Climate Projection And Runoff Simulation In The Kaidu River Basin Based On Statistical Downscaling-SWAT Hydrological Model

In this paper,the Kaidu River,a typical basin in arid inland areas,is used as a research area.Based on the daily precipitation,the daily maximum temperature,the daily minimum temperature,and NCEP reanalysis data from the four meteorological sites in the Kaidu river basin and surrounding areas from 1961 to 2010.Use the Statistical Downscaling Model?SDSM?and the Automated Statistical Downscaling model?ASD?to downscale the global climate model simulations for large-scale future global climate change.Select the typical representative concentration path emission scenarios?RCP2.6,RCP4.5 and RCP8.5?from the latest the phase 5 of the Coupled Model Intercomprasion Project?CMIP5?of CanESM₂ model.Obtain climate scenarios in the basin and analyze future climate changes.Then apply the SWAT hydrological model to the upper reaches of the basin and complete the calibration of the model,verify model simulation results.Finally,future climate elements under different scenarios are estimated to be the input conditions of the SWAT hydrological model after downscaling.Estimated future changes in runoff of the Kaidu river basin under different climate scenarios.The study initially reached the following conclusions.1.The suitability of the model?1?The precipitation,the maximum temperature and the minimum temperature data were determined and validated by the predictors selected by the ASD model.During the calibration period?1961-1990?,the selected forecasting factor has better ability to explain the maximum temperature and minimum temperature.However,the simulation effect on precipitation is not satisfactory.During the validation period?1991-2010?,the simulated r and RMSE values for precipitation,maximum and minimum temperatures were small,and they all performed best at the Buluntai site.The ASD model has certain applicability in the study area.?2?SDSM model can simulate precipitation and temperature well.The forecasting factors selected in SDSM model can explain the maximum temperature and the minimum temperature during the calibration period,among which Bayinbulak has the best simulation effect on the temperature.For the simulation of precipitation,the Korla site has the best simulation results.During the validation period,the maximum and minimum temperatures can also be better explained,and the simulated effect of temperature is better than that of precipitation.It shows that the SDSM model can simulate precipitation and temperature well.The SDSM model has certain applicability in the study area.?3?Both predictors selected by the two statistical downscaling models have good explanatory power for precipitation.The selected forecasting factors can better explain the maximum and minimum temperatures.However,compared with the two models,in addition to the Bayinbuluke station,the SDSM model was superior to the ASD model at the other sites.During the calibration period,the SDSM model has a good simulation of temperature.During the validation period,the SDSM model simulates precipitation and air temperature with relatively small r and RMSE values and is clearly superior to the ASD model.?4?The SWAT model can better simulate the runoff changes in the Kaidu river basin,and in the study area has a certain degree of applicability.The sensitivity parameters are changed and calibrated.The results demonstrate that the observed runoff and the simulated runoff have a good fitting effect.Nash-Sutcliffe efficiency coefficient?NS?at the calibration and validation period are all greater than 0.57,and the determination coefficient?R²?and relative deviation?RE?can also meet the requirements of the model,indicating that the SWAT model can well simulate the hydrological process of the basin in the study area.2.Forecast of future climate scenariosComprehensive comparison of two models,the accuracy of SDSM model simulation is higher.Then use the SDSM model to estimate the future temperature and precipitation.The results show that in the future three climate scenarios,the predicted precipitation,maximum temperature,and minimum temperature all exhibit the characteristics of RCP8.5>RCP4.5>RCP2.6.From the analysis of precipitation,under the RCP4.5 scenario,the increase was slower?0.84 mm/10a?during the period of 2061-2100,showing a significant decline.In other climate scenarios,the prediction of precipitation shows fluctuant.From the analysis of the maximum temperature and the minimum temperature,under the RCP2.6 scenario,during the period 0f 2061-2100 may decrease by 0.09°C/10a and 0.04°C/10a respectively.The lowest temperature in the RCP4.5 scenario may decline to 0.07°C/10a during the2061-2100.In the other climate scenarios,the temperature has shown an upward trend.3.Future runoff prediction simulationThe SDSM model was used to simulate the temperature and precipitation data of different future scenarios in the Kaidu River Basin.The input data of the SWAT hydrological model was used to analyze the future runoff changes in Kaidu River Basin under different climate scenarios.The results show that in the future three climate scenarios,compared with the baseline period,the annual runoff in the RCP2.6 scenario has decreased by 5.14%,and the RCP4.5 scenario has not changed significantly to 0.022%,and RCP8.5 scenario estimates increased by 10.74%.