An interactive sampling procedure is proposed to optimize environmental risk assessment. Subsequent sampling stages were used as quantitative pre-information. With this pre-information probability maps were made using indicator kriging to direct subsequent sampling. In this way, optimal use of the remaining sampling stages was guaranteed. Interactive sampling was applied to a lead-pollution in the Dutch city of Schoonhoven. Environmental risks were quantified by the probability of exceeding the intervention level. The data and sampling schemes were stored in a GIS. Using six conditional simulations of stochastic fields, interactive sampling schemes were compared to conventional sampling schemes by calculating type I and type II errors. The interactive schemes had much lower type I errors than the conventional schemes, and comparable type II errors. Moreover, the interactive sampling schemes left a smaller fraction of the not-sanitated area polluted than the conventional ones did. They predicted almost 70% of the area correctly, as compared to 55% by conventional schemes.
Read full abstract