新一代GPM IMERG卫星遥感降水数据在中国南方地区的精度及水文效用评估

Evaluation on the accuracy and hydrological performance of the latest-generation GPM IMERG product over South China

新一代GPM(Global Precipitation Measurement)IMERG(Integrated Multi-satellit E Retrievals for GPM)卫星遥感反演降水数据产品提供了覆盖范围更大、精度和时空分辨率更高的新的降水数据来源。本文以北江流域为例,基于高精度融合降水数据产品定量评估了新一代准实时"early-run"和"late-run"IMERG产品(IMERG-E和IMERGL)以及非实时后处理的"final-run"IMERG产品(IMERG-F)的精度,并结合VIC(Variable Infiltration Capacity)分布式水文模型评估了该系列产品的水文效用。结果表明:(1)研究区域内非实时IMERG-F产品有着较高的精度,网格尺度上日尺度相关系数为0.65,相对偏差为5.87%,优于上一代3B42-V7产品,而准实时的IMERG-E及IMERG-L产品的日相关系数则为0.6左右,精度同样令人满意;流域平均尺度上各产品精度均进一步提高;(2)情景I(基于站点观测降雨数据率定模型)的水文模拟结果表明,非实时IMERG-F产品表现出较好的水文效用,纳西效率系数为0.622;准实时的IMERG-E及IMERG-L产品虽总体上表现较差,但在汛期表现较好,说明该产品有应用于短期洪水预报的潜力;(3)情景II(基于3B42-V7数据率定模型)的结果表明,各IMERG产品的水文模拟表现均有显著改善,说明3B42历史数据适合在使用IMERG产品进行水文模拟时用于水文模型率定。

Latest-generation GPM （Global Precipitation Measurement） IMERG （Integrated Multi-satellitE Retrievals for GPM） satellite-based precipitation quantitative estimation products provide a new source of precipitation data. This study evaluates the accuracy and performance of the near-real-time ＂early run＂ and ＂late run＂ IMERG products （here after IMERG-E and IMERG-L） and the post-real-time ＂final run＂ IMERG product （here after IMERG-F） and evaluates the hydrological performance of the three IMERG products with the V1C （Variable Infiltration Capacity） distributed hydrological model during 2014-2016 over the Beijiang River Basin, a mid-size basin in South China. The results show that： （1） The post-real-time IMERG-F product presents acceptable accuracy over the study area, with relatively high daily CC （correlation coefficient） of 0.65 at the gridcell scale, relatively low BIAS （relative bias） of 5.87%, while the accu- racy of the near-real-time IMERG-E and IMERG-L products is relatively poorer but is still satisfactory, with CCs of approximately 0.6; performances of the 3 products are apparently improved at the basin scale. （2） Under the scenario I （VIC model calibrated by gauge data） in the hydrological simulation evaluation, post-real-time IMERG-F product presents satisfactory hydrological performance, with high NSCE （Nash-Sutcliffe Coefficient of Efficiency） of 0.622; by comparison, near-real-time 1MERG-E and IMERG-L present poorer hydrological performance, but perform well specially during the flood season （May to September） of 2015, indicating that the potential for the flood forecasting of the near-real-time IMERG products; （3） Under the scenario II （model recalibrated by the TMPA 3B42-V7 data）, hydrological performances of the three IMERG products are all significantly improved, indicating that the 3B42 data is suitable for the model calibration when simulating with the IMERG products.