An autonomous decision support system for manganese forecasting in subtropical water reservoirs

Abstract

Manganese monitoring and removal is essential for water utilities in order to avoid supplying discoloured water to consumers. Traditional manganese monitoring in water reservoirs consists of costly and time-consuming manual lake samplings and laboratory analysis. However, vertical profiling systems can automatically collect and remotely transfer a range of physical parameters that affect the manganese cycle. In this study, a manganese prediction model was developed, based on the profiler's historical data and weather forecasts. The model effectively forecasted seven-day ahead manganese concentrations in the epilimnion of Advancetown Lake (Queensland, Australia). The manganese forecasting model was then operationalised into an automatically updated decision support system with a user-friendly graphical interface that is easily accessible and interpretable by water treatment plant operators. The developed tool resulted in a reduction in traditional expensive monitoring while ensuring proactive water treatment management.