基于QDM方法的江淮流域极端降水模拟能力评估

Performance Evaluation of Quantile Delta Mapping for Extreme Precipitation Simulation Over the Yangtze-Huaihe River Basin

本文利用江淮流域1961—2005年76站夏季逐日降水观测资料和BCC-CSM1.1(m)模式模拟数据,构建基于经验和理论分布函数的分位数增量映射(Quantile Delta Mapping,QDM)偏差订正模型eQDM、dQDM,并评估其对江淮流域逐日降水及极端降水的订正效果。研究结果表明:相较于理论分布,基于经验分布的eQDM方法显著改善模式对观测降水的概率分布和空间分布特征的模拟能力;除连续干日订正效果欠佳外,其余四个极端降水指数的均方根误差明显降低,其气候态基本与观测一致,大部分站点的相对误差保持在-20%~20%。

The observed summer daily precipitation in 76 stations over the Yangtze-Huaihe River Basin(1961-2005)and the simulation results of BCC-CSM1.1(m)model in the same period are used.Quantile deltamapping(QDM)bias correction method based on empirical and theoretical distribution functions(expressedas eQDM and dQDM)are introduced to evaluate their correction effects on daily precipitation,especiallyextreme precipitation over the Yangtze-Huaihe River Basin.The results showed that compared with thedQDM,the eQDM significantly improved the model's capacity to simulate the probability distribution andspatial distribution characteristics of observed precipitation.In addition to the poor performance for thecontinuous dry days,the root-mean-square error(RMSE)of the other four extreme precipitation indexeswere significantly reduced,and the climate state were basically consistent with observation with the relativeerror from-20%to 20%at most of stations.