Study On The Uncertainty Of Runoff Simulation Driven By Multi-source Data

In recent years,climate change and human activities have destroyed the hydrological consistency of the watershed.The hydrological series show the characteristics of nonlinearity,weak correlation and non-stationary,which bring great uncertainty to the runoff simulation.At the same time,the complex correlation of parameters and the uncertainty of objective function will also affect the runoff simulation.The diagnosis of hydrological variability and quantitative analysis of uncertainty in runoff simulation are helpful to improve the accuracy of hydrological simulation,which has great significance for runoff forecast and water resources planning and management.Aiming at the uncertainty of runoff simulation process,this study analysed sensitivity of parameter and did runoff simulation using different objective functions.We also compared the performances of GPM precipitation data and CMFD precipitation data with observed precipitation in runoff simulation to explore the impact of precipitation input on runoff simulation.The main content and results of the paper are as follows:(1)Trend and variability analysis of precipitation and runoff.The trend test and variation diagnosis of precipitation and runoff series in Luanhe River Basin are carried out by using several parameter test and non-parameter test methods.The annual distribution of precipitation and runoff is uneven.And the concentration period is from July to August.The decreasing trend of precipitation is not significant,while the runoff shows a significant downward trend.The variation points of precipitation and runoff series are consistent in 1979 and 1998.(2)The uncertainty analysis in the process of parameter calibration.We established the SWAT model of Panjiakou Reservoir upstream in Luanhe River Basin.Four different types of objective functions(including NSE,b R~2,PBIAS and KGE)were selected for model calibration and verification.It is found that the parameter sensitivity will change with the increase of iteration times and the change of sampling range.The parameter ranges and optimal values obtained by calibration are significantly different under different objective functions.As the objective function,NSE and KGE are more robust in runoff deterministic and uncertain simulation of each station,showing higher simulation accuracy and lower simulation uncertainty respectively.(3)Comparision and analysis of runoff simulation accuracy under different precipitation inputs.By evaluating the accuracy through muli index,GPM data and CMFD data have high correlation with observed precipitation.However,both of them overestimate the precipitation by 3.7%and 2.6%respectively.In general,the two precipitation products can be applied to runoff simulation in Luanhe River Basin.The GPM data can effectively improve the runoff simulation accuracy of sparse watershed with rainfall stations and perform well in the runoff uncertainty simulation.Therefore GPM data has a good application prospect in the runoff simulation,especially in the areas which are lack of precipitation data and have complex terrain.