## Abstract

This paper presents a stochastic approach to delineate the capture zones for the well field 'Het Rot' at Nieuwrode (Belgium). The conductivity of the production aquifer is modelled as a random space function (RSF). For the conductivity of the overlying semi-pervious unit and the top phreatic aquifer homogeneous parameters are used. The parameters of the stochastic model and the homogeneous parameters are treated as random variables to account for the fact that they are unknown. Conductivity measurements for the production aquifer update the prior distributions for the structural parameters of this layer using Bayes' theorem, yielding posterior parameter distributions. Conditional realisations of the production aquifer conductivity field are generated using parameter sets sampled by Monte Carlo from the posterior distributions of the structural parameters. The realisation are combined with values for the other unknown parameters, sampled by Monte Carlo from the corresponding prior distributions. In a second application of Bayes' theorem, head observations are used to assign probability-based weights to the parameter realisations. Using particle tracking the 25-year capture zone and the well catchment are determined for each realisation, which are weighted by the associated probability of the parameter realisation. Statistical analysis of the set of weighted capture zones results in a capture zone probability distribution.

Original language | English |
---|---|

Pages (from-to) | 199-202 |

Number of pages | 4 |

Journal | Aardkundige Mededelingen |

Volume | 12 |

Publication status | Published - 2002 |

Externally published | Yes |

Event | First Geologica Belgica International Meeting - Leuven, Belgium Duration: 11 Sep 2002 → 15 Sep 2002 |

## Keywords

- Bayesian inference
- Capture zone
- Groundwater
- Spatial stochastic approach