Nowcasting and Forecasting Beach Bacteria Concentration Using EPA's Virtual Beach Software
Abstract
Beaches in the United States of (North) America are subject to closure when bacterial counts exceed water quality criteria. Many authorities base these decisions on water samples that typically require at least 18 hours to analyze. This persistence approach, or model, often leads to erroneous decisions due to the great variability in bacterial concentrations. Beaches are closed when they could be open and vice versa, their true status unknown until the next day. Studies show that mathematical models based on multi-variable linear regression (MLR) principles can produce better estimates, or nowcasts, using real-time explanatory variables, such as turbidity, cloud cover, and rainfall. To make such models generally available, the Environmental Protection Agency (EPA) is developing a program called Virtual Beach (VB). VB is public-domain software for developing site-specific predictive models. It features capabilities that make it possible with reasonable effort to develop, and compare the performance of, static and dynamic MLR models. The results of tests on 2006 Huntington Beach, Lake Erie beach data are presented. In addition to nowcasting, the work begins to address the question, can weather and water forecasts be used to forecast beach conditions in advance? A preliminary affirmative answer is provided based on an analysis of the Huntington Beach data, with weather forecasts for nearby Cleveland-Hopkins international airport, and NOAA lake condition forecasts. We encourage those engaged in beach monitoring and management to request VB, applying the nowcast and forecast models developed with it to their locations of interest. Disclaimer: Although this work was reviewed by EPA and approved for presentation, it may not necessarily reflect official Agency policy.
- Publication:
-
AGU Spring Meeting Abstracts
- Pub Date:
- May 2007
- Bibcode:
- 2007AGUSMOS23G..05F
- Keywords:
-
- 0399 General or miscellaneous;
- 0430 Computational methods and data processing;
- 0458 Limnology (1845;
- 4239;
- 4942);
- 0466 Modeling;
- 4840 Microbiology and microbial ecology (0465)