Dr. Daniel Obenour, Civil Construction and  Environmental Engineering, NC State

Title: Enhancing Ecological Assessment and Prediction through Probabilistic Modeling: Examples from the Dead Zone

Abstract: In experimental and field-based ecological research, statistical tests are often employed to assess the veracity of data-derived findings. In many cases, these tests provide a quantification of uncertainty with respect to some system state or biophysical rate. Ecological modeling research, which is typically concerned with numerous system states and rates, could also benefit from probabilistic methods to rigorously test hypotheses and quantify uncertainties. Such methods would help build confidence in model forecasts and provide richer information for environmental decision making. Nonetheless, perhaps due to the complexity of ecological systems, progress toward uncertainty quantification in ecological modeling has been limited. Using the hypoxia (or “Dead Zone”) problem in the northern Gulf of Mexico as a case study, I demonstrate the application of probabilistic methods to estimate hypoxic extent and to mechanistically predict its interannual variability. First, a novel geostatistical modeling framework is developed to simulate BWDO and hypoxic layer thickness across the Gulf shelf, providing probabilistic estimates of hypoxic area and volume over a 27-year study period. Second, a probabilistic and parsimonious mechanistic model, driven by nutrient loading, flow, and weather is developed to predict the interannual variability in hypoxic extent over this same study period. This work provides new insights into short and long-term trends in Gulf hypoxia, and the biophysical factors controlling its variability.