The Design And Development Of System Based On Mid-and-long Term Runoff Forecasting At The Three Gorges

Hydrological system which is always influenced by climate, geographical conditions and human activity is a complicated and large-scale nonlinear system, and its change law is also very complicated with lots of uncertainty, randomness, grey, fuzziness and chaos. Runoff prediction especially of the long-term runoff prediction for this system which can make policy-makers to give an overall consideration as soon as possible, is the key link to maximum the beneficial result when conflicts arise between different water-use sectors such as flood control and drought, water stored and wasted and so on.This study firstly gave a detailed review on the current situation and development trend of the long-term runoff prediction and its system software both at home and abroad, and then a new ensemble runoff prediction method based on wave analysis was proposed in this paper. Several models of long-term run-off forecasting are considered, namely Multi-variable linear regression model (MLR), Auto-regressive model (AR), Nearest neighbor bootstrap regressive model (NNBR), Support vector machine model (SVM), Wavelet analysis (WA). Firstly, the original annual runoff series is decomposed into different component sequences with WA, and then the MLR, AR, NNBR and SVM models are used to get the forecast results of different component sequences, at last they are reconstructed with WA again. The observed discharges getting from Yichang station, Yangtze River basin, are used to test the performance of the proposed method. The results confirm that better discharge estimates can be obtained by the multi-model approach based on wavelet analysis compared with the single models.To making the results of this research can be applied to the practical operation of the Three Gorges reservoir, a long-term runoff forecasting system software with a system integration of hydrological information inquiry function, data retrieval and management function and long-term runoff prediction function. This runoff forecasting system software is powerful, user-friendly interface and easy to operate. The entire system integrates hydrological information query and management, hydrological forecasting and hydrological forecasting model base-specific database module, and several other main parts. Hydrological information query and management module enables real-time hydrological data and information inquiry function, and these data into a hydrological forecasting dedicated database for fast retrieval and real-time hydrological forecasting hydrological models provide a powerful data support. Hydrological statistics and analysis module can analyze the frequency and statistics characteristic of yearly and monthly runoff. Long-term runoff prediction module contains several runoff forecasting models, such as WAR model, WNNBR model, WAR-SVM model, WNNBR-SVM model, and so on. The forecasting precision of this module can be observably raised, because the integration of Wavelet analysis method and traditional methods which can better reflects the runoff change characteristics.