TRMM卫星降水数据在天山山区的校正方法研究

Correction Method of TRMM Satellite Precipitation Data in Tianshan Mountains

本文基于全球降水观测(Global Precipitation Measurement,GPM)及天山山区40个气象站点数据,提出了一种新的热带降水测量任务(Tropical Rainfall Measuring Mission,TRMM)卫星数据误差校正方法(MERGE),并与广泛使用的地理差异分析(Geographical Different Analysis,GDA)方法进行对比,通过交叉验证来评价两种方法对山区年/月降水的校正效果。结果表明,MERGE方法不仅能提高TRMM的空间分辨率(0.05°),且在年(平均Bias降低了21%)、月(平均Bias降低了37%)尺度上的校正结果都优于GDA方法;校正后TRMM降水的误差在不同月份存在很大差异,即在降水较多的雨季(4—10月)误差较大(MAE>3mm),而在干季(11月至次年3月)误差较小(MAE<3mm)。MERGE方法不仅能有效改善高海拔地区的降水误差(平均误差降低了59%),且对原始TRMM数据的依赖性较小(R2=0.47),即使在原始TRMM数据表现不佳的局部区域,也能有效降低其误差。

Based on Global Precipitation Measurement（GPM）and 40 weather stations in Tianshan Mountains,a new error correction method（MERGE）is proposed and compared with the widely used geographical different analysis（GDA）method through cross-validation.The results indicate that both GDA and MERGE methods can improve the accuracy of Tropical Rainfall Measuring Mission（TRMM）significantly.The MERGE method not only improve the spatial resolution of TRMM,but also has higher accuracy than GDA method on annual and monthly scales.The error of corrected TRMM data varies greatly in different months.It has larger error（MAE〉3 mm）in the rainy season（April-October）,but less error（MAE〈3 mm）in the dry season（November-March）.Besides,the MERGE method can effectively improve the error in high altitude area and less depends on the original TRMM data,with the R2=0.47.So,even in local areas where the TRMM data have poor performance,this method can also reduce the error effectively.