Modeling uncertainty and imprecision in water resource systems (original) (raw)

Water resource systems are fraught with stochastic uncertainties and imprecision due to insufficient knowledge. Realistic, plausible and relatively simple models should be found to reflect these characteristics before practitioners would be ready to apply the theory. It is proposed that a joint probabilistic/fuzzy set approach may help in this respect. Thus uncertainty and imprecision in risk analysis, an important area of water resource systems analysis, are to be modeled simultaneously. The case of health risk due to groundwater contamination is considered. Fuzzy set geostatistics can be used to describe imprecision in the spatial variability of exposure; that is, contaminant concentration. The consequence of an exposure is typically described by dose-response relationships, generally based on a few animal experimental data. Fuzzy regression is shown to be applicable for encoding imprecision in such dose-response relationships. More generally, it is shown how risk management can be performed by trading off risk and cost under uncertainty and imprecision, leading to results that decision makers can readily visualize.