Projection And Uncertainty Attribution Analysis Of Future Dry-wet Conditions In Northwest China

Since the industrial revolution,population and greenhouse gas emissions have increased significantly,and global average temperatures have generally risen.Due to the intensification of the water cycle driven by global warming,the precipitation and river runoff in Northwest China have increased in recent years,and the drought situation has been alleviated.Due to the increased instability of the climate system,the Northwest climate may change greatly in 21 century.Therefore,the projection of future evapotranspiration and dry-wet conditions in Northwest China is very important for scientifically addressing climate change.In recent years,great progress has been made in the study of potential evapotranspi-ration（PET）,actual evapotranspiration（AET）,and dry-wet conditions under future climate scenarios.Current research usually feeds future climate data simulated by global climate models（GCMs）into potential evapotranspiration models and drought index formulas under future climate scenarios.However,in the calculation process,there are many uncertainties that cannot be completely eliminated.Therefore,it is necessary to use a variety of representative scenarios,models,and GCMs to jointly project,and obtain a data set composed of multiple prediction results and analyze the statistical characteristic.This paper took Northwest China as the study area and used three climate scenarios（SSP1-2.6 scenario,SSP2-4.5 scenario,and SSP5-8.5 scenario）,six well-performing PET models（PETM）,and six global climate models of CMIP6 to characterize the uncertainty from scenarios,models,and GCMs.Then,this paper performs PET prediction based on multiple modes,multiple scenarios,and multiple models,and obtains 108 sets of PET prediction results.Meanwhile,the AET forecast was carried out based on multiple scenarios and multiple models,and 18 sets of AET forecast results were obtained.The future evolution rules of PET and AET and the multi-source uncertainties were analyzed.Based on future PET and AET data,108 sets of future evapotranspiration stress index（ESI）data were obtained,and the change in ESI was used to measure the evolution of wet and dry conditions.The relative importance of each source of uncertainty was analyzed according to Ward hierarchical clustering and ANOVA.The main research contents and results are as follows:（1）Spatial and temporal changes of future PET in Northwest China and the relative importance of the uncertainty sources in PET projectionThe future potential evapotranspiration in Northwest China would increase,and the regional average growth rates in the mid-term future（2041-2070）and long-term future（2071-2100）would be 7.8%and 11.3%,respectively;SSP uncertainty was the most important source of uncertainty affecting the results of PET projection.Specifically,the PET growth rate based on the SSP5-8.5 scenario,the random forest model,and the ACCESS-ESM1-5 were higher,which based on the SSP1-2.6 scenario,PM[CO₂]model and MIROC6 were lower.In the mid-term future,the contribution rate of PETM uncertainty was the highest;while in the long-term future,the contribution rate of SSP uncertainty was the highest,as high as 61.5%.As for seasonal distribution,PETM and SSP had higher contribution rates in warm seasons（summer and autumn）and cold seasons（winter and spring）,respectively.（2）Spatial and temporal changes of future AET in Northwest China and the relative importance of the uncertainty sources in AET projectionThe future AET in Northwest China showed an increasing trend,and the regional average growth rates in the mid-term and long-term future were 18.4%and 24.8%,respectively;GCM uncertainty was the most important source of uncertainty affecting the AET projection results.Specifically,the AET growth rate based on the SSP5-8.5 scenario and the EC-Earth3 climate model were higher,and the AET growth rate based on the SSP1-2.6 scenario,the ACCESS-ESM1-5 climate model,and the GFDL-ESM4 climate model were lower.Besides,the annual distribution of AET increments in the MIROC6model was different from other models.The study also shows that differences between outcomes（both spatially and intra-annually）caused by SSPs increase over time.The relative importance of each source of uncertainty in the AET forecast was ranked by GCM,SSP,and interaction.The contribution rate of GCM uncertainty would be 55.83%and50.38%respectively in the future two periods.The contribution rate of the uncertainty from GCM was 55.83%and 50.38%respectively in the mid-term and long-term future.（3）Future dry-wet variation in Northwest China and the relative importance of the uncertainty sources in ESI projectionThe future ESI in Northwest China would increase,which indicated that the drought would be alleviated,and the GCM uncertainty had the highest relative importance in the ESI projection.Specifically,the aggregate means of the mid-term and long-term future average△ESI（ESI change relative to the historical period）were 0.41 and 0.52,respectively.In the projection,the△ESI based on the random forest model and the EC-Earth3 climate model were higher,and the△ESI based on the PM[CO₂]model and ACCESS-ESM1-5 were lower.Most projection results showed that the△ESI in winter was the highest,followed by spring,which means that the drought conditions would be relieved more obviously in winter and spring.The hierarchical clustering analysis of 108sets of prediction results showed that most of the prediction results were clustered into a sub-cluster according to different GCM types,and only one sub-cluster was dominated by SSP.The quantification results of the contribution rate of uncertainty sources showed that the contribution rate of GCM uncertainty was the highest,with an average value of 53.85%and 49.98%in the mid-term future and long-term future,respectively;The contribution rate of SSP uncertainty was the second,which would be 21.68%and 28.43%respectively in the future two periods.Seasonally,the contribution rates of GCM and SSP increase over time in warm and cold seasons,respectively.