Predicting Structural Change in Environmental Systems: a Complement to Model-based Forecasting

摘要

Simulation models are commonly employed for predicting both the impacts of natural stressors and the outcomes of anthropogenic actions on the environment. Such modelbased forecasting depicts a forward approach to environmental systems analysis. This paper presents a complementary backward approach that uses prescribed future scenarios to identify the dominant components of the modeled system. In particular, two Monte Carlo methods for uncertainty analysis – namely, Regionalized Sensitivity Analysis (RSA) and Tree-Structured Density Estimation (TSDE) – are applied in detecting structural change between observed past and speculated future ecological behavior in a small southeastern US Piedmont reservoir (Lake Oglethorpe, GA). Results indicate that sediment-water-nutrient interactions are key elements of structural change that distinguish between the past and future scenarios analyzed. General implications for predicting structural change in environmental systems are discussed.