Temporal change detection of mangrove species utilizing Sentinel-2 Satellite data for various study sites in India.

Abstract

The mangrove ecosystem, considered fragile, requires regular monitoring due to its crucial role in coastal ecosystems. This study focuses on conducting a detailed, species-level analysis of mangrove vegetation over time, utilizing moderately high-resolution Sentinel-2 satellite data. Supervised classification utilizing Random Forest (RF) algorithm was used to discriminate mangrove vegetation on a species level. Upon investigation we could find the east coast mangroves have a higher species diversity compared to west coast. Only two true mangrove species such as Avicennia marina and Rhizophora mucronate were common mangrove species for each coast. The exceptional spatial and temporal resolution of Sentinel 2 has proven crucial in detecting nuanced alterations in land cover that would have otherwise been challenging to identify using conventional approaches. The field hyperspectral data collected used as the reference spectra gives insights to the reflectance spectra of these mangrove vegetation. Although most of the sites studied shows a conservation in mangrove vegetation, it is observed that few stands of mangroves are converted to other mixed mangrove patches within the study time. In study sites such as Thane and Pichavaram, a decrease in the mangrove vegetation cover accompanied by mud deposition was observed. In the Indian Sundarban site Lothian Island, the occurrence of new stands of vegetation was observed. Validation of the study is conducted by field surveys and visualization using google earth pro. Species mapping is done by Ground Truths (GTs) reported by onsite visits and studying the existing literature. Normalised Difference Vegetation Index (NDVI), changes are monitored for the study years, both seasonally and yearly to conclude on the health of vegetation and to correlate with the changes.