A novel approach for recreational water quality along the Texas beaches: Implication of a predictive neural network model for real-time nowcasting

Prof. Jongsun Kim, Ph.D., Virginia Institute of Marine Science, Department of Biological Science, Gloucester Point, VA, United States and Chuan-Yuan Hsu, Texas A&M University, Department of Oceanography, College Station, United States
Abstract:
One of the critical environmental issues in coastal GOM is recreational beach water quality. The BEACH Act of 2000 by US EPA announced the enterococci level as an index of beach conditions for recreational activities and swimming. Beach closures are associated with the presence of high bacterial contamination caused by deteriorating water quality. In this study, we developed a novel model system, which is a combination of deep learning model and ocean dynamic numerical model to nowcast/forecast the levels of enterococci in Texas coastal recreation water at any time of the day. We developed our predictive model for using real-time environmental data (e.g., Weather forecast, wind speed, direction, stress curl, river discharge rate, nutrients, seawater salinity, temperature, and various types of antecedent rainfalls). As preliminary results, we collected the data over 100 stations from the 67 recreational beaches along the Texas coast from 2005 to 2019 by weekly during swimming season and bi-weekly during non- swimming season period, respectively. The results of our predictive model can be utilized to nowcast/ forecast the water quality of beach and can be compared with laboratory test analysis. Thus, this model approach is first time to nowcast/forecast the water quality in the recreational beach and might be useful tool to reducing the risk of human health.