ERA5再分析降水数据在长江三角洲的性能评估

采用长江三角洲1961—2018年171个气象站逐日实测降水数据评估ERA5再分析降水数据对雨日的探测能力,分析ERA5模拟降水量的准确性及ERA5对极端降水指标的刻画能力。结果表明：ERA5再分析降水数据能够再现降水过程,有较高的命中率（POD）和公正先兆评分（ETS）,误报率（FAR）较低,在研究区南部的探测性能优于北部,ERA5对小雨为主的雨日数探测偏多17.68%;ERA5高估区域多年平均降水量约15.75%,均方根误差（RMSE）较小,降水量较大时误差也较大,且在南部的误差略高于北部,在月、季、年尺度均表现出较好的线性相关性;ERA5对PRCPTOT、RX5d、R95P、R99P、R20和R25 6个极端降水指标随时间变化表现出先高估后低估的趋势,其中强度指标南部比北部高估得多,频率指标北部比南部高估得多;RX1d、SDII、CDD 3个指标一直被低估,北部比南部低估得多;CWD和R10 2个指标持续被高估,没有明显的南北差异。

Performance Evaluation of ERA5 Reanalysis Precipitation Data in the Yangtze River Delta

Using the daily measured precipitation data of 171 meteorological stations in the Yangtze River Delta from 1961 to 2018, this study evaluated the detection ability of ERA5 reanalysis precipitation data for rainy days, analyzed the accuracy of ERA5 precipitation simulation and the ability of ERA5 to describe extreme precipitation indices. Assessment results showed that ERA5 reanalysis precipitation data could reproduce the precipitation process, with high probability of detection (POD), small equitable threat score (ETS) and low false alarm ratio (FAR). The detection performance in the south of the study area was better than that in the north. The percent of false alarm occurrences for rainy days of ERA5 was about 17.68%, mainly occupied by light rain. ERA5 overestimated the regional multi-year average precipitation by about 15.75%, and the root mean square error (RMSE) was small. The error was also large when the precipitation was large. And the error in the south region was slight higher than that in the north region. ERA5 had a good linear correlation on monthly,seasonal and annual scales. ERA5 showed a trend of overestimation and then underestimation over time for six extreme precipitation indicators such as PRCPTOT, RX5d, R95P, R99P, R20 and R25. And the intensity indices over estimated much higher in the south than those in the north, while the occurrence indices had opposite trend. The three indices RX1d, SDII and CDD had been underestimated, and there was a significantly large underestimation in the north than that in the south. While ERA5 had a continuously overestimation for CWD and R10 with no significant difference between the north and the south.