Cyanobacterial mats of eastern Exmouth Gulf, and the wider Pilbara intertidal region are largely unprotected within the environmental regulatory framework, primarily due to the inundation linked dormancy cycle and associated dynamic spatial extent and biological output. Here, we utilised supervised machine learning, with a combination of satellite imagery and ground truthing data to predict the spatial extent of cyanobacterial mats over time, accounting for tidal cycle, seasonality and freshwater input. Three distinct cyanobacterial habitat classes were identified, including, for the first time, identification of areas that may support cyanobacterial habitat if inundation conditions are conducive. These classes facilitate a more accurate environmental impact assessment, not only for spatial extent and direct loss, but the nutrient contribution and productivity to the Exmouth Gulf. Further, this method raises a novel approach to the management of habitats with dynamic boundaries, particularly for the design and implementation of management plans and environmental protection outcomes.