Predictive Modeling for Beach Water Quality
Imagine that weather forecasting went something like this – you measure temperature, wind speed and direction, relative humidity and rainfall for one day. Then, the next day, you report those values as the weather for today and also for the next several days, until you have the resources and money to collect another set of measurements. Ridiculous? Of course. But that pretty much describes the way we do beach monitoring. The reported beach water quality is based on data we collected yesterday, or perhaps as long ago as last week.
One way to make beach monitoring more current is to develop and use rapid testing methods that can give results in two to four hours instead of 24 hours or longer. But that still is not real-time, and it still falls well short of predicting what the water quality will be tomorrow when you want to go to the beach.
In order to address this need, scientists, local beach mangers, county and state health departments, environmental organizations and the U.S. EPA have been working on developing predictive models that can be used to forecast beach water quality with similar accuracy to forecasting the weather.
The computer models that are being developed (and in some locations, already used) are not intended to replace beach monitoring. Beach monitoring data is one of the inputs to the model and is also used to calibrate and “truth check” the model. The model estimates fecal indicator bacteria levels in the surf-zone in real time, using both water quality data (most recent samples and historic trends) and environmental conditions, such as rainfall, tide levels, solar radiation, wind, storm drain flows, and swell conditions.
Predictive modeling has been in use for several years at certain beaches on the Great Lakes, using models such as SwimCast, NowCast and Virtual Beach. Development and testing of predictive modeling for beach monitoring has also been done at ocean beaches in Texas and in Southern California.
EPA has developed a guidance document, Six Key Steps to Developing and Using Predictive Tools at Your Beach, to encourage states’ use of modeling as a predictive tool to make timely beach advisory or closure decisions and issue same-day public notifications. EPA has also conducted research on predictive models for beach notifications, including how and where they are being used, and the differences between models.
In California, researchers at Stanford University have worked with the environmental organization Heal the Bay to develop and test predictive models. The research collaborative is overseeing a pilot program that predicts good or poor water quality for the day at three beaches that have historically struggled with bacterial pollution: Doheny State Beach in Orange County, Santa Monica Beach at the Santa Monica Pier, and Arroyo Burro Beach in Santa Barbara County.
Through this work, there is some promise that beach water quality forecasting will some day become as timely and accurate as weather forecasting.