ATAK and IDW and the raw MSWEP data without spatial downscaling are used as the background fields to merge the gauges observed rainfall respectively.It is found that the monthly fused precipitation obtained using the three different kinds of background fields are generally similar in spatial patterns,and the accuracy statistics are also quite close.However,among the three kinds of merged precipitation datasets,there are still some differences in the spatial continuity and details.In the case of dense ground rainfall gauges within the study area,the influence of background field differences on the merging results of MSWEP and gauge observed precipitation cannot be completely eliminated,and even may be amplified locally.Therefore,strengthen the comparison of spatial downscaling methods for global precipitation data and choosing the appropriate method is vital.These conclusions provide important references for the spatial downscaling of global precipitation datasets and accurate estimation of rainfall fields.