Inland water-quality monitoring increasingly draws on satellite observations, in-situ sensors, laboratory measurements, environmental models, and artificial intelligence. However, these components are often developed and operated separately, limiting their collective value for timely, physically interpretable, and decision-relevant forecasting. AquaWatch Australia is a CSIRO-led national capability integrating Earth observation, in-situ sensing, modelling, data analytics, and digital delivery to support water-quality monitoring and forecasting. This presentation uses selected AquaWatch-derived products to demonstrate one pathway for connecting heterogeneous observations, process-based models, and data-driven prediction within an operational inland-water workflow. Earth observation provides spatial context, in-situ and laboratory measurements anchor interpretation and validation, and process-based and data-driven models combine these information streams to generate current-state assessments and forecasts.
The case study brings together three complementary components. First, hyperspectral reflectance is transformed into a stable Chlorophyll Index observation operator to track chlorophyll-a dynamics and identify the development and persistence of bloom conditions. The index is not assumed to provide a universally transferable chlorophyll-a concentration estimate; application across contrasting water types requires local validation and, where necessary, recalibration and uncertainty characterisation. Second, automated calibration is applied to a one-dimensional hydrodynamic temperature model to improve the reproducibility and scalability of simulations of vertical thermal structure, stratification, and mixing. Third, these hydrodynamic outputs are incorporated as physically meaningful predictors in hybrid process–AI models for dissolved oxygen and mixing-regime prediction, helping data-driven models retain links to the processes governing water-quality change.
These components operate within governed data pipelines, fixed modelling environments, web services, dashboards, and human-in-the-loop interpretation, rather than as isolated research algorithms. Together, they illustrate how water-quality observations and models can be connected to shift assessment from retrospective reporting towards anticipatory decision support. The case study demonstrates how stable observation operators, calibrated process models, hybrid prediction, and operational delivery can deliver an integrated inland water-quality forecasting system.