Study On The Short-term Hydrological Forecasting Of Zhexi Watershed

On the basis of the existing hydrologic basic theory and technology, it is necessary to know the distribution regularities of Earth’s water and do research on the impact of human activities on the water cycle system. It has an important impact on the achievement of the sustainable development of water resources. The main task of hydrological forecast is to make full use of the pre-existing hydrometeorological information to predict the future hydrological regime. Precise hydrological forecast is the premis of decision-making of reservoir optimal operation scheme and is closely related to the economic and social development, safety of life and property. In addition, the accuracy of the hydrologic forecast can be verified in a short period. Therefore, we build a hydrological forecasting system for Zhexi watershed and the main research contents and results are as follows:(1) Global optimization algorithms SCE-UA algorithm and MOSCDE algorithm were applied to calibrate hydrological models’ parameters. Taking Xiaoxi-Zhexi watershed as a research area, we established Xin’anjiang model and tank model and used single and double objective parameter optimization methods to study the difference between these methods. The result further verifies the necessity of research on the multi-objective parameter set. Then we selected certain coefficient, flood peak relative error and flood volume error as the objective functions to calibrate the hydrologic model and guarantee the forecast accuracy of flood hygrograph, flood peak and flood volume. The result testifies the reliability and validity of the multi-objective parameter calibration.(2) Multi-objective parameter calibration is usually applied to obtain a a series of Pareto non-inferior solutions.To choose the best paremeter for the series of Pareto non-inferior solutions, we combined the Pareto preference ordering and minimax regret principle to select optimal parameters. This method is not limited by the size of the Pareto non-inferior solutions and can reduce the workload effectively. This method was applied to the Zhexi watershed and the result showes that this method can quickly and effectively choose the best paremeter for the hydrological model.(3) To consider the impact of Xiaoxi Reservoir on runoff prediction, we used BP neural network to build the relationship between inflow, waterlevel and outflow, which can find the regulating law of Xiaoxi reservoir. In addition, peak identification is introduced to establish BP neural network model which is based on peak identification. The result indicates that this model can effectively reduce the peak flood error and improve the forecast accuracy of outflow of Xiaoxi reservoir.(4) Based on these research results, we choosed Xin’anjiang model and Tank model as hydrological forecasting models and used MOSCDE algorithm to calibrate hydrological models’ parameters. Finally we established a hydrological forecasting system for Zhexi watershed. In addition, series and parallel correction methods were applied to realize real-time correction of hydrological forecast for Zhexi watershed.Compared with the previous hydrologic forecasting system of Zhexi watershed, hydrologic forecasting system of Zhexi watershed established in this paper has significantly improved hydrologic forecasting accuracy and has a certain value in engineering applications.