Hydrological Process Simulation Of Alpine And Arid Grassland Ecosystem Based On Improved SWAT Model ——A Case Study Of The Upper Bayin River

Grassland is the main land cover type and one of the most important ecosystem types in alpine and arid areas of China.Affected by climate,topography,human activities and other factors,grassland coverage has obvious spatial differences.Grassland with different coverage had different ability of interception and transpiration,which affected regional water cycle.However,the spatial variability of grassland vegetation coverage is seldom considered in the studies of eco-hydrological processes in alpine and arid inland river basins.Bayin River is a typical alpine and arid inland river basin and the fourth largest inland river in Qaidam Basin.The main types of land cover are grassland and wasteland.The spatial heterogeneity of grassland coverage is obvious in this watershed.Hydrological model is an effective means to reproduce and predict eco-hydrological processes at large regional scale.SWAT(Soil and Water Assessment Tool),a large-scale hydrological model with physical basis,has been widely used in the simulation of hydrological processes under different land use,climate and management measures and achieved good results.Leaf area index(LAI)is calculated in the plant growth module of SWAT model,which is used to represent vegetation growth status and vegetation coverage.However,this module only considers the impact of temperature while ignoring precipitation,terrain and human activities when calculating LAI.In addition,in areas lacking meteorological data,LAI calculated by SWAT model could hardly reflect its spatial heterogeneity.Compared with LAI calculated by eco-hydrological model,LAI based on remote sensing has the advantages of easy acquisition,wide coverage and spatial heterogeneity,and is more suitable for expressing vegetation coverage characteristics at regional scale.This study firstly established SWAT models based on CMORPHv1.0(Climate Prediction Center Morphing technique),CHIRPS V2.0(Climate Hazards Group Infrared Precipitation with Stations data),MSWEP V2(Multi-Source Weighted-Ensemble Precipitation)products and measured precipitation with altitude correction.in order to solve the problem of lack of meteorological data in the study area,and evaluated the runoff simulation effect corresponding to each precipitation data.On this basis,the optimal precipitation data suitable for the study area was selected.Secondly,based on GLASS(Global Land Surface Satellite)LAI and Landsat,ESTARFM(Enhanced Spatial and Temporal Adaptive Reflectance Fusion Mode)model was used to obtain LAI with high spatio-temporal resolution(1day/30m)that fit the spatio-temporal scale of SWAT model.And replaced the original LAI estimation module of SWAT model.Finally,the improved SWAT model was used to simulate the eco-hydrological process in the upper reaches of bayin River.The improved SWAT model was calibrated and validated using three kinds of river runoff,sediment content and high spatial and temporal resolution evapotranspiration data,and its simulation effect was compared with that of the original SWAT model.This study mainly draws the following conclusions:(1)The runoff simulation effect of SWAT model corresponding to measured data based on altitude correction is obviously better than that of CMORPHv1.0,CHIRPS V2.0 and MSWEP V2 products.Except CMORPHv1.0,the other three simulation results have certain reliability.Compared with CHIRPS V2.0 and MSWEP V2 precipitation products,the corresponding NSE value of SWAT model based on the measured data of altitude correction is increased by 0.11-0.26,R~2value increased by0.07-0.14,and PBIAS value decreased by 14.40%-24.30%.(2)Compared with the original GLASS LAI,the high spatial-temporal resolution LAI obtained in this study has clearer texture and contour,showing more spatial details.In addition,there is a significant positive correlation with GLASS LAI on a monthly scale,with R~2reaching 0.95.Compared with the LAI calculated by the original SWAT model,the high spatio-temporal resolution LAI obtained in this study have more obvious variability in both space and time,and the average monthly LAI values vary from 0 to 10 over the years.The LAI calculated by the original SWAT model mainly depended on land cover type,and the average monthly LAI values for many years ranged from 0 to 2.(3)The improved SWAT model is superior to original SWAT in simulating monthly runoff.The difference of monthly runoff simulated by the improved SWAT model was between-12.08-8.61m~3/s compared with the measured value,and the error is reduced by-0.37-1.53m~3/s compared with the original SWAT model.The R~2value of the improved SWAT model is greater than 0.90,the NSE value is greater than 0.89,and the PBIAS value is between 0%and 4%during the rate period and validation period.Compared with original SWAT,the mean R~2value and mean NSE value of improved SWAT increased by 0.025,and the mean PBIAS value decreased by 2.85%.(4)The improved SWAT model is superior to the original SWAT in simulating Monthly sediment content.When the simulated value is underestimated,the error of the improved SWAT decreased by 80194.98 t compared with the original SWAT.The mean R~2and NSE values of the improved SWAT are 0.90 and 0.875,respectively,and the PBIAS ranges from-15.5%-30%.The mean R~2,mean NSE and PBIAS of original SWAT are 0.89,0.83 and-16%-36%,respectively.Compared with the latter,the mean R~2value and mean NSE value of the former increased by 0.01 and 0.045 respectively,while the PBIAS value decreased by-0.4%to 6%.(5)The improved SWAT model is superior to the original SWAT in simulating moon ET.The HRU area of improved SWAT with R~2value greater than 0.5 at verification and greater than 0.7 at rate period increased by 9.48%and 226.20 km~2compared with original SWAT,respectively.In rate periodic and validation period,The total area of HRU of improved SWAT with NSE value greater than 0.5 increased by 578.19 km~2compared with original SWAT.The HRU of the improved SWAT model with R~2and NSE values greater than 0.5 and PBIAS between-20%and 20%are 37%and 54%,respectively,increased by 8%compared with the original SWAT model.At the sub-basin scale,12 and 19 sub-basins of improved SWAT met the above three conditions in the rate period and validation period,respectively,2 more than the total number of original SWAT sub-basins.