Résumé
This work deals with a rainfall-discharge model applied to a well known karst aquifer. A new approach is developed in order to minimize the fitting parameters: here, some of the model parameters do not result from a simple fitting, as it was the case with earlier models, i.e., some of them were assessed from the hydrograph analysis. The conceptual model of the functioning is based on a production function based on a simple calculation of effective rainfall and a transfer function consisting of two reservoirs. A slow discharge reservoir transfers the low flow and a rapid discharge reservoir feeds the high flow. The model has three fitted parameters plus one for its initialisation. Three parameters are deduced from the hydrograph analysis over the entire time series. For example, the recession coefficient of the slow discharge reservoir is determined from the hydrodynamic analysis of the recession [Mangin, A., 1975. Contribution à l’étude hydrodynamique des aquifères karstiques. 3ème partie. Constitution et fonctionnement des aquifères karstiques. Annales de Spéléologie, 30 (1), 210–124].
This model was tested over a ten years period on the Fontaine de Vaucluse French karst system. The first hydrological year is used for fitting the model; the nine other cycles validate the modelling. The good quality of the model is proved by the Nash criterion of 92.3% on the validation period. Moreover, the simulation results were validated by a statistical analysis of measured and simulated time series.
The model successfully simulates both the high and low flow at the same time. Also it estimates the water volumes available in the different parts of the aquifer and it proposes a management tool capable of predicting the evolution of the discharge in different climate conditions.