Document Type
Presentation
Publication Date
8-1-2014
Abstract
Under the background of demand for accurate and reliable flood forecasting, various methodologies are used to model floods. However, not all the phases of the hydrograph can be predicted by any models, even though the global optimum may be reached. In order to exploit the distinct information provided by different models, an ensemble approach is proposed to improve the forecasting accuracy and reliability. The ensemble precipitation estimates from a Weather Research Forecasting (WRF) model were used to as inputs to model the rainfall-runoff process in Taiwan. A Dynamic Evolving Neural-Fuzzy Inference System was applied to combining the predictions of the ensemble members based on the forecasting performance for different water levels of the combined members. Using sophisticated models to address the performance of different runs is shown to be a potential way to improve the accuracy of flood forecasting.
Comments
Session R41, Flood Early Warning Systems