中国融合降水产品在太行山区的质量评估

山区地形复杂,气象站点稀少导致降水实测资料匮乏,卫星降水产品是实测资料的有效补充。以太行山区为研究区,采用国家气象站点观测降水为基准数据,通过连续评价指标(R、RMSE、RB)和分类统计指标(POD、FAR、CSI)对国家气象信息中心研制的中国区融合降水产品进行质量评估,并探讨海拔、坡度和坡向与产品表现的关系。结果表明:融合降水产品与地面实测数据之间有着良好的相关性(R>0.7),但相比原始产品相对偏差和均方根误差仍较大。在降水探测能力上,融合产品提升明显,全序列站点日降水命中率达到0.65,误报率为0. 33,特别是对夏季降水的探测精度更高(POD=0.76,FAR=0. 32,CSI=0. 56)。海拔对融合前后的产品精度有显著(p <0. 05)的影响,回归分析显示,在20°～30°的坡度范围存在产品表现的拐点,融合前后降水产品均在南坡表现较差。融合降水产品在经过地面观测数据的校正之后,降水产品精度和降水捕捉能力都有显著的提升,能够较为准确地反映太行山区的降水分布特征。

Quality assessment of Chinese merged precipitation product over Taihang Mountains

The complex terrain and sparse weather stations lead to limited precipitation measurement materials in mountainous areas, and satellite precipitation monitoring is an effective supplement for precipitation information. Selecting the Taihang Mountains as the study areas and taking precipitation data from rain gauges of the national meteorological stations as the reference, the quality of merged percipitation monitoring product developed by national meteorological information center of China were evaluated using the continuous statistical metrics (R, RMSE and RB) and categorical statistical metrics (POD, FAR and CSI), and the relationship of elevation, slope and slope direction with product performance was also discussed. The result showed that the merged product had a high correlation (R>0.7) with the measured data; however, it had higher RB and RMSE compared with non-merged product. Meanwhile, the merged product showed a significant improvement in precipitation detection capability. The POD of whole-sequence diurnal scale reached 0.65 and the FAR was 0.33. Particularly, merged product had better performance in detecting summer precipitation (POD=0.76, FAR=0.32, CSI=0.56). Elevation had a significant (p<0.05) influence on the accuracy and precipitation detection ability for both products. The regression analysis showed that there was a performance turning point in the slope range of 20° to 30°. Both merged and non-merged products performecl poorly on the south slope. In general, after the calibration of rain gauge data, the accuracy and precipitation detection ability of the merged product have been significantly improved, which can reflect the precipitation distribution characteristics of Taihang Mountains correctly.