Abstract. For enhanced public safety and water resource management, a three-dimensional operational lake hydrodynamic forecast system called COASTLINES (Canadian cOASTal and Lake forecastINg modEl System) was developed. The modelling system is built upon the Aquatic Ecosystem Model (AEM3D) model, with predictive simulation capabilities developed and tested for a large lake (i.e., Lake Erie). The open-access web-based platform derives model forcing, code execution, post-processing and visualization of the model outputs, including water level elevations and temperature, is in near real-time. COASTLINES currently generates 240-h predictions using atmospheric forcing from 15 km and 25 km horizontal-resolution operational meteorological products from the Environment Canada Global Deterministic Forecast System (GDPS). Simulated water levels were validated against observations from 6 gauge stations, with model error increasing for longer forecast times. Satellite images and lake buoys were applied to validate forecast lake surface temperature (LST) and the water column thermal stratification. The forecast LST is as accurate as hindcasts, with a root-mean-square-deviation < 2 ℃. COASTLINES predicts storm-surge events and up-/down-welling events that are important for flood water and drinking water/fishery management, respectively. Model forecasts are available in real-time at https://coastlines.engineering.queensu.ca/. This study provides an example of the successful development of an operational forecasting system, entirely driven by open-access data, that may be easily adapted to simulate aquatic systems or to drive other computational models, as required for management and public safety.