ABSTRACT
Drainage basins in many parts of the world are ungaged or poorly gaged. Moreover, some existing stream-gage networks are in a state of decline due to cost issues or maintenance diffi culties. Note also that the behavior of drainage basins is often nonstationary, modifi ed directly or indirectly by a wide range of human activities and/or development projects. Under such conditions, the estimation of hydrological variables is diffi cult and highly uncertain. Prediction of these variables is however essential for effective management and use of water resources, whether for the design of hydraulic works or the identifi cation of hydrological risks related to extreme events (e.g., fl oods, low fl ows). Real-time forecasting of these hydrological variables is another equally crucial task in hydrology.
