Adaptive Real-time Forecasting and Monitoring of Water Quality in a Crucial Link to Regional Water Supply

Real-time monitoring networks are becoming increasingly prevalent in environmental systems of operational and/or management significance as the technology for live in-situ online data collection becomes more accessible. Additionally, ecosystem and water resource pressures have continued to grow under an expanding anthropogenic footprint and projected climate pressures into the foreseeable future. A prototype adaptive real-time monitoring-integrated learning modelling system was developed and applied for an intensive study to improve the understanding of the mixing dynamics in a regionally crucial drinking water supply reservoir in Queensland, Australia. This was accomplished through the deployment of real-time temperature and dissolved oxygen monitoring in the reservoir to be used in parallel with an existing government-run streamflow gauging station and online water level monitoring data. Short-term water quality forecasts were externally forced by operational forecasts from Australia’s national weather service’s ACCESS-C model. An adaptive learning catchment model was developed and linked for each inflow arm of the reservoir using the Australian Water Balance Model approach to adapt and learn in real-time from live in-situ data. This framework enabled the automated online communication to researchers and managers around the current performance of the inflow predictions and the confidence expected in the current forecasts. Moreover, this live learning catchment model was coupled with a real-time adaptive three-dimensional hydrodynamic model of the reservoir in AEM3D assimilating and training using data from the deployed real-time temperature and dissolved oxygen monitoring system. The complete system facilitated the online adaptive forecasting of mixing dynamics in the reservoir and the automated identification of features of interest for water quality profiling, as well as dynamically monitoring the areas most valuable for model learning development to improve system-wide understanding and forecast certainty through addition into the live dataset for ongoing training and evaluation.