| Water loss and Soil erosion are serious in the Loess Plateau region.Studying the influence of different factors on the simulation results of hydrological models and using models to study the effects of land use and climate change on regional runoff can provide reference for the scientific deployment and rational use of water resources.Based on this background,this study selected the upper reaches of the Liujia River hydrological station in the upper Beiluo River in the Loess Plateau area as the research area,SWAT distributed model was established,Parameter sensitivity analysis and optimization were conducted to determine the suitability of the model in the study of the basin.The impact of subwatershed numbers and rainfall point distribution on the simulation results of the SWAT model in the basin was investigated,and the runoff forecasting under land use and climate change in the basin was conducted.After research and analysis,the main results were as follows:(1)Different subwatershed numbers was divided based on different catchment area threshold settings.The simulation results of runoff,sediment,and nutrient under different delineation was analyzed,and finally the reasonable subwatersheds number was determined.From the most crude subwatershed delineation to more precise delineation,The changes in runoff,sediment and nutrient load in the upper Liujia river hydrological station above the Beiluo river all showd a significant increase and then decrease,and finally tended to be stable.It showed that the lower subwatershed delineation level was not stable to the model results.Considering the precision and efficiency of the model simulation,it was determined that the reasonable subwatershed delineation level was about 25 in the study area.(2)Based on the ArcSWAT module building project,The basin was divided into 25subwatershed and 377 hydrological response units.And then the total water outlet was determined.A monthly scale SWAT hydrological model was constructed for the study area from 1980 to 1990.The sensitivity of 26 parameters such as CN2(runoff curve number),CANMX(canopy maximum water discharge),SOL_Z(maximum root depth),etc.related to the watershed simulation were obtained by global and one-at-a-time sensitivity analysis methods based on SWAT-CUP software.According to the SUFI-2 algorithm,the results of the monthly calibration of the model parameters:The determinatical coefficient(R~2)was 0.56 and the Nash-Sutcliffe effciency coefficient(NS)was 0.56 during the calibration period,The determinatical coefficient(R~2)was 0.52 and the Nash-Sutcliffe effciency coefficient(NS)was0.38 during the verification period,and both R~2 and NS in the calibration period were greater than 0.5.The calibration results were satisfactory.The verification period R~2 was greater than0.5 and the NS was greater than 0.36.The verification result was acceptable.Therefore,the SWAT model is suitable for the study of watershed simulations.(3)Based on the results of the simulation of 25 rainfall stations,when the density of rainfall stations was 1,5,10,15,20,and 25,the relative error of runoff simulation was relatively small,and the relative error fluctuation did not exceed 23%.The relative error of sediment simulation is relatively large,and the relative error fluctuation amplitude was a maximum of 72%.Rainfall stations were selected at different locations in the upper,middle and lower reaches of the river basin,the maximum relative error of runoff simulation occurred when the rainfall station was selected in the downstream of the river basin,and the simulation result when compared with the density of 25 rainfall stations was 5.4%smaller.The maximum relative error of sediment simulation occurred when the rainfall station was selected upstream of the river basin,and the simulation result was 25%smaller than the density of 25 rainfall stations.Therefore,we should choose more rainfall stations,pay attention to the site distribution in the upper,middle and lower reaches of the river basin,reduce calculation and simulation errors caused by rainfall station distrubution,and improve accuracy.(4)Based on the above studies,land use changes in 1980,1995,and 2000 were analyzed and runoff predictions were made.The area of cultivated land increased,and the area of the forest and grassland decreased,the area of water areas,the urban,rural residential areas and unused land did not change substantially from 1980 to 2000.Using 1980 as a benchmark:from 1980 to 1995,the area of pasture and hay increased the most,making it more effective in maintaining the precipitation,and runoff prediction result was decreased.The area of increased arable land from 1980 to 2000 was larger than the total area reduced from pasture and hay,as a result,the intense evaporation on the surface of the arid and semi-arid areas was enhanced,and the runoff prediction result was reduced.Runoff simulation forecast under different climate scenarios found:Maintaining the same temperature,with the increase of precipitation,and the simulated value of runoff will increase.Under the premise of constant precipitation,with the increase of temperature,the evapotranspiration of the watershed will increase,and the runoff simulation value will reduced.When the precipitation and temperature change slightly,but the increase and decrease trend is opposite,it will cause larger change in runoff.It needs to be pointed out that,although this paper studied the impact of the subwatershed delineation and rainfall point distribution on model simulation results,and on this basis,runoff forecast under land use and climate change was carried out.No further research was done on the simulation accuracy of the model under different time scales and different algorithms.The development of new algorithms to improve the simulation accuracy of the model is the next step to be studied. |
PDF Full Text Request