On the models of real-time forecasting for flood drainage in a low-lying area
2000
Cao, F. (Niigata Univ. (Japan)) | Toyota, M. | Misawa, S.
For establishing a quick and accurate forecasting system for flood drainage in low-lying areas, it is necessary to progressively improve runoff forecasting precision. The choice of runoff model will influence runoff forecasting precision directly. This paper discussed three types of flood models: the Parallel Tank Model (PTM); the Series Tank Model (STM); and the Back-Water Tank Model (BWTM). Filtering forecast-systems were constructed and were operated on the Kalman Filter technique. Comparisons for forecasting the water level in a pump-station up to 3 hours ahead were performed using the floods of August, 1995 and June, 1978 in the Kyukiyamakawa, the Imai, and the Odoorikawa basin in Niigata Prefecture, Japan. The forecasting precision of each model was discussed for the complete flood hydrograph and for the flood peak. The results show that the BWTM flood forecast accuracy was higher than that for the PTM and the STM, particularly for 3 hours ahead. Furthermore, in analyzing the structure of the models, the following result was obtained. The BWTM method can reflect the characteristics of drainage holdback correctly for flood drainage in a low-lying area. Using the BWTM method results in a high degree of forecasting accuracy. Also, the BWTM method has the function of change-over holdback, in the gravitational drainage application, which aids high accuracy
Показать больше [+] Меньше [-]