Analysis of the Informativeness of Spectral Vegetation Indices for Satellite Mapping of Open Sands and Deflated Territories of the Northwestern Caspian Sea

Abstract

Unfavorable hydrothermal conditions and irrational use of agricultural land, including excessive grazing pressure, lead to the degradation of steppe and desert landscapes up to the complete loss of vegetation and upper soil horizons. This serves as a prerequisite for the intensification of deflation processes and the emergence of dust storms, which provoke further growth in the area of open (moving, blowing) sands and areas devoid of vegetation. Remote sensing data, including spectral vegetation indices obtained on their basis, are widely used to monitor the prerequisites and consequences of intensive deflation processes. This study aims to assess the information content of 45 different vegetation indices calculated using Sentinel-2 and Landsat satellite data for mapping open sands in the Northwestern Caspian region, where degradation processes of pastures on sandy and sandy loam soils have intensified in the last decade. As a result of comparison with aerial survey data on three test sites with a total area of 23.1 hectares, which was carried out three times during 2023, the most informative vegetation indices were established. For Sentinel-2 data with a spatial resolution of 10 m, NDVI and spectral radiance coefficients in the visible range are recommended, and for data with a resolution of 20 m and Landsat, vegetation indices based on the visible and shortwave infrared ranges (NDSAI, BUI, NBR2, BAEI) are recommended. The results of this study can be used in mapping open sands using satellite remote sensing data and developing automated methods for their classification, including those based on machine learning.