基于HASM方法对气候模式气温降水的降尺度研究——以黑河流域为例

Downscaling Simulation of Annual Average Temperature and Precipitation of CMIP5 Outputs by using HASM——A case study in Heihe River basin

基于空间平稳性分析,引入地理因素结合回归分析及高精度曲面建模方法(HASM)对黑河流域多年平均气温、降水给出了降尺度模拟。基于过去器测资料的验证,提出了CMIP5模式资料的合理降尺度方法。比较了降尺度结果与站点实测值的差异,同时比较了所给出的方法与经典插值方法的模拟精度。最后,基于历史时期T1(1976—2005年)的降尺度方法结合RCP2.6、RCP4.5及RCP8.5不同情景下未来时段T2(2011—2040年)、T3(2041—2070年)、T4(2071—2100年)CMIP5模式结果,对降尺度方法进行了修正,给出了未来时段气温的降尺度模拟公式,并基于此对上述3种情景下多年平均气温的CMIP5模拟结果进行了降尺度模拟。结果表明:本文所提出的降尺度方法模拟结果与站点观测值具有较好的相关性,且精度高于其他经典插值方法。对未来时段的模拟结果表明,升温最快的是RCP8.5情景,在2071—2100年,除祁连山地区外,大部分地区年平均气温大于10℃。

Based on the spatial stationarity analysis,this study proposed a new statistical downscaling method with a combination of regression method and high accuracy surface modeling method（HASM）,by considering geographical and topographical factors.The downscaling results were compared with observations from meteorological stations and the results produced from classical interpolators in 1976-2005（T1）.The downscaling formula for future scenarios was given and the downscaling results for future temperature and precipitation under RCP2.6,RCP4.5 and RCP8.5 scenarios in 2011-2040（T2）,2041-2070（T3）,and 2071-2100（T4）were obtained.It showed that the results of the proposed downscaling method were approximate to the real values and the accuracy of the method was higher than other classical methods.Downscaling simulations of the future scenarios indicated that the temperature has the highest rising under RCP8.5 scenario.Except for Qilian mountains,temperature is more than 10 ℃in 2071-2100.