Predigested Tank Model and Its Application in 10 days Averaged Inflow Forecast of the Longyangxia Reservoir during Raining Seasons
HU Xing lin (Gansu Hydrology and Water Resource Bureau, Lanzhou Gansu 730000, China)
The Longyangxia Reservoir is the biggest reservoir in the upper reaches of the Yellow River. It is located the eastern Tibetan Plateau. After the reservoir being completed a long-period drought occurs in the Tibetan Plateau. Less inflow comes into the reservoir. According to the flood control program, however, it is necessary to drain much water in the flood season for the reservoir security. In one aspect, managers must assure enough water after flood season to produce electricity. And in another aspect, they must assure the reservoir security. As the future rainfall and water income is unknown, the reservoir should not store much water in the flood control stage. Therefore a very high accuracy of long-and-middle inflow forecast is urgent. Tank model is a certainty mathematics model, which is used to forecast runoff in a basin. In this model only consider the input and output process, ignoring the physical process of basin water transformation. So for a very large basin with a complicated hydrologic and climate situation, this model can be well applied for flood control. The Tangnag Gauging Station is located in the upper reaches of the Yellow River, gauging the main inflow to the Longyangxia Reservoir. According to the underlying surface condition and runoff yield characteristics of the working basin, three independent runoff yield areas can be divided from the working basin. Each runoff yield area is predigested to be a single opening linear water tank, of which the input is precipitation and the output is runoff. According to the water balance theory in flood period a model of calculating the runoff in flood period is worked out. Combining the three runoff yield areas, the runoff averaged over a flood period can be yielded. Thus, a medium-term runoff forecast model was put forward. Optimum seeking method is applied, and forecast parameter is determined by regression analysis. This model can be used to predict the mean inflow to the Longyangxia Reservoir for a period of ten days. Through fitting history data and testing the forecast it is found that the model is of a satisfying precision. The model calibration accuracy reaches to 95% or more. An actual forecast applied in July~September 1999 showed that in this model the maximum prediction error was 15 2%, and the minimum one was 3.7%. The error is lower than the allowance, 20%, of the Standard 《Hydrologic Information Forecast Code》.