Monte Carlo and fuzzy approaches to environmental risk assessment in plant production

Using N-leaching under pasture conditions as an example, suitable techniques of risk assessment to reduce possible hazards to the environment due to agricultural operations are evaluated. Unlike in the case of industrial plants where often the worst scenario is considered, in agriculture usually small possibilities or probabilities to violate limiting values can be tolerated, when the risk is low and the overall majority of values are clearly off limits. Thus distributions of the occurance of hazard are needed and these distributions must be achieved for assumed uncertainties in inputs (management) and in production parameters (soil, weather etc.). A first assessment is done by numerical experiments (Monte Carlo techniques) or by concentrating on possible extremal points in the parameter space (a-cut fuzzy approach). The results of both techniques can be considerably improved, when additional knowledge (assumed correlations or empirical rules) are incorporated. With both techniques, there is no reformulation of the underlying model. With an increase in computer power, the methods appear suited for risk assessment relying on machine power instead on programming efforts.