Semi-distributed lumped model of a karst system under active management

In this paper inverse modeling is used to characterize the regime of a karst aquifer subjected to extensive pumping in a conduit located upstream of its main outlet. The systemic approach uses a transfer model that is based on computing the convolution integral of up to several signals, e.g., efficient rainfall, pumping, to simulate flow rates and groundwater levels in both the karst conduit and the carbonate matrix at the aquifer outlet and in several parts of the catchment area. The model is a semi-distributed lumped model which simulates the hydrological response of a heterogeneous karst aquifer made up of different hydrologic compartments, and is applied to the Lez karst system, France. Groundwater is abstracted near the system’s major outlet at a higher rate than the low-water spring discharge, thereby mobilizing stored groundwater during low-water periods (‘active management’). The model’s results are very satisfactory, especially for the karst system outlet, where the water levels are particularly well reproduced. The model can also simulate the natural, i.e., non-pumping, state of the karst system and thereby estimate the impact of active management on the water resource.