Evaluation of UAV and satellite-derived NDVI to map maritime Antarctic vegetation

• Published in: Applied Geography
• Main Authors: Sotille, Maria E., Bremer, Ulisses F., Vieira, Gonçalo, Velho, Luiz F., Petsch, Carina, Simões, Jefferson C.
• Format: Article in Journal/Newspaper
• Language: English
• Published: Elsevier 2020
• Subjects: Vegetation mapping; NDVI; Antarctica; UAV; Landsat Sentinel-2; Remote sensing; Antarctic; Antarctic Peninsula; Hope Bay; The Antarctic; Antarc*
• Online Access: http://hdl.handle.net/10451/44561; https://doi.org/10.1016/j.apgeog.2020.102322

Expansion of Antarctic vegetation in ice-free areas underlines the need for effective remote sensing techniques to properly monitor the changes. Detection and mapping of vegetation remains limited in the Antarctic environment given the complexity of its surface coverage. Some cryptogamic species exhibit low reflectance in the nearinfrared region and are not easily detected by vegetation indices, such as the normalized difference vegetation index (NDVI). In addition, spectral reflectance of Antarctic vegetation is highly variable according to seasonal conditions, which may influence NDVI results. As ultra-high resolution aerial imagery allows for a detailed analysis of vegetation and enables the validation of satellite imagery, in this study we assess the ability of the NDVI from unmanned aerial vehicle (UAV), Sentinel-2, and Landsat 8 to identify vegetated areas in the ice-free environment of Hope Bay, Antarctic Peninsula. NDVI classification with class ranges set by statistical parameters (i.e., mean and standard deviation) is performed. The results show that different sensors provide different NDVI values for the same vegetation class. NDVI classification enabled the identification of areas showing vegetation cover, which are in accordance with the manually mapped areas in the UAV image. Correspondence in vegetation distribution and classes can be observed across all classifications, demonstrating that aerial and satellite imagery may be used for Antarctic vegetation monitoring. A close association between NDVI classes and Antarctic vegetation type is identified, where lichens are generally classified in lower probability classes, and algae and moss in higher probability classes. This article shows the potential of NDVI applied to Antarctic vegetation and the significance of data statistical parameters in the selection of thresholds, reducing the need for groundtruth information in remote areas. info:eu-repo/semantics/publishedVersion