Habib, E. and Reed, D., 2013. Parametric uncertainty analysis of predictive models in Louisiana's 2012 Coastal Master Plan.

This study presents an assessment of uncertainty associated with the predictive models utilized in Louisiana's 2012 Coastal Master Plan. In this context, model uncertainty was defined as the deviation of model prediction from the actual ecosystem response to certain proposed projects. The focus is on parametric-related uncertainties, which are due to imperfect knowledge about parameters and relationships used within the models. Due to the large number of models used in the master plan, a reduced set of model parameters (34) was identified as the most uncertain. A limited sampling experiment was designed on the basis of stratified sampling from predefined simple probability distributions of the selected parameters. Two phases of analysis were conducted. The first phase focused on examining the impact of parameter uncertainties on model predictions and comparing such uncertainties with the predicted impacts of individual projects. The second phase focused on comparing model uncertainties in predicting the future-without-action conditions vs. a proposed draft version of the master plan. The study attempts to answer some key questions that are relevant for model development and planning and project selection aspects: How uncertain are the models in predicting changes in key ecosystem metrics? Does the uncertainty vary spatially across the coast and temporally into future years? How does parameter-induced uncertainties compare with those due to other exogenous large-scale drivers? How can the uncertainty analysis inform decisions? Finally, the paper discusses implications of uncertainties for using the models as prediction tools, and highlights critical data gaps and modeling development efforts needed for future analysis.