Risk-Averse Importance Sampling of Tree Attributes in High-Risk Forested Areas

Abstract This study develops the theory of risk-averse importance sampling and explains its potential application to forest inventory estimation through use a heuristic simulation. When risk-producing elements landscape are known, strategy can be created that results in fewer samples high-risk areas. Our simulation shows for certain populations, highly effective at reducing both risk field crew members (requiring only 10% plot visits riskiest category) sample variance relative simple random sampling. The method is shown especially helpful when population values interest decreases with increasing risk, reduction mean square error (MSE) 84% 99% these cases. also showed opposite effect on MSE expected increase risk. By safety, should improve frequency accuracy observations, potentially leading even bigger gains estimate precision. We recommend any time hazardous conditions result high number missing observations reasonably accurate characterizations risks developed. Study Implications: During collection data, safety personnel always primary importance, but never has been component design. provides low-risk probability areas interested practitioners. forested area known described functional form. decrease It hypothesized if defined will reduce field, response rates measurements increase, and, turn, lead final estimates. Risk-averse recommended it likely higher level observational success.