Using typhoon characteristics to improve the long lead-time flood forecasting of a small watershed

This paper describes the development of flood forecasting models to improve the long lead-time flood forecasting performance. To investigate the influence of typhoon characteristics on flood forecasting, four different types of model inputs are designed to yield 1- to 6-h ahead forecasts of runoff. The most appropriate lengths of model inputs, namely rainfall, runoff and typhoon characteristics, are then determined. A performance comparison of models with and without typhoon characteristics is made. In addition to the conventional performance measures, based on the viewpoint of early warning three measures regarding effective warnings are also proposed to evaluate the forecasting performance. With typhoon characteristics as input to the model, the forecasting performance is significantly improved for both medium and long lead-time forecastings, especially when the rainfall data are not available. It is also found that the forecasting model without typhoon characteristics cannot provide any effective warnings for long lead-time forecasting.