Evaluation of hydrological models for scenario analyses: Signal-to-noise-ratio between scenario effects and model uncertainty

Abstract

Many model applications suffer from the fact that although it is well known that model application implies different sources of uncertainty there is no objective criterion to decide whether a model is suitable for a particular application or not. This paper introduces a comparative index between the uncertainty of a model and the change effects of scenario calculations which enables the modeller to objectively decide about suitability of a model to be applied in scenario analysis studies. The index is called "signal-to-noise- ratio", and it is applied for an exemplary scenario study which was performed within the GLOWA-IMPETUS project in Benin. The conceptual UHF model was applied on the upper Ouémé basin. Although model calibration and validation were successful, uncertainties on model parameters and input data could be identified. Applying the "signal-to-noise-ratio" on regional scale subcatchments of the upper Ouémé comparing water availability indicators for uncertainty studies and scenario analyses the UHP model turned out to be suitable to predict long-term water balances under the present poor data availability and changing environmental conditions in subhumid West Africa.