You are here:
MULTIPLE REGRESSION MODELS FOR HINDCASTING AND FORECASTING MIDSUMMER HYPOXIA IN THE GULF OF MEXICO
Citation:
GREENE, R. M., J. C. LEHRTER, AND J. D. HAGY. MULTIPLE REGRESSION MODELS FOR HINDCASTING AND FORECASTING MIDSUMMER HYPOXIA IN THE GULF OF MEXICO. ECOLOGICAL APPLICATIONS. Ecological Society of America, Ithaca, NY, 19(5):1161-1175, (2009).
Impact/Purpose:
Multiple regression models developed to describe the relationship between the area of bottom water hypoxia along the northern Gulf of Mexico and Mississippi-Atchafalaya River nitrate concentration, total phosphorus (TP) concentration, and discharge.
Description:
A new suite of multiple regression models were developed that describe the relationship between the area of bottom water hypoxia along the northern Gulf of Mexico and Mississippi-Atchafalaya River nitrate concentration, total phosphorus (TP) concentration, and discharge. Variability in hypoxic area during midsummer was best described by models that incorporated May discharge, May nitrate and February TP concentrations; models using spring averages (April – June) for discharge and nitrate concentration and winter averages (January – March) for TP concentration performed equally well. The ensemble of these regression models predicted the observed hypoxic zone area within +30%. The models were used to forecast the response in mean hypoxic area to various scenarios in which riverine nutrient concentrations and freshwater discharge were altered via a Monte Carlo approach. The Monte Carlo simulations were also used to evaluate the number of years needed to discern statistically significant responses to nutrient reduction scenarios. Hindcast models using historical Mississippi River discharge and nitrate concentrations during May indicated that large-scale hypoxia (e.g., several thousand km2) was likely present on the shelf intermittently as early as 1955. Model forecasts predicted that a 45% nitrate and TP concentration reduction would be required to reduce hypoxic area to approximately 5,000 km2, the coastal goal established by the Mississippi River/Gulf of Mexico Watershed Nutrient Task Force. Monte Carlo simulations showed that a statistically significant reduction in midsummer hypoxic area could be detected after 5 years with 50% nitrate reductions or 45% nitrate and TP reductions. However, if nutrient reductions were achieved gradually (e.g., over 10 years), a protracted period would be required to resolve a significant downward trend in nutrients against the background of natural variability. In this gradual change scenario, it appeared unlikely that a significant decreasing trend in hypoxic area would be detected until several years after nutrient reductions were fully achieved. The multiple regression models and statistical approaches applied provide improved capabilities for evaluating dual nutrient management strategies to address Gulf of Mexico hypoxia and a clearer perspective on the strengths and limitations of approaching the problem using regression models.