Dissertation > Astronomy,Earth Sciences > Geophysics > Hydrological Sciences (water sector physics) > Hydrological forecasting > Medium - and long - term runoff forecasting

Research on Mid-and-Long Term Runoff Forecasting Technology and Application System

Author LiuJia
Tutor LiuJianShe
School Donghua University
Course Environmental Science
Keywords mid-and-long term runoff forecasting technology system forecasting system probability forecast Three Gorges Reservoir
CLC P338.2
Type Master's thesis
Year 2011
Downloads 185
Quotes 3
Download Dissertation

The runoff forming process is the fundamental part of the water cycle, and the runoff is the basic element of water balance. Based on the objective laws of hydrology, runoff forecasting is the task to predict the future flows and provide the foundation of the water resource scientific dispatching. The mid-and-long term runoff prediction system can provide future year/month inflow information which is helpful for the water resources planning and deployment. It could also help the authority to be more efficiency to manage water resource while facing benefits conflict among flood control, water utilization, ecosystem and various department needs.(1) Technical system of the mid-and-long term runoff forecasting technology system is established, and it is formatted the technology framework, the basic processes about the "qualitative analysis of the forecast object→determine predictable time→determine predictable factors→optimize the varieties of forecast model and set its parameters→test the performance of the model→probability forecast". Considered the characteristics of flood season and dry season, separate forecast was used to improve forecast accuracy. Based on different forecasting models, the mid-and-long term runoff forecasting system considered the complexity and randomness of various prediction objects. The error analysis and the prediction results accuracy analysis show that the system could effectively improve the runoff prediction precise. Based on the Bayesian analysis, the mid-and-long term runoff probability prediction model was built. Combined with mid-and-long term runoff forecasting model, the model which considered the probability distribution in the form of quantitative hydrological forecasting uncertainty, presents different confidence intervals with corresponding confidence level. It provides the necessary basic information for the decision making of the deployment.(2) Developed the mid-and-long term runoff forecasting system. Firstly, the mid-and-long term runoff forecasting database including many subsystems and data is established, and the database management system (DBMS) uses SQL Server 2000. The data model uses the relational model. A database table framework is const诉讼ructed to achieve the unified information management during the medium and long term runoff forecasting. The analysis information is just like the basic information data, forecast model and its parameters, forecast accuracy evaluation results. It provides data and decision support for better implementation of the mid-and-long term runoff forecasting.Secondly, by MATLAB software, it uses graphical user interface tools (GUI) to design mid-and-long term runoff forecasting system interface, including a database interface, forecast interface, result analysis interface. Database interface is a management platform including the historical data of forecasting object and basic information. By the database interface, the functions, such as information query, adding, import, export, deleting, etc. could be conducted. On the basis of prediction interface, it could be done to qualitative and quantitative analysis for data, and make timely-accurate runoff prediction for the future months and years. The function of prediction results interface, mainly compare and analyze different prediction results, determine the best forecasting model scheme and the Bayesian probability forecast method analyzed the result.(3)The mid-and-long term runoff forecasting system was applied about Three Gorges Reservoir. The mid-and-long term runoff forecasting system is used for the inflow forecasting of Three Gorges Reservoir. The inflow of Three Gorges Reservoir is qualitative analyzed and the forecasting period (the non-flood season) is determined. Forecasting factors are determined by correlation coefficient method. BP neural network model, RBF neural network model, multiple linear regression model and the combined forecasting model are used to predict Three Gorges Reservoir inflow. Based on the accuracy analysis in the prediction results and Bayesian model, the paper analyzes deterministic prediction results and provides a probabilistic forecast results. It provides reliable information for the operation and water management of Three Gorges reservoir.

Related Dissertations
More Dissertations