不同降水产品在长江流域的偏差校正研究

Bias Correction of Different Precipitation Products in the Yangtze River Basin

基于长江流域135个站点1983-2020年的月降水量观测数据,分别对ERA5-Land、PERSIANN-CDR和M SWEP等三种降水产品进行了适用性评估,并对比使用了多种偏差校正方法来提高降水产品的观测精度。结果表明,MSWEP对于长江流域的降水模拟效果较好,空间相关系数高达0.91,相对偏差仅为1.64%;ERA5-Land则对长江流域降水高估严重,精度较差。经偏差校正处理后,三种降水产品的适用性均得到了不同程度的提升,尤其是ERA5-Land,KGE(Kling-Gupta Efficiency)指标值从校正前的0.72增加到了0.82。对比不同偏差校正方法的应用效果发现,线性缩放法和分段伽玛分布校正法在长江流域表现最优,且后者更加稳健。本研究可为降水产品及气候模式输出等的后处理工作提供指导。

Based on the monthly observations of 135 stations in the Yangtze River basin during 1983-2020,this study evaluated the performance of three precipitation products,ERA5-Land,PERSIANN-CDR,and MSWEP respectively.Furthermore,bias correction methods were used to improve the quality of the precipitation products.The results reveal that MSWEP shows a good performance in the Yangtze River Basin,with a spatial correlation coefficient of 0.91 and relative bias of 1.64%,while ERA5-Land overestimates precipitation seriously.After bias corrections,the performance of all three precipitation products has significantly improved,especially for ERA5-Land,with the KGE increasing from 0.72 to 0.82.Comparing the different bias correction methods,it is found that linear scaling and separated three Gamma distributions perform best,and the latter is more robust.This study can provide support for the post-processing of precipitation products and climate model outputs.