The mapping and monitoring of habitats and vegetation conditions through remote sensing and machine learning is an important aspect of managing our environments at scale. New remote sensing platforms allow for fine-scale mapping of habitats at regional and potentially even global scales. However, mapping marine ecosystems with remote sensing is challenging and we lack scalable methods to assess the extent and changes in coastal marine habitats reliably. Here, we present a contemporary approach for mapping submerged vegetation using a novel aquatic vegetation index derived from the Sentinel-2 MultiSpectral Instrument. Using spatially balanced survey designs and remote underwater imagery, we test methods to classify submerged aquatic vegetation at regional geographic scales. Our results indicate that this novel index provides a consistent, scalable and accurate method for mapping submerged vegetation. Using several different validation approaches, models that included the novel vegetation index achieved almost perfect agreement (Cohen’s kappa > 0.80) for classifying submerged vegetation. We demonstrate the applicability of this novel vegetation index for mapping aquatic vegetation across the mid-west coastal environment of Western Australia. This region supports some of the most extensive kelp forests, seagrass beds and most valuable invertebrate fisheries on the planet.