About project
Title: Hybrid flow forecasting models
Project ID: APVV - 0443-07
Acronym of the project: HYBRID
Duration of the project: 06/2008 - 06/2011
Key words: hydrological forecasting, hybrid models, flow routing, nonlinear time series models, regime-switching models, regional estimation
Project abstract: Hydrological forecasting for integrated management of waters and protection against hydrological extremes needs to meet increasing accuracy standards and requires uncertainty estimates of the forecast. The project aims to approach flow forecasting with a hybrid modeling approach by decomposing the problem into component processes and using process physics models where appropriate and data driven models (time series models and methods of hydro-informatics), where these are best suited. Such an approach will be used with different time steps for typical scenarios in the forecasting of inputs into rainfall-runoff models and flow routing models and for forecasting of the errors of these. Nonlinear time series models including regime switching models, artificial neural networks and machine learning approaches will be developed for the data driven component of the hybrid modeling framework. Further methods of regional transfer of hydrological model parameters will be investigated for the process models in order to enable forecasting in ungauged basins.