Mapping and assessment of vegetation types in the tropical rainforests of the Western Ghats using multispectral Sentinel-2 and SAR Sentinel-1 satellite imagery

Detailed mapping and regular monitoring of tropical rainforests is important for conservation and management of highly fragmented tropical rainforest habitats and biodiversity. Several studies have observed that it is highly challenging to map different vegetation types in tropical rainforests due to large environmental heterogeneity, high topographical variability and near constant cloud cover. In the present study, we assessed the capability of optical multispectral Sentinel-2 MSI bands, their derived NDVI and textures, and SAR Sentinel-1 bands and their textures to discriminate different vegetation types in the tropical rainforests of the Western Ghats using maximum likelihood and random forest classification. We also compared the results of our classification with previous such maps of the study area. Finally, we evaluated the magnitude of habitat fragmentation by using derived landscape metrics. Our classification had high accuracy (>75%), especially compared to previous classification efforts. Our results emphasise the significance of using vegetation indices and textures for vegetation type classification in the Western Ghats. Furthermore, the results suggest that rainforest habitats and other agro-ecosystems, suitable as habitats for numerous plant and animal species, are highly fragmented and require top conservation priority. High spectral and spatial resolution, continuity, affordability and access makes Sentinel-2 one of the best options for regular monitoring of tropical rainforests.