Author(s): K. W Choi; Joseph H. W. Lee
Linked Author(s): 李行伟
Keywords: Beach water quality; Hydrodynamic model; Statistical model; Real-time forecast
Abstract: Escherichia coli (E. coli) concentration is adopted as the main indicator of beach water quality in Hong Kong due to its high correlation with swimming associated illnesses. As part of the WATERMAN system - coastal water quality forecast and management system for Hong Kong, both 3D deterministic hydrodynamic and multiple linear regression (MLR) models have been developed to provide daily water quality forecasting for eight marine bathing beaches in Tsuen Wan, which are only about 8 km from the Harbour Area Treatment Scheme (HATS) outfall discharging 2.5 million m3/d of chemically enhanced primary treatment (CEPT) effluent. The forecast performance of the two models on the compliance of beach water quality objective is studied for a three-year period (2016-2018). While the 3D model is process-based and the MLR model is data- driven, both models have comparable performance with an overall accuracy of about 70-80%. The model forecasts of E. coli concentrations are significantly correlated with the sampling data obtained from the regular monitoring programme. In general, the MLR models have slightly higher overall accuracy and better correlation with the observation, but by its very nature can predict the E. coli level only around the standard sampling time on daily basis (around 11 am). On the other hand, the 3D model can capture the diurnal variation of beach water quality and provides better prediction for the observed water quality exceedances.