Enhancing Coastal Vegetation Mapping by Integrating Elevation with UAS-Based Multispectral Imagery: A Case Study from Jupiter Inlet Lighthouse ONA, Florida

Coastal vegetation plays a critical role in shoreline stabilization, habitat preservation, and maintaining biodiversity. Accurate mapping of these communities is essential for informed ecosystem management, particularly in ecologically sensitive coastal areas. This study investigates the application of machine learning for mapping vegetation communities using high-resolution multispectral imagery acquired by an unmanned aerial system (UAS)-mounted MicaSense RedEdge-MX sensor at the Jupiter Inlet Lighthouse Outstanding Natural Area (ONA), FL. Four dominant vegetation communities, oak scrub, palm, sand pine, and mangrove, were classified using both pixel-based and object-based approaches with support vector machine and random forest algorithms. To enhance classification performance in spectrally similar communities such as mangrove and sand pine, a high-resolution digital elevation model derived from airborne LiDAR (USGS 3DEP) was incorporated as an additional input feature. Results show that integrating elevation data improved classification accuracy by up to 15%, achieving a maximum overall accuracy of 90%. The inclusion of elevation was particularly effective in distinguishing ecologically distinct yet spectrally similar vegetation communities. These findings highlight the value of combining spectral and elevation data to improve coastal vegetation mapping and offer practical guidance for land managers and remote sensing practitioners in similar environments.