The South Aral Seabed is an extreme dryland ecosystem undergoing rapid transformation yet remains misrepresented or absent in global land cover datasets. Conventional vegetation indices, specifically the Normalized Difference Vegetation Index (NDVI), perform poorly in such environments due to their limited ability to distinguish sparse vegetation from highly reflective saline and sandy soils. This study evaluated the effectiveness of the Modified Soil Adjusted Vegetation Index (MSAVI) for improving land cover classification in the South Aral Seabed and conducted a decadal analysis of land cover change between 2013 and 2023 using Landsat 8 imagery (30 m resolution). A spectral index-based classification framework was developed, combining MSAVI with the Normalized Difference Water Index (NDWI) and Salinity Index 1 (SI1) to reduce spectral confusion between vegetation, saline soils, and surface water. The MSAVI-based classification achieved an overall accuracy of 77.96% (Kappa coefficient=0.71), supported by 313 field-collected validation points from 2023. While the multi-index approach enabled finer discrimination of ecologically important classes, particularly separating salt pans from solonchak soils, it resulted in a lower overall accuracy (73.80%), highlighting a trade-off between class separability and classification performance. Land cover change analysis revealed a highly dynamic landscape, with 52.96% of the study area transitioning between classes over the decade. Transformed areas (16,893 km 2 ) exceeded stable zones (15,004 km 2 ), driven primarily by rapid desiccation and salinization. Solonchak soils increased at an annual rate of 5.58%, while surface water bodies declined by 4.83% per year. Concurrently, sparse or distressed vegetation increased by 1.43% annually, reflecting ongoing afforestation efforts. This study provides the first MSAVI-based and medium-resolution land cover baseline for the South Aral Seabed and demonstrates that soil-adjusted vegetation indices are essential for reliable dryland classification where conventional indices fail. The proposed spectral index framework offers a replicable methodology applicable to other global drylands facing similar land degradation and restoration challenges.
Alikhanova et al. (Sun,) studied this question.