Environmental Sampling, Observational Economy, and Extreme Values
Abstract. Issues in environmental sampling and ecological monitoring can involve extreme values as the main inferential target, or as a design tool for cost -effective sampling. Although conventional sampling methods address the problem of estimating the population mean with a desired precision, classical procedures are not always cost-effective for studies involving extreme values. In this paper, we review some procedures that allow :
1. inference on sample extreme values based on sample means;
2. inference on population mean based on sample extreme values; and
3. inference on both population mean and extreme values,
while maintaining observational economy. The methods are illustrated with examples in compliance monitoring and enforcement, stock assessment based on fisheries data, and hazardous waste site characterization.