Development of a nowcasting system using machine learning approaches to predict the level of fecal contamination at recreational beaches in Korea

Authors: PARK, YONGEUN - University Of Ulsan College Of Medicine; KIM, MINJEONG - University Of Ulsan College Of Medicine; Pachepsky, Yakov; CHOI, SEOUNG-HWA - University Of Ulsan College Of Medicine; CHO, JEONG-GOO - University Of Ulsan College Of Medicine; CHO, KYUNG HWA - University Of Ulsan College Of Medicine

Submitted to: Journal of Environmental Quality
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/25/2018
Publication Date: 3/22/2018
Citation: Park, Y., Kim, M., Pachepsky, Y.A., Choi, S., Cho, J., Cho, K. 2018. Development of a nowcasting system using machine learning approaches to predict the level of fecal contamination at recreational beaches in Korea. Journal of Environmental Quality. 47(5):1094-1102.

Interpretive Summary: Beaches are being closed when the concentration of bacteria indicator organisms exceed regulation limits. It takes a day at current technologies to determine whether the critical indicator bacteria concentration limits are exceeded. The decisions on beach water safety are globally made based on outdated information. An alternative way to evaluate the beach water quality is to rely on relationships between the indicator bacteria concentrations to environmental variables and water treatment plant parameters. These relationships are known to be very complex. To describe them mathematically for the forecast purposes, we applied two artificial intelligence techniques. It appeared that both techniques could provide an acceptable accuracy in predicting the microbiological quality of beach water, and could be used to determine the most important predictors. Different predictive relationships were found for two difference beaches. Results of this work can be of use for professional in the field of microbiological quality of surface waters in that they demonstrate the technology for development of site-specific time and cost-efficient microbiological water quality monitoring and assessment means.

Technical Abstract: Microbial contamination in beach water poses a public health threat due to water-borne diseases. To reduce the risk of exposure to fecal contamination, informing beach-goers in advance about the microbial quality of the water is important. Currently, determining the level of fecal contamination takes 24 h. The objective of this study is to predict the current level of fecal contamination (i.e., enterococcus (ENT) and Escherichia coli (E. coli)) using readily available environmental variables. Artificial neural network (ANN) and support vector regression (SVR) models were constructed using data from the Haeundae and Gwangalli Beaches in Busan City. The input variables included the tidal level, air and water temperature, solar radiation, wind direction and velocity, precipitation, discharge from the wastewater treatment plant, and suspended solid concentration in beach water. The dependence of the fecal contamination on the input variables was statistically evaluated; the precipitation, discharge from the wastewater treatment plant, and wind direction at the two beaches had significant effects on the changes in the two bacterial concentrations (p-values < 0.01). Both models performed well; the performance of the ANN model for predicting ENT at Gwangalli Beach was only significantly higher than that of the SVR model with the training dataset (p-value < 0.05). The sensitivity analysis results demonstrated that SVR appeared to be more relevant for interpreting the cause-and-effect relation between the bacterial concentrations and input variables. This study suggests an effective prediction method to determine whether a beach is safe for recreational use.