Task 2: Improve existing beach health models
Task Chiefs: Donna Francy, Richard Whitman
Collaborators: Cuyahoga County Board of Health, Cuyahoga County Sanitary Engineers, Northeast Ohio Regional Sewer District, Ohio Department of Health, University of Toledo, National Park Service, Ogden Dunes, Indiana, Portage, Indiana, Indiana Department of Environmental Management, Environmental Protection Agency, NOAA, Michigan State University
Period of Project: March 2008 - present
Task 2 References
Ogden Dunes research image collection
At Edgewater, Cleveland, OH, ORPP scientists are working to improve the accuracy of a predictive model and maintain real-time measurements of model variables.
Task 2 goal and objectives
The overall goal of Task 2 is to improve existing models so that they provide more accurate responses and operate in a real-time fashion. Specific objectives for 2008 are to:
- Identify additional variables that may improve the predictive capabilities of the models
- Work toward including instrumentation that enables beach managers to use models without having to travel to the beach to collect data
- Determine if afternoon samples are needed to predict changing water-quality conditions accurately
- From information collected in 2008, make recommendation for improving existing models and expanding to other beaches
Task 2 approach
Ohio
Work at two Ohio beaches where operational models are already in place (Huntington, Bay Village and Edgewater, Cleveland) and at one beach where local cooperation is already in place (Maumee Bay State Park (MBSP), Oregon)
Installing telemetry at Ogden Dunes, Indiana.
Indiana
Work at adjacent beaches at the National Park Service and Ogden Dunes, Indiana
Tasks
-
Purchase, install, and maintain the equipment needed to obtain accurate near-real-time data.
- Ohio, Huntington: Install nearshore buoy or other data collection platform to measure wave height and currents (ADP), and turbidity; install onshore PAR sensor and rain gage; install telemetry to transmit data.
- Ohio, Edgewater: Add turbidity meter to existing buoy, install an ADP in the lake bottom.
- Indiana, Ogden Dunes/NPS: install turbidimeter, ADP with telemetry in the lake, turbidimeter in Burns Ditch. Measure E. coli in replicates 4 days/week along transects and Burns Ditch.
- Ohio: If funding permits, conduct synoptic studies of ADP current profiles to determine the relations between beach characteristics and ADP measurements at one point.
- Collect data from mid-May to Labor Day for Ohio, mid-June to mid-August in Indiana. Ensure that high quality data are collected.
- All beaches. Data are collected each morning through existing local programs. USGS will design a QA/QC plan for each agency, perform QA/QC checks of field and laboratory activities, and provide reference samples for E. coli and turbidity analyses
-
Add afternoon collection of samples for E. coli and explanatory variables at Huntington, Ohio.
- Work with local agencies to obtain summer interns to collect and analyze afternoon samples.
- Indiana will review retrospective data to determine the affects of sunlight E. coli populations
-
Deploying nearshore ADCP at Ogden Dunes.
Work with Ohio Department of Health (ODH), the agency that oversees the beach monitoring program in Ohio, to collect high quality data and identify additional beaches that may benefit from predictive models.
- Compile data on explanatory variables and run and validate existing models for Indiana and Ohio. Monitor the daily performance of the models throughout the season.
-
Determine the performance of the models by calculating percent correct and percent of false positives
and false negatives (based on the actual concentration of E. coli) in Ohio, the approach in Indiana will be judged against anticipated confidence intervals for the dependent variable.
- Compare relative importance values and rankings of independent variables, variation of dependent variables and solar inactivation between Indiana and Ohio. Compare models. Independently cross-validate Ohio and Indiana models using agreed upon procedures.
- Explore binomial regression and regression trees as optional approaches to both Ohio and Indiana models
- Begin to develop new models for the recreational season of 2009 at Ohio and Indiana beaches. Begin to develop models for MBSP.
- Begin to write reports or articles on the work done during 2008.
|