Document Type


Publication Date



Beach water issues are gaining worldwide attention due to their impact on health and other environmental problems. The Ontario beaches require beach managers to issue swimming advisories when water quality standards are exceeded since users of recreational waters may be exposed to elevated pathogen levels through various point and non-point sources. Typical daily notifications rely on microbial analysis of indicator organisms (e.g. Escherichia coli), which require 18-24 hours to provide an adequate response. This research evaluated the use of Artificial Neural Networks (ANNs) and Evolutionary Polynomial Regression (EPR) for real time prediction of E.coli in the beach waters of Toronto (Ontario, Canada). The nowcasting models were developed in combination with readily available real-time environmental and hydro-meteorological data during the bathing season (June-August) of 2008 to 2012 for Toronto’s three beaches. The performance of ANN and EPR models was assessed in terms of their ability to accurately match observed E.coli concentrations and to correctly predict when the beach water quality standard - primary contact recreation of 100 CFU/ 100 ml - is exceeded. In the ANN models the feed forward back propagation algorithm was used for the analysis. Different combinations of predictor variables were used and the best results were obtained when stream flow, various combinations of antecedent rainfalls, lake level, solar radiation, past counts of E.coli, wind direction and speed were used. The results of the developed models were compared with those of the conventional method and statistical models, and it was found that the predictions of ANN models slightly outperforms with better accuracy. The best performing ANN models on each beach are able to predict approximately 60% to 90% of the E.coli concentrations, whereas the EPR models return a correct prediction as high as 77%. The developed methodologies offer a promising alternative to traditional methods for the protection of health in bathing beaches.


Session R68, Early Warning and Nowcasting Approaches for Water Quality in Riverine and Coastal Systems



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.