Karst recharge-discharge semi distributed model to assess spatial variability of flows

Abstract

Aquifer recharge assessment is a key factor for sustainable groundwater resource management. Although main factors of the spatial and temporal variability of recharge are known, taking them into account in a distributed or semi-distributed model is still a challenging task. This difficulty is increased in karst environments. Indeed, recharge of karst aquifers also depends on the organization of the karst network, which is both highly heterogeneous and difficult to characterize.We developed a reservoir model to simulate the spatial and temporal variability of recharge on karst watersheds. Special attention was paid to the link between model parameters and measurable or qualitative environmental factors of recharge. The spatial variability of soil reservoir capacity was estimated by multifactorial modelling (neural network). Intrinsic vulnerability indices were used to constrain the partitioning between slow and fast flows within the karst aquifer. Comparison of simulated and measured discharge at the outlet was used to calibrate and assess recharge model.The karst hydrosystem of the Fontaine de Vaucluse is renowned for its significant heterogeneity and anisotropy, which has so far limited the application of 2D or 3D modelling. The model developed was successfully applied to this system. Our results showed that the annual recharge is very heterogeneous on the test site. Spatialization of recharge improves discharge modelling as evidenced by increased KGE (from 0.8 to 0.9) and more realistic flows during drought periods. It is therefore essential to spatialize recharge in karst hydrogeological modelling to improve predictive capacity and better understand functioning of the whole hydrosystem.