Impact Of Two Statistical Downscaling Methods On Projecting Climate Change In Northeast Tibetan Plateau

As climate change intensifies,in recent decades,especially in the last 10 years,from the Arctic to the Tibetan Plateau,from inland river basins to Siberia,the hydrological impact of climate change has been increasingly visible,which has brought great challenges to regional and basin water resources management,and how it will change in the future has been received increasing attention.General circulation models(GCMs)provide valuable information for forecasting future climate change.However,the resolutions of most current climate models are not detailed enough,it is difficult to be directly applied to the assessment of regional or watershed climate change impacts.Therefore,downscaling technology plays a vital role in climate change and hydrological research,among them,statistical downscaling is the most widely used.Numerous studies have used the Delta and automated statistical downscaling(ASD)methods to downscaling the GCMs and drive hydrological models to evaluate the impact of future climate change on hydrology and water resources.However,quantitative research on the uncertainties of projected climate change caused by different downscaling methods is insufficient,and few studies have addressed how the differences in the projected future climate change trends of different downscaling methods.In this study,the Qilian Mountains and the Hexi Corridor,which are sensitive to climate change,are selected to compare and analyze the impact of ASD and Delta methods on projecting future climate change.Firstly,according to the annual precipitation characteristics of 19 meteorological stations,the Qilian Mountains and Hexi Corridor region are divided into three areas: semi-arid region,intermediate region and arid region.And using ASD model and Delta method to downscale the data of four GCMs(bcc-csm-1-1m,Nor ESM1-M,IPSL-CM5A-MR,MRI-CGCM3)to Qilian Mountains and Hexi Corridor region.Subsequently,this paper evaluated the applicability of the two methods in the semi-arid region,intermediate region and arid region and the entire Qilian Mountains and Hexi Corridor region respectively on the daily,monthly and yearly scale.And then this study compared and analyzed the difference between the downscaling results of the two methods,and investigated the impact of the different weights of the ASD and Delta downscaling methods on the project of future climate change trends.Finally,analyzed the projected precipitation and air temperature trends by weighted average of ASD and Delta downscaling methods in Qilian Mountains and Hexi Corridor region from2021 to 2100.The main conclusions are as follows:(1)The construction of ASD model in the Qilian Mountains and Hexi Corridor region suggested that surface upward latent heat flux(hfls)and surface upward sensible heat flux(hfss)being two vital predictors that affect precipitation and air temperature in the Qilian Mountains and Hexi Corridor region.These two factors rarely appear in ASD model in other regions.(2)The difference between the two methods in the Qilian Mountains and Hexi Corridor region was approximately 7-40 mm,and the absolute difference decreased as rainfall decreases,which indicates that the downscaling methods has a big influence in the simulation of precipitation.The precipitation in the Qilian Mountains and the Hexi Corridor region will maintain a steady fluctuation state in the coming decades,which different from the result of a significant increase in precipitation projected in previous studies.This indicates that the optimistic of future precipitation need to be cautious.(3)In term of air temperature,both the ASD model and the Delta method can accurately capture the temperature feature of the observed stations in the Qilian Mountains and the Hexi Corridor region.The projected air temperature by the weighted result of ASD model and Delta method will increase by 0.2 ℃/decade-1 in the forthcoming decades(2021-2100),and it will not exceed 2 ℃ by 2100 s under the RCP4.5 scenario.This result is consistent with previous studies,which indicates that temperature is less affected by different downscaling methods.(4)This study shows that different downscaling methods and different climate region are important sources of uncertainty on projecting future climate change scenarios.The weights of two methods in the three regions indicate that the downscaling results of the ASD model in the area with more precipitation has more uncertainty than the Delta method.For temperature,the ASD model is significantly better than that of the Delta method for the downscaling the temperature in the Qilian Mountains and Hexi Corridor region.