基于多模式耦合的赣江流域设计暴雨估算

Design rainstorm estimation of Ganjiang River Basin based on ensemble of multi-GCM

基于3种不同排放情景下LARS-WG天气发生器内嵌的6种大气环流模式结果,预估了21世纪3个时期赣江流域不同重现期下的设计暴雨值变化。研究表明,使用LARS-WG方法得到的暴雨系列,能得到可信的频率分析结果,基于贝叶斯模型平均方法的多模式耦合能降低预估结果的不确定性。赣江流域绝大多数站点未来设计暴雨值相对基准期减小,且变化幅度随重现期的大小而变化。设计暴雨增加或部分增加和减少的站点分别位于赣江流域上中游和中下游。在未来气候变化影响下,赣江流域按照我国现有规范计算的设计暴雨整体偏大。

The output data of six General Circulation Models under SRA1 B,SRA2 and SRB1 emission scenarios were downscaled by statistical downscaling model（LARS-WG） to predict the change of design rainstorms with different return periods of Ganjiang River Basin during three future periods.The results indicated that the reliable frequency analysis results can be obtained from rainstorm series generated by LARS-WG and the application of Bayesian Model Average can diminish the uncertainty of the estimation.Design rainstorms tend to decrease relative to the baseline period in most stations of the basin and the variation ranges change with return periods.The stations which exhibit overall increase or partial increase of design rainstorm are located in upper and middle reaches,while the rest stations which exhibit decrease are located in middle and lower reaches.Under the impact of future climate change,the design rainstorm in Ganjiang River Basin will be overall overestimated in accordance with our current standard.