Combining stochastic simulations and inverse modelling for delineation of groundwater well capture zones

In hydrogeology, protection zones of a spring or a pumping well are often delimited by isochrones that are computed using calibrated groundwater flow and transport models. In heterogeneous formations, all direct and indirect data, respectively called hard and soft data, must be used in an optimal way. Approaches involving in situ pumping and tracer tests, combined with geophysical and/or other geological observations, are developed. In a deterministic framework, the calibrated model is considered as the best representation of the reality at the current investigation stage, but result uncertainty remains unquantified. Using stochastic methods, a range of equally likely isochrones can be produced allowing to quantify the influence of our knowledge of the aquifer parameters on protection zone uncertainty. Furthermore, integration of soft data in a conditioned stochastic generation process, possibly associated with an inverse modelling procedure, can reduce the resulting uncertainty. A stochastic methodology for protection zone delineation integrating hydraulic conductivity measurements (hard data), head observations and electrical resistivity data (soft data) is proposed.