From 1 - 1 / 1
  • Plot data Mc Nemar: To enable comparisons with the 1961 and 2004 survey results, the Lambert Conformal Conic projection from the 2004 survey was used to precisely georeference and trim the RGB image across a 1-m2 grid, generating a total of 3,458 1-m2 grid cells. For each grid cell moss, lichen, or algae/cyanobacteria cover was extracted as one of the four cover classes: Heavy (>40%), Patchy (10–40%), Scattered (less than 10%), and None (0%) for the survey years 1962, 2004 and 2018. Ground truthing: To test the overall accuracy of cover classifications and ensure consistency with 2004 survey methodologies, a ground-truthing approach was performed. Photographs were taken of individual cells along eight transects, running west to east across the plot at 0.5, 1.5, 15.5, 16.5, 28.5, 29.5, 116.5 and 117.5 m distance from the NW corner. Each grid cell could be identified individually with an x/y coordinate in the centre and was surrounded by a rectangular frame parallel to the outer edge of the plot. A total of 174 photographs were taken and archived with Antarctica New Zealand. For each photographed grid cell, the presence of each functional group of vegetation and their cover class was assessed visually. Orthomosaic image: Aerial images were obtained using a DJI Matrice 600 Pro hex-rotor remotely piloted aircraft system equipped with a Canon EOS 5Ds camera (image size: 8688×5792 pixels, focal length: 50 mm, pixel size: 4.14 μm) on November 28, 2018. The flight altitude was 30 m above ground level, and a total of 10 ground-control points were included to provide accurate geo-referencing. An orthomosaic photo and accompanying DEM was generated with the acquired aerial images using Agisoft PhotoScan (now known as Metashape by Agisoft LLC, https://www.agisoft.com/) RELATED PUBLICATION: https://doi.org/10.1029/2022EF002823 GET DATA: https://doi.org/10.7488/ds/3417