Improvements to Shellfish Harvest Area Closure Decision Making Using GIS, Remote Sensing, and Predictive Models

Abstract

Currently, many states use precipitation information to regulate periodic closures of shellfish harvest areas based on a presumptive relationship between rainfall and bacteria concentration. We evaluate this relationship in four South Carolina estuaries and suggest new predictive models that integrate remote sensing precipitation data with additional environmental and climatic data. Model comparisons using Akaike’s information criterion, tenfold cross-validation, and model r 2 values show substantial and consistent improvements using integrated precipitation, salinity, and water temperature data as predictors. These models may be useful for shellfish area closure regulation support. The model development approaches used here may also be useful in estimating bacteria concentration at beaches and can serve as the basis for developing near-real-time estimates and forecast predictions of bacteria levels for closure decision-making tools.