Abstract:
A bivariate model for real-time flood forecasting is developed from Burg's multivariate channel analysis. The model may be applied to any bivariate hydrologic processes, provided that there exists cross-correlation between processes. The model was tested for short-term flood forecasting for various rainfall-governed events from five climatologically different watersheds. The model is particularly suited for flood forecasting in the data-scarce regions, since it does not require prior parameter estimation.
