Study And Application On Real Time Flood Forecast Method Of Huanren Reservoir Based On Knowledge

Flood forecast, as a vital non-engineering measure for protecting flood and optimal utilizing the quantity and hydropower of water resources, has played an important role in reservoir operation decision making in flood and no-flood seasons. However, following the increasing population shifts and rapid socio-economic development, the construction of medium reservoirs, small reservoirs, and ponds on upstream of river has changed the underlying surface of river basin, which caused the dropping of precision of the existing flood forecast plan. For improving the precision of the current short-term flood forecast method and efficiently guiding the real-time reservoir operation, the paper takes Huanren Reservoir for study area, and studies the human activity imposing on runoff yield via analyzing the variation of runoff coefficients of river basin. The specific objectives of this paper are listed as follows:(1) Based on the characteristic of Huanren River Basin, the three proper runoff forecast models are selected. Then, by performing the credible analysis, the imprecise data are taken out. Finally, the optimization algorithm(GAs) is used for calibrating DHF and Xin'anjiang model parameters, and history floods are tested by the optimized parameters. The results of simulation and testing of the three runoff forecast models show that eligible rate of testing is less than eligible rate of simulation; the volume of forecasted net rainfall is more than the volume of real net rainfall, especially in the flood sample of having lesser rainfall of prophase. In addition, the non-eligible flood sample has seldom no variation and is mainly in the first one.(2) By analyzing the non-expectation of the three proper runoff forecast models, the influence of human activity is considered as the central factor of causing the forecasting results more than the real ones. Hence, the variation of rainfall and runoff are analyzed, and the results show that the decrease of runoff is mainly cased by human activity not by climate change. Then the variation of runoff coefficients involving annual runoff, runoff of flood season and similar flood sample is analyzed, the results illustrate that runoff coefficients are in the decreasing state, the underlying surface is influenced by human activity, which affects the runoff yield and caused the decline of precision of flood forecast.(3) For improving the forecasting precision, by analyzing the influence factors of runoff yield of the sample flood and the degree of impact for runoff, the rule of guiding forecast is developed. Then the model parameters are calibrated again based on the rule, and the new model parameters and rule are applied to test the 13 flood between the year 1995 and 2003. The results show that the precisions of forecasting of every model reach the excellent level.(4) For solving the problem of model parameters calibration relates to multiple models and various manner of data, the system of multiple parameters calibration is designed by using software development technology such as the design of interface universal and object persistence. In the same platform, the system implements many functions which includes multi-database management and maintenance, data processing and analysis, artificial trial and error of muti-forecasting models parameters, and automatic optimization rate and evaluation.Finally,the conclusions were drawn and further research perspective were discussed.