GPM卫星降水数据在天山山区的适用性分析

The Applicability Evaluation of Three Satellite Products in Tianshan Mountains

卫星降水产品在山区的适用性具有较高的不确定性。论文基于天山山区2014—2015年167个气象站点数据,应用较为广泛的两套卫星降水产品——TRMM与CMORPH,选用均方根误差（RMSE）、相关系数（R）、相对误差（PB）,以及分类统计分析指标（错报率FAR、探测率POD、公正先兆评分ETS、频率偏差BIAS）等,评估了新一代卫星降水产品——Global Precipitation Measurement（GPM）在天山山区的适用性。结果表明：相对于以上两种产品,GPM在山区的精度最高。具体表现在：1）3套产品在降水较多的夏秋季均表现较好,相对于TRMM与CMORPH,GPM与观测数据的相关系数最高（R≥0.6）,相对误差最小（PB≈10%）;2）在整个天山山区,GPM相对于其他两套产品表现出较低的误差范围（-55%~55%）;3）GPM在不同的高程带内,均表现出同观测站点较低的误差与较高的相关系数;4）3套降水产品均表现出对探测弱降水事件较高的准确性（POD≈0.58）和较低的错误率（FAR≈0.63）,但综合分析4种指数,GPM表现最佳,能够以较准确的精度和较低的误差估测降水系统。

The uncertainty of the satellite rainfall products limits their applicability in mountainous areas. This paper evaluates the applicability of a new rainfall product—Global Precipitation Measurement（GPM） in Tianshan Mountain area based on the meteorological stations data and commonly used precipitation products—TRMM and CMORPH. The data of a total of 167 weather stations from 2014 to 2015 were used to evaluate and compare the accuracies of GPM, TRMM and CMORPH. Root mean square error（RMSE）, Correlation coefficient（R）, Percent bias（PB） and Classification statistical analysis index（FAR, POD, ETS,BIAS） were used to assess the applicability of GPM. The result shows that： GPM has the higher accuracy than the other two commonly used satellite rainfall products. Specifically： 1）Although three data sets perform well in summer and autumn, GPM has the highest correlation（R≥0.6） and the smallest relative error（PB≈10%） between the observed data, compared to TRMM and CMORPH. 2） Throughout the Tianshan Mountains, GPM shows the lower error range（- 55%- 55%）. 3） In different elevation zones, GPM shows the lower relative error and higher correlation coefficient. 4） Three sets of data all show higher accuracy（POD≈0.58） and lower error rate（FAR≈0.63） when detecting the weak precipitation. The comprehensive statistical analysis of four indexes indicates that GPM performs the best and estimates the precipitation with higher accuracy and lower error.