Uncertainty Management for Environmental Risk Assessment Using Geostatistical Simulations

Abstract

Geostatistical simulations are very popular in the petroleum and mining industries as they address some issues where “kriging-like” techniques fail. The multiple capabilities of geostatistical simulations have also proven to be of major interest to the environmental sciences. Non-linear estimation techniques such as disjunctive kriging, uniform conditioning or conditional expectation may be convenient for solving problems like the estimation of the probability of exceeding thresholds and contaminated volumes. But if these problems involve multiple point statistics or non-stationary cases, these techniques are not sufficient. Besides, in many situations, the multivariate aspect of the problem cannot be ignored and co simulation methods turn out to be the most efficient solution. The powerful contribution of geostatistical simulation methods to environmental issues is illustrated with applications in the domains of air pollution, soil contamination and hydrogeological modelling. The first example shows how simulations can quantify the risk of exposure of a city’s population to air polluted with NO₂. The second example deals with soil contamination with poly-cyclic aromatic hydrocarbons at former industrial sites. The last example is taken from a national program for the storage of nuclear waste. When faced with the complexity of today’s environmental risk assessment issues, optimal decision making requires knowledge of the prevailing uncertainties. Geostatistical simulations provide an assessment framework as well as solutions to achieve this goal.