You are here:
Now-cast modeling workshop for predicting bacteria levels in shellfish harvesting waters and recreational beaches
Citation:
Cyterski, M. Now-cast modeling workshop for predicting bacteria levels in shellfish harvesting waters and recreational beaches. Class at University of South Carolina, Baruch Institute, Georgetown, SC, May 13 - 14, 2019.
Impact/Purpose:
Class at University of South Carolina, Baruch Institute
Description:
Fecal indicator bacteria (FIB) are the primary method for measuring the water quality of recreational waters in the United States. Statistical models are often developed to make predictions about daily FIB concentrations that could potentially impact human health at specific beach locations. Virtual Beach (VB) is a software tool that facilitates the development of empirical models for the prediction of microbial indicator levels at recreational beaches. The software is primarily designed for beach managers without considerable statistical expertise who are responsible for making decisions regarding beach closures or the issuance of swimming advisories due to microbial contamination. VB has been applied to develop models for predicting a suite of water quality indicators at freshwater and saltwater beach sites across the United States, but primarily in the Great Lakes region. The tool currently provides three analytical techniques for model development: multiple linear regression (MLR), partial least squares regression (PLS), and generalized boosting (GBM). VB uses an integrated mapping component to determine the geographic orientation of the beach to decompose wind/current/wave speed and direction into along-shore and onshore/offshore components. A new web-based version of VB will include logistic regression for modeling binary outcomes and allow users to have immediate access to the latest updates without the burden of downloading and installing a new executable. The new version will provide faster data processing and an enhanced visualization interface, allowing the user to analyze model outputs dynamically.