Articles | Volume 17, issue 6
Research article
03 Jun 2013
Research article |  | 03 Jun 2013

A critical assessment of simple recharge models: application to the UK Chalk

A. M. Ireson and A. P. Butler

Abstract. Quantification of the timing and magnitude of point-scale groundwater recharge is challenging, but possible at specific sites given sufficient high spatial and temporal resolution field observations, and a suitable physically based model. Such models are generally too computationally intensive and have too many unknown parameters to be practically applicable within distributed, larger-scale hydrological or groundwater models. This motivates the need for simpler recharge models, which are widely used within groundwater models. However, it is important that these models are able to capture adequately the unsaturated zone flow processes. We perform an inter-comparison of recharge simulated by a detailed physically based model and a simple recharge model, with both models applied to a field site in the fractured porous Chalk in the UK. Flow processes are simulated convincingly using a dual permeability, equivalent continuum, vertically heterogeneous, Richards' equation model, applied to a 2-D hillslope transect. A simple conventional recharge model was then calibrated to reproduce the water table response simulated by the physically based model. The performance in reproducing the water table was surprisingly good, given the known discrepancies between the actual processes and the model representation. However, comparisons of recharge fluxes simulated by each model highlighted problems with the process representations in the simple model. Specifically, bypass flow events during the summer were compensating for recharge that should have come from slow, continual drainage of the unsaturated zone. Such a model may still be useful for assessment of groundwater resources on a monthly basis, under non-extreme climatic conditions. However, under extreme wet or dry conditions, or under a changed climate the predictive capacity of such models is likely to be inadequate.