Rapid Remote Sensing Updates of Landcover for Storm and Hurricane Forecasts Training and Test Dataset

Coastal vegetated ecosystems play a vital role in reducing hurricane impacts on nearby communities. Different types of vegetation and landcover create varying levels of drag that help lessen hazards such as flooding and erosion. However, coastal landcover is constantly changing due to storms, tides, waves, and human activities. Existing satellite landcover products, such as the annual NLCD or C‑CAP, lack the spatial and temporal resolution needed to capture these rapid changes. To better support coastal modelers, high‑resolution, spatially detailed Manning’s n maps are needed to estimate the drag forces produced by different landcover types. To meet this need, we developed a random forest landcover classification model capable of generating on‑demand, near‑daily, 3‑meter resolution coastal landcover maps for project areas across the North Carolina Outer Banks, the Mississippi River Delta, and parts of the Florida Gulf Coast. The model was trained using harmonized Planet SuperDove 8‑band imagery. This data release includes the training and test datasets used in model development, consisting of Planet 8‑band surface reflectance data and corresponding landcover labels. To account for seasonal vegetation changes, the release includes training and test points for both cool‑season and warm‑season models.