| As an important part of flood control system in Nenjiang River and Songhua River,Nierji Reservoir undertakes the tasks of flood control,urban life and industrial and agricultural water supply in Nenjiang River.Flood is an important manifestation of water resources in the flood season of the basin.Improving the accuracy of flood forecast and extending the flood forecast period can provide important help for flood control decision-making.Therefore,it is of practical significance to carry out the flood simulation and forecast research of Nierji Reservoir in flood season to improve the flood control capacity of the reservoir.Due to the problems of rainfall and obvious heterogeneity of spatial distribution of underlying surface conditions in Nierji Basin,the prediction accuracy of inflow flood of Nierji Reservoir is low and difficult.In view of the above problems,this paper takes Nierji basin as the research object,combined with the characteristics of rainfall and runoff in the basin,XAJ hydrological model is used as the flood forecasting model to explore the applicability of XAJ model in the flood forecasting of Nierji reservoir,analyze the influence of spatial distribution heterogeneity of rainfall and underlying surface conditions on the flood forecasting of Nierji reservoir,and formulate the flood forecasting scheme of Nierji reservoir in different prediction periods.In order to maximize the prediction accuracy of inflow flood of Nierji Reservoir,effectively extend the forecast period of inflow flood,and provide data support for flood control and disaster reduction in Nenjiang River Basin.The main research contents are as follows :(1)In order to construct the XAJ hydrological model suitable for Nierji Basin,this paper analyzes the hydrological characteristics of Nierji Basin.Firstly,the basic hydrological data of Nierji Basin were reviewed.And on this basis,typical floods were selected to determine the flood forecast period.Then,we take the typical floods as an example to analyze the hydrological characteristics such as storm flood,runoff yield and confluence in detail.Then,starting from the actual hydrological and geographical conditions of the basin and the flood control task of the reservoir,the difficulties and requirements of the Nierji inflow flood forecast are analyzed in detail.Finally,according to the actual demand of inflow flood forecasting and the needs of traditional hydrological model evaluation,we determin the evaluation indexes and evaluation criteria suitable for inflow flood forecasting of Nierji Reservoir.(2)In order to explore the applicability of the traditional centralized XAJ hydrological model in flood forecasting of Nierji reservoir,this paper constructs a three-source XAJ hydrological model.And we use the NSGA-II multi-objective genetic algorithm to automatically optimize the model parameters with the relative error of flood peak flow,peak time error,runoff depth relative error and Nash efficiency coefficient as the objective function.The flood inflow of Nierji Reservoir in flood season from 2006 to 2019 is simulated.The results show that it is difficult to reasonably characterize the spatial heterogeneity of underlying surface conditions in Nierji Basin by using a set of parameter prediction models in the whole basin,which leads to poor prediction accuracy of the lumped XAJ hydrological model.(3)In view of the spatial heterogeneity of rainfall and underlying surface conditions in Nierji Basin,this paper constructs a distributed XAJ model.Firstly,divid the Nierji Basin into multiple unit basins,and the surface rainfall of each unit basin is calculated respectively to realize the distribution of model input.Then,construct the XAJ model of each unit,and the model parameters are optimized independently to realize the distribution of model parameters according to the measured runoff data of the corresponding unit basin.Finally,the effects of uneven spatial distribution of rainfall and spatial heterogeneity of underlying surface conditions on Nierji inflow flood forecast are discussed.The results show that after considering the uneven spatial distribution of rainfall and underlying surface conditions,the qualified rates of flood runoff depth prediction in regular and validation periods are 77.8% and 100.0%,respectively.The qualified rates of flood peak prediction are 66.7% and 100.0%,respectively.The average Nash efficiency coefficient is only 0.469 and 0.406.(4)In order to formulate the inflow flood forecasting scheme suitable for Nierji Reservoir,this paper studies the correlation between the measured flow per unit basin and the inflow of Nierji Reservoir,based on the characteristics of different flood propagation time in different unit basins in Nierji Basin.Then,through a special way,the measured flow information of the unit basin is fused with the prediction information of the unit basin,and the flood forecasting scheme of the Nierji reservoir in the 1～7days prediction period is constructed.The improved scheme ensures that the measured flow information of a certain number of unit basins in the 1～5days forecast period is integrated with the flood forecasting information of the lower unit basins.The results show that the improved prediction scheme has better prediction accuracy than the original hydrological prediction scheme when the prediction period is the same.Under the premise of sacrificing some forecast accuracy,the forecast scheme can extend the forecast period to 7days.For the forecast scheme with the forecast period of 1 day,the qualified rate of flood peak flow forecast is 100.0%,the qualified rate of runoff depth forecast is 92.9%,the qualified rate of peak time forecast is 92.9%,the average Nash efficiency coefficient is 0.850,and the accuracy level of forecast scheme reaches grade A.The prediction accuracy of runoff depth and flood peak in 2～7days forecast period is reduced,but it is still above grade B. |
PDF Full Text Request