Skip to main content

Tools and Training for Predicting Beach Conditions

Use the links on this page to access publications and references related to our work with beach pathogen indicators and water quality models, including a list of useful articles and documents [PDF] dealing with statistical modeling of E. coli and other pathogen indicators.

Virtual Beach Software

Virtual Beach is a software package designed to construct site-specific multiple-linear regression (MLR) models for predicting pathogen indicator levels at recreational beaches. The MLR analyses outperform persistence models at beaches where conditions, such as weather, hydrology and human and animal traffic levels change significantly from day to day. Virtual Beach was developed by scientists with the Ecosystems Research Division of the National Exposure Research Laboratory at the EPA's Office of Research and Development.

The Virtual Beach software can be downloaded from the U.S. EPA website.

Workshops

Wisconsin DNR staff conducted two hands-on training workshops for local beach managers in 2009.

  • An initial pilot workshop, hosted by Concordia University's Center for Environmental Stewardship, prior to the opening of the beach season.
  • A second at the 2009 State of Lake Michigan/Great Lakes Beach Association Conference.

Learning Modules

For beach managers or others interested in building multivariate statistical models to “nowcast” pathogen indicator levels at their beach, the following learning modules were developed for the State of Lake Michigan-Great Lakes Beach Association workshop.

Learning Module I - Model Building [PDF]
In this module you will learn how to:

  • format and import data tables;
  • evaluate data using scatter plots;
  • transform variables;
  • exclude unwanted observations and variables;
  • check for multicollinearity (non-independence);
  • convert wind speed and direction into “longshore” and “onshore;" and
  • create interaction terms (combined variables).

Learning Module II - Model Evaluation and Nowcasting [PDF]
In this module you will learn how to:

  • fit models;
  • identify influential outliers;
  • identify best, unbiased models;
  • make single-day predictions with 95% confidence intervals; and
  • make real-time predictions ("nowcasts").

Example Data

The DNR compiled example data to be used with the learning modules. The data are from the 2003, 2004 and 2005 beach seasons for Red Arrow Park Beach in Manitowoc.

Other Resources

Check out these resources for additional guidance on building and evaluating models.