Understanding Nowcasting Data

The State of Ohio uses standards for E. coli to assess whether the water quality is acceptable for swimming or other recreational activities. E. coli is a bacterium that is not usually disease causing but is found in the gastrointenstinal tract of healthy warm-blooded animals, including humans, and their feces. Its presence is an indicator that disease-causing organisms are also likely to be present. If the level of E. coli goes over the “single-sample bathing-water standard” of 235 colonies per 100 milliliters, the beach manager will issue a water-quality advisory. A milliliter is a unit of volume in the metric system; 100 milliliters is equivalent to just under ½ cup.

What is commonly done now to monitor water quality at beaches is to collect a water sample from the swimming area, transport the sample to a laboratory, and determine numbers of E. coli in the sample. The problem is that it takes as least 18 hours to grow E. coli in the laboratory. Unfortunately, by the time E. coli results are compiled, beach water-quality conditions may have already changed. By contrast, it takes less than an hour to estimate recreational water quality using the predictive model and transmit results through the nowcasting system.

A predictive model is a mathematical system that uses easily measured environmental and water-quality “variables,” such as wave height and rainfall, to estimate levels of E. coli. Predictive models are developed from several years of measurements taken at a particular beach, and all models are beach specific. We can measure these variables each morning, enter them into a computer program, and obtain the probability of exceeding the bathing-water standard.

Predictive models work like a weather forecast in the nowcasting system. If there was a 70-percent chance of rain, would you go on a picnic? A 15-percent chance of rain? Similarly, if there was a 70-percent chance that the standard will be exceeded, would you head to the beach and swim in the water? Beach managers decide on a threshold probability to issue beach advisories; thresholds are based on examinations of past data. For Huntington during 2008, the threshold is either 27 or 28 percent, depending on the date. Probabilities equal to or above the threshold would indicate to the beach manager that E. coli concentrations are most likely above the bathing-water standard, and an advisory would be issued. Probabilities below the threshold indicate that E. coli concentrations are most likely below the standard. And again, because the nowcast assessments of recreational water quality are ready within an hour, they are better than the old methods that take 18 hours to complete.