Hydrological models are traditionally developed to better describe river discharge dynamics. Model calibration is usually based only on discharge at the catchment outlet, while internal fluxes are often ignored. As a result, in spatially distributed models, inconsistencies may arise between simulated fluxes in different modelling units. When this happens, the model may reproduce discharge correctly but remain an unrealistic representation of the hydrological system. This PhD thesis hypothesises that the physical realism and spatio-temporal robustness of conceptual hydrological models can be improved by explicitly enhancing their spatial consistency. Starting from a lumped modelling framework, this thesis identifies both the need and the strategy to achieve spatial consistency in simulated fluxes. Actual evaporation (AE) and inter-catchment groundwater flows (IGFs) are the most uncertain components in discharge-only calibrations, as errors in one can be compensated by the other to close the water balance. The work therefore focuses on (a) constraining AE through multi-objective calibration, and (b) constraining IGFs by revising model structure and assumptions. Applications are conducted in the Somme catchment, strongly influenced by groundwater processes, and further evaluated in the Meuse catchment.
Shu-Chen Hsu (Fri,) studied this question.