Browsing by Author "M. Arasumani"

Now showing 1 - 1 of 1

Mapping the fraction of invasive species to prioritise restoration areas of coastal wetland vegetation communities in a Ramsar site using Sentinel-2 images with machine learning regression
(Springer Nature, 2025) M. Arasumani; M. Kumaresan; Ayushi Gupta; A. R. Anaswara
Coastal wetlands are critically important ecosystems that support biodiversity, provide carbon storage, and deliver essential ecosystem services; however, these wetland ecosystems are increasingly threatened by invasive plant species such as Prosopis juliflora and Prosopis chilensis, which invade and reduce the native vegetation communities of tropical coastal wetlands. This study concentrated on fractional cover mapping of invasive and native vegetation of the Point Calimere Ramsar Site (PCRS), India, based on regression-based unmixing of Sentinel-2 spectral temporal metrics (STM) to conserve and restore the native vegetation communities in the PCRS. The entire seasonal Sentinel-2 STM was used to capture the native and non-native vegetation fractions. The synthetic training data were generated for each landcover and uploaded into the Google Earth Engine (GEE) to predict the fractional cover across the PCRS using the Support Vector Regression (SVR). The fractional cover maps effectively delineated the spatial extent of native vegetation communities, including mangroves, tropical dry evergreen forests, coastal grasslands, and mudflats, along with the distribution of invasive species such as P. juliflora and P. chilensis. Validation against field data and high-resolution satellite imagery showed a high accuracy of R2 values greater than 0.9 and less MAE < 10% for major native and non-native vegetation classes. In addition, a Restoration Priority Index (RPI) and an Invasive Species Removal Index (IRI) were developed to guide conservation efforts and invasive species management in the study area. The findings indicate that regression-based unmixing combined with spectral temporal measurements is an excellent method for monitoring and managing tropical coastal wetlands under invasion pressure. It also offers valuable insights into ecosystem restoration and biodiversity conservation. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2025.