9 - Determination of capture zones of wells by Monte Carlo simulation

Summary

ABSTRACT

Effective protection of a drinking water well against pollution by persistent compounds requires the knowledge of the well's capture zone. This zone can be computed by means of groundwater flow models. However, because the accuracy and uniqueness of such models is very limited, the outcome of a deterministic modeling exercise may be unreliable. In this case stochastic modeling may present an alternative to delimit the possible extension of the capture zone. In a simplified example two methods are compared: the unconditional and the conditional Monte Carlo simulation. In each case realizations of an aquifer characterized by a recharge rate and a transmissivity value are produced. By superposition of capture zones from each realization, a probability distribution can be constructed which indicates for each point on the ground surface the probability to belong to the capture zone. The conditioning with measured heads may both shift the mean and narrow the width of this distribution. The method is applied to the more complex example of a zoned aquifer. Starting from an unconditional simulation with recharge rates and transmissivities randomly sampled from given intervals, observation data of heads are successively added. The transmissivities in zones that do not contain head data are generated stochastically within boundaries typical for the zone, while the remaining zonal transmissivities are now determined in each realization through inverse modeling. With a growing number of conditioning data the probability distribution of the capture zones is shown to narrow. The approach also allows the quantification of the value of data. Data are the more valuable the larger the decrease of uncertainty they lead to.