Comparison of vegetation indices derived from a consumer-grade UAV and the Google Earth Engine platform for seasonal mangrove canopy assessment

Abstract

Mangrove forests, though classified as evergreen vegetation, exhibit substantial year-round litterfall, necessitating species-level phenological understanding for accurate remote sensing classification. This study compared monthly phenological variations across four mangrove classes viz., red (Rhizophora mangle), white (Laguncularia racemosa), and black mangrove (Avicennia germinans) in both shrub and fringe conditions, using consumer-grade UAV data and Sentinel-2 imagery from Google Earth Engine (GEE) over a 15-month period. Visible (RGB) data from UAV orthomosaics and Sentinel-2 were analysed using four RGB-vegetation indices, revealing similar temporal trends between platforms, though with consistently lower intensity values in UAV-derived data. Notably, the weakest correlations occurred in Rhizophora mangle and the shrub Avicennia germinans classes. These findings demonstrate UAVs´ potential for high-resolution phenological monitoring where spaceborne sensors lack sufficient resolution, while highlighting species-specific limitations for mangrove forest assessments.