Study On Hydrological Forecasting And Parameter Regionalization In Ungauged Area Of Guangzhou

The accuracy of flood forecasting is closely related to the integrity of data and the coverage of monitoring equipment in the study area.The area without data in China accounts for a large proportion,and the problem of runoff forecasting in the area without data is also a hot issue.Aiming at the problem of flood forecasting and parameter estimation in the ungauged area of Guangzhou,this thesis is based on the Xin’anjiang model and the Liuxihe model.The Liuxihe Reservoir,Huanglongdai Reservoir,Jiuwantan Reservoir,Helong Reservoir,Fuyuan Reservoir and Furongzhang Reservoir in Guangzhou are taken as the research areas.The forecast results show that the Xin’anjiang model and the Liuxihe model reach the second-level accuracy standard,and the application effect is good.This thesis presents a parameter estimation method based on CART regression tree,and compares the application results of distance method,multiple regression method and parameter average method.Assuming that the six reservoirs in the study area are ungauged areas in turn,the model parameters are estimated based on the basin eigenvalues of the remaining reservoirs,and the CART regression tree method is the best.The Sankeng Reservoir,Tianhu Reservoir and Maodun Reservoir in the ungauged area of Guangzhou City are used to forecast the flood in the flood season of 2021,and good forecast results are obtained.This method has certain reference significance for parameter estimation of Xin ’anjiang model and Liuxihe model in ungauged areas of Guangzhou.(1)In this thesis,based on the three-source Xin ’anjiang model,the flood forecasting of six reservoirs such as Liuxihe Reservoir in Guangzhou City is carried out,and the simulation results all reach the level B accuracy standard.Among them,the qualified rate of Liuxihe Reservoir was 72.4 %,and the certainty coefficient was 0.78.The qualified rate of Huanglongdai reservoir is 80.9 %,and the certainty factor is 0.82.The qualified rate of flood simulation of Fuyuan Reservoir is 68.75 %,and the certainty factor is 0.75.The qualified rate of Helong reservoir is 75 % and the certainty factor is0.78.The qualified rate of Jiuwantan Reservoir is 81.8 % and the certainty factor is 0.82.The qualified rate of Furongzhang Reservoir was 78.6 %,and the certainty coefficient was 0.77.(2)In this thesis,based on the Liuxi River model,the flood forecast of Liuxi River Reservoir and other six reservoirs in Guangzhou is carried out.In the simulation of flood peak flow,the qualification rate of Liuxihe reservoir is 84.6 %,and the certainty factor is 0.86.The qualified rate of Huanglongdai reservoir is 100 % and the certainty factor is 0.85.The qualification rate of Fuyuan Reservoir is 55.6 %,and the certainty factor is 0.76.The qualified rate of Jiuwantan Reservoir is 83.3 %,and the certainty factor is 0.83.The qualified rate of Helong reservoir is 71.4 %,and the certainty factor is 0.79.The qualified rate of Furongzhang reservoir model was 0.875 %,and the certainty coefficient was 0.84.The qualification rate of Fuyuan Reservoir is 55.6 %,and the certainty factor is 0.76.(3)Based on the parameter estimation method of CART,it is assumed that six reservoirs such as Liuxihe Reservoir are ungauged areas,and the parameters are estimated according to the basin eigenvalues of the other five reservoirs.In the parameter estimation of Xin ’anjiang model,the average relative error of CART method is 27.9 %,the distance similarity method is 50.63 %,the parameter average method is109.33 %,and the multiple regression method is 34.04 %.In the parameter estimation of Liuxi River model,the average relative error of CART method is 32.5 %,the distance similarity method is 38.19 %,the parameter average method is 96.42 %,and the multiple regression method is 48.8 %.The CART parameter estimation method is better than the distance similarity method,the parameter average method and the multiple regression method in the common interval.(4)In this thesis,the CART method is used to estimate the parameters of the hydrological model of Sankeng Reservoir,Tianhu Reservoir and Maodun Reservoir in the ungauged area of Guangzhou City,and to simulate the flood season in 2021.In the parameter estimation of Xin ’anjiang model,the deterministic coefficients of Sankeng Reservoir,Tianhu Reservoir and Maodun Reservoir are 0.72,0.78 and 0.83 respectively,and the flood peak simulation errors are 31.4 %,27.4 % and 38.5 % respectively.In the parameter estimation of Liuxihe model,the deterministic coefficients of Sankeng Reservoir,Tianhu Reservoir and Maodun Reservoir are 0.63,0.65 and 0.71 respectively,and the flood peak simulation errors are 59.7 %,39.7 % and 27.1 % respectively.The results show that the CART method has the best application effect in the parameter estimation method,and the Xin ’anjiang model is better than the Liuxihe model.