Integrating UAV and Satellite Multispectral Remote Sensing to Monitor Terrestrial Vegetation Dynamics in the Fildes Peninsula

Recent studies report a rapid expansion of vegetation cover in Antarctic ice-free regions; however, relying solely on satellite observations introduces significant uncertainties requiring rigorous calibration and validation. For example, discrepancies in the Normalized Difference Vegetation Index (NDVI) derived from independent sensors—such as Landsat-8 and Sentinel-2—can lead to highly inconsistent assessments of vegetation cover and greenness. To address this, this study integrates Unmanned Aerial Vehicle (UAV) and satellite multispectral remote sensing to provide a reliable, accurate, and multi-scale quantification of the spatiotemporal dynamics of vegetation over large regions in Antarctica. Our case study focused on the Fildes Peninsula and its nearby islands, located at the southwestern part of the King George Island, is a typical ice-free region in sub-Antarctic. In February 2026, we deployed DJI Mavic3 multispectral edition to collect 5-cm-resolution aerial images over southern Fildes Peninsula, Ardley Island and Nelson Island. The images encompassed diverse settings including typical moss beds, bare grounds, penguin habitats, lakes, rivers and pro-glacier regions. They provide important data for a regression-based calibrate for Landsat and Sentinel-2 multispectral images. After calibration, Landsat-series and Sentinel-2 images are used to construct decadal time series of terrestrial vegetation cover in the past decades. The remote sensing results will be used together with field-based and laboratory measurements of carbon, nutrients, and containments in vegetation, soil, and water for understanding large-scale biogeochemical processes in rapidly-changing Antarctic terrestrial ecosystems.