数据精度检验是卫星遥感降水数据应用中的关键问题之一。目前对 T RMM 数据精度检验多以偏差和相关 性分析等方法为主, 对不同雨量等级的估计检验比较缺乏。针对湘江流域, 引入模糊综合评分方法以解决不同雨量 等级降水估计的检验, 同时利用泰森多边形法、 模糊综合评分方法、 相关系数法和散点斜率法, 以研究区内 14 个气 象站点实测降水数据为/ 真值0, 对 TRMM 3B42 降水数据在日、 月尺度分别进行了定性和定量精度检验。结果表 明: 研究区 TRMM 3B42降水数据在月、 日尺度上相关系数分别达到了 0. 93、 0. 48, 前者偏差值 Bias 比后者低 84%, 精度更优; 在不同降水量级的预测中, 精度由好到差依次为小雨、 暴雨、 大雨、 中雨, 其中小雨预报精度最好, 达到良 好水平, 其他雨量级预报精度为中等水平。

The precipitation data predicted by satellite remote sensing technique are widely used, and the accuracy test of precipi 2 tation data is one of the key problems in application. Deviation and correlation analyses are the main methods for the accuracy test of TRMM data, and the accuracy test of different levels of precipitation is lacked. The fuzzy comprehensive evaluation was introduced to perform the accuracy test of different levels of precipitation in the Xiang River Basin. Based on Thiessen polygons method, fuzzy comprehensive evaluation, correlation coefficient method, and scatter points slope method, the accuracy of TRMM 3B42 data at the monthly and daily time scales from 1998 to 2009 was validated using the measured precipitation data at 14 gauge stations. The results showed ( 1) the correlation coefficients of T RMM 3B42 precipitation data reach 0. 93 at monthly time scale and 0. 48 at daily time scale, and the Bias of former data is 84% lower than that of latter data; and ( 2) for different levels of precipitation data, the prediction accuracy of light rain is the best, followed by heavy rain, rainstorm, and moderate rain.

国家高技术研究发展计划( 863 计划) 课题( 2012AA12A309) ;中国水利水电科学研究院科研专项( 减集 1314; 减集1350)