Author(s): W. Thoe; H. C. Wong; K. W. Choi; J. H. W. Lee
Keywords: No Keywords
Abstract: Currently beach advisories for Hong Kong are based on Escherichia coli (E. coli) concentrations sampled at sparse intervals of 3-14 days during the bathing season. As it takes more than 1 day to carry out a standard E. coli measurement, beach management decisions are based on information at least 1-2 days old. It is desirable to develop beach forecasting methods to supplement and enhance the routine beach water quality monitoring. Based on the regular monitoring data during 1990-1997 and 2002-2006, a statistical analysis has been carried out to identify the critical hydro-meteorological factors that affect beach water quality, including rainfall, global solar radiation, tide level, wind, salinity, water temperature, and past E. coli level. Artificial Neural Networks (ANN) and Multiple Linear Regression (MLR) models are developed for two beaches in Hong Kong, Big Wave Bay and New Cafeteria, with different hydrographic and background pollution conditions. The MLR and ANN models are able to explain 30-50%of the observed variation in bacterial concentration. In terms of predicting exceedance and compliance of the water quality standard, the models are proved to be superior to purely relying on past monitoring data. The performance of the predictive models for real time forecasting of beach water quality has been tested by daily sampling during June-July 2007; the models are able to track the daily changes in E. coli. At Big Wave Bay, the models can successfully predict the compliance of water quality standard, with an overall predictability of around 85%. The ANN model is superior to the MLR model in predicting non-compliance of E. coli levels. The present study demonstrates that real time forecasting of beach water quality is a realistic possibility.