Determination of a well head protection zone by stochastic inverse modelling

Abstract

Capture zones of wells play an important role in the protection of groundwater against pollution by persistent chemical compounds. Groundwater flow models can be used to delineate capture zones. As, however, the accuracy and uniqueness of such models may be poor, the outcome of a deterministic modelling exercise is likely to be unreliable. In such a case, stochastic modelling presents an alternative for the delineation of capture zones. In this paper, two methods are compared: the unconditional and conditional Monte-Carlo simulation. In each method, realizations of the aquifer characterized by a combination of recharge rates and transmissivity values are produced. For each realization, the capture zone of the well is determined by particle tracking. By superposition of all capture zones produced, a probability distribution is obtained that describes the probability of a point on the ground surface to belong to the capture zone. This probability is given by the fraction of catchments among all realizations which contain the point. Conditioning the calculation with measured heads usually implies a zoning or spatial resolution of the transmissivity distribution. The need for identification of large scale features for a meaningful zonation is stressed. For a given zoned aquifer, the influence of successively adding measured heads used to condition the stochasticsis studied. With an increasing number of conditioning heads, the probability distribution of the capture zone is shown to narrow. This method allows the quantification of the value of measured head data.