不同源降水数据在天山南坡阿克苏河流域适用性分析

Assessing the applicability of different sources of precipitation data in Aksu River Basin on south slope of Tianshan Mountains

降水是水热循环以及气候变化研究的重要环节,降水资料的准确与否直接影响流域尺度的水文过程研究。本文基于2000—2015年天山南坡阿克苏河流域气象站点观测降水数据,对比分析了Tropical Rainfall Measuring Mission(TRMM)降水数据集和Global Land Data Assimilation System(GLDAS)两种具有代表性的降水格网数据集在阿克苏河流域的适用性。结果表明:TRMM3B43数据在阿克苏河流域的整体表现优于GLDAS-2数据。两种数据的精度在月尺度上表现最优,相关系数分别为0.938和0.901,通过了0.01的显著性检验;在季节尺度,TRMM3B43数据各季节与站点插值的拟合度要优于GLDAS-2数据,但二者均呈现出高估冷季降水而低估暖季降水的趋势;在年尺度上,两种数据表现较差。在空间分布上,两种数据类型均能够反映出阿克苏河流域降水自西北向东南递减的空间分布趋势。并且两种数据在平原区的表现均优于山区,低估高海拔地区降水而高估低海拔地区的降水。

Precipitation is an important part of the hydrothermal cycle and climate change research. The accuracy of the precipitation data will affect directly the study on hydrological process at the watershed scale. This paper takes Aksu River Basin, Xinjiang, China, a typical basin on the southern slope of Tianshan Mountains as the research object, based on the observed precipitation data from 2000 to 2015. Firstly, it was interpolated into the grid data with ANUSPLINE interpolation method; then selecting the typical sites, two kinds of grid data sets corresponding to the precipitation of the site were extracted. Finally, by using Pearson correlation coefficient, root mean square error and relative deviation, the applicability of two representative precipitation data sets （TRMM, Tropical Rainfall Measuring Mission and GLDAS, Global Land Data Assimilation System） in Aksu River Basin is compared. The results show as follows： （ 1 ） The overall performance of the TRMM3B43 data in Aksu River Basin is better than that of GLDAS-2. （2） The accuracy of the two data sets hits their best at the monthly scale, and the correlation coefficients were 0.938 and 0.901 respectively, and it passes the significance test of 0.01; at the seasonal scale, the fitting 0fTRMM3B43 data is better than GLDAS-2 data, the spatial consistency for both data sets was good. But both of them show the tendency of overestimating the precipitation in the cold season and underestimating the precipitation in the warm season; at the annual scale, the two data sets show poor performance. （3） In terms of spatial distribution, two data sets can reflect the precipitation pattern which showed a descending trend from northwest to southeast in Aksu River Basin. The two data sets achieved better results for the plain areas than for mountainous areas. They underestimated the precipitation in high altitude and overestimated precipitation in low altitude. This study is of great significance to understand the spatial distribution of precipitation in the basin, to carry out reasonable water resources allocation and runoff simulation of watershed.