中文摘要：

分别使用SDSM、SVM、LARS-WG等3种统计降尺度模型对淮河流域蚌埠以上流域在HadCM3模型A2情景下2070-2099年的日降水特征进行了预估,并应用Tebaldi法集成3种模型的结果,量化分析了各月平均日降水与最大日降水变化特征的不确定性.多模型集成的结果显示,平均日降水在总体上呈现出干旱季节增大,而湿润季节减小的趋势,且在干旱季节,平均日降水变化的不确定性更为显著;最大日降水的变化更为一致,除11月与12月外,最大日降水均呈现出较为显著的减小趋势.淮河流域的实例分析表明,Tebaldi法为分析比较降尺度模型的特点、概率预估降水变化以及量化评估因应用不同降尺度模型所带来的不确定性提供了一种灵活有效的途径.

英文摘要：

In this study,SDSM,SVM and LARS-WG are used to project future precipitation in 2070-2099 under A2scenario of HadCM3 in the Huaihe River basin above Bengbu station.Then,a Bayesian multimodel ensemble method proposed by Tebaldi et al.is applied for probabilistic ensemble of results from single downscaling models;and an uncertainty analysis focused on the projected changes of mean daily precipitation and maximum daily precipitation in each month is performed.Results of the multimodel ensemble show that mean daily precipitation is generally projected to increase in dry months and decrease in wet months;and stronger uncertainty have been found in dry months.On the other hand,results of maximum daily precipitation are more consistent.Significant decreasing trend is observed in all the months except November and December.The case study suggests that the Tebaldi method can serve as a flexible and viable tool for comparing downscaling methods,providing probabilistic projections and quantifying uncertainty derived from choosing different downscaling models.