Effects Of Objective Functions And Temporal Resolutions On Hydrological Modeling

Flood forecasting is a very important non-engineering measure in flood control.Accurate flood forecasting is of vital importance for decision making in flood control operation.In the flood forecasting modelling process,the objective function and time resolution are two key factors which impact the forecasting performance.Thus,the study of model responses to different objective functions and the investigation of the relationship between model parameters and time resolution contribute to the improvement of forecasting efficiency especially for the poorly gauged areas.In this study,the Xin'anjiang model is selected as the flood forecasting model for the Jianxi river basin to examining the effects of objective functions and temporal resolutions on hydrological model parameters and prediction performance,the main contents and conclusions of the study are as follows.(1)The Sobol' method is used to analyze the sensitivity of model parameters to different objective functions and different time resolutions.The results show that parameters KE,SM,KI,KG,CI,CG,N and NK are sensitive with their sensitivitity values and sensitivity order varying with time-steps and objective functions.(2)The SCE-UA algorithm is used to calibrate the XAJ model.The performance of different objective functions and their combinations by using the aggregated distance measure in calibrating flood forecasting model is assessed and compared.The results show that no single objective function could represent all the characteristics of the shape of the hydrograph simulataneously,the mutl-objective functions perform better.The projection pursuit method is used to visually compare the model performance using objective functions with different thresholds of peakflows,and the results show that when the threshold values are between 0.4 and 0.6 of the peakflow value,the models perform better for all muti-objective functions.(3)The Bayesian method and DREAM algorithm are used to derive the posterior distribution of the sensitive parameters with 2-,4-,6-,8-,12-and 24-hourly time-step.The results show that the parameters' posterior distributions change regularly with time-steps,among them the parameters KE,SM,KI,KG,CI and CG present linear relation,the parmaeters N and NK display power functional relation,and the residual model parameter ? shows a logarithmic function relationship.The quantitative relationship between parameters and time resolutions are derived and used to transfer the parameters from one time-step to other time-steps.The results show that the simulations based on the transformed parameters perform well.The parameter transformations are verified to have good capacity for runoff simulation.