Développement d'un modèle précipitation-débit géomorphologique pour les zones tempérées : Application au bassin versant de la Chamberonne
The goal of this master thesis was to improve an existing rainfall-runoff model to better represent the dynamics of runoff generation at the catchment scale. The used model is a so-called geomorphological model, which is a spatially explicit hydrological model: it simulates the dominant processes at the hillslope or subcatchment scale and routes the individual contributions along the river network. Within the existing model framework, the main contribution of this master thesis was the implementation of the concept of saturated contributing areas at the subcatchment scale. This concept was implemented through an empirical relationship between soil moisture and the percentage of saturated areas, based on the information contained in the Topographic Wetness Index defined by Beven and Kirkby in TOPMODEL. In addition, the model was extended by including a snow melt and accumulation module and its numerical implementation was improved by implementing a Backward Euler scheme, which overcomes potential problems of poor numerical stability of the original formulation. The model was developed and tested for the Chamberonne catchment. As for any hydrological model, the model calibration procedure represented a key to analyze the model performance and to detect structural deficiencies. Here, the used algorithm was the so-called DREAM algorithm, a stochastic search algorithm that yields optimal parameter sets as well as parameter distributions. The overall hourly model performance for this relatively small catchment is satisfying; the model is able to reproduce the timing and magnitude of the main observed discharge peaks. The obtained results highlighted, however, some model performance problems caused by urban areas. In fact, the model performance is much better for the rural areas than for the entire basin containing significant portions of urban areas. Despite of this, the hourly simulations of the calibrated model clearly highlight the dominance of the dispersion induced at the transition from the hillslopes to the channels in comparison to the geomorphological dispersion induced by the stream network. Accordingly, the newly introduced curve relating the saturated areas to the soil moisture plays an important role in conjunction with the residence times associated with the different types of flow. To further improve the simulation of these hillslope-scale flow generation mechanisms, future research could study the effect of other types of contributing area relations or propose spatially varying relationships at the sub-catchment scale.