Real-Time Flood Forecasting System: Case Study of Hsia-Yun Watershed, Taiwan

Abstract

Typhoons and heavy storms frequently cause severe inundation in Taiwan. For early flood warning purposes, a real-time flood forecasting system was acknowledged as a pivotal tool and developed in this study. The system integrates four parts: (1) a gray-based rainfall prediction model, (2) the antecedent hydrological condition estimation method, (3) a deterministic runoff model, and (4) a runoff updating algorithm. The developed system was applied to the Hsia-Yun watershed in Taiwan. The time-invariant parameters, which characterize terrain features, were calibrated in advance. Additionally the time-variant parameters were modeled to change with temporal hydrological condition of the watershed. The gray theory was performed to predict the rainfall intensity in the next 1–3 h, which thus enables the runoff model to forecast incoming runoff in the next 1–3 h. By consulting the real-time feedback flow observation, a runoff updating algorithm was adopted and was helpful in improving the performance of the simulated hydrographs. In real-time application, the system can be initiated or terminated by detecting if the cumulative rainfall depth is larger or less than a predefined threshold value. The idea of the developed system has recently been constructed with a visual interface for real-time flood forecasting by authorities. The overall results confirm that the developed system is competent to forecast flash floods.