Improvement and comparative assessment of a new hydrological modelling approach to catchments in Africa and the USA

Abstract

Assessing water resources is still an important issue especially in the context of climatic changes. Although numerous hydrological models exist, new approaches are still under investigation. In this context, we investigate a new modelling approach based on the Physics Principle of Least Action. A first version of a Least Action based hydrological model, in its deterministic version has already given very good results in simulating the Bétérou catchment in the Ouémé basin, Benin. The thesis presents new hypotheses to go further in the model development with a view of widening its application. The improved version of the model MODHYPMA was applied on 21 subcatchments in Africa, in Bénin, Côte d’Ivoire, Ethiopia; and in the USA. Its performance was compared to two well-known lumped conceptual models, the GR4J and HBV models. The model was successfully calibrated and validated; it showed a good performance in most catchments. The analysis revealed that the three models have similar performance and timing errors. But in contrary to other models used in this study, MODHYMA is subject to a less loss of performance from calibration to validation. The parameter uncertainties were analysed using the GLUE methodology. It is concluded that model uncertainties are higher during high flows and that uncertainty analysis should include the uncertainty of the discharge data. In order to explore the possible transferability of our model for ungauged basins studies, we then intended to investigate how model parameters could be estimated from the physical catchments characteristics. We relied on statistical methods applied on calibrated model parameters to deduce relationships between parameters and physical catchments characteristics. These relationships were further tested and successfully validated on gauged basins that were considered ungauged.