A rolling flood forecast method for river basins with newly-built meteorological and hydrological station network

Abstract

Destructive flash floods occurred more frequently in the small and medium river basins in China recently. However, meteorological and hydrological station networks in such river basins were usually poor. Some of them were newly-built only several years ago so that long-series observations are unavailable; and therefore, it is impossible to gain the most suitable parameters for flood forecast from the historical data directly. This paper developed a rolling flood forecast method for such regions, taking the Leli River basin in Guangxi Province, China, as the study area. The Digital Yellow River Integrated Model (DYRIM) was adopted to simulate the streamflows of the Tianlin hydrological station for each flood during the study period, and the model parameters were rolling optimized in real time as follows. First, the parameters were calibrated with the observed rainfall and streamflow data of the first flood, and they were used to forecast the flood caused by the next rain. Second, when the rain came true, the parameters were modified with the newly-observed rainfall and streamflow data if the simulation result obtained with the parameters of the last flood was not satisfied; and the new parameters would be used to forecast the next flood. Through repeating the above two steps for each flood, the parameters may be optimized constantly; and finally, the value ranges of the parameters could be obtained. From a sample demonstration, it can be concluded that this flood forecast method was feasible; it would be valuable for the flood forecast of river basins with newly-built meteorological and hydrological station network.