Spatial Analytic Framework-Based Prediction and Analysis of Cyclone-Resilient Mangrove Species with Sentinel-2 Data

Mangroves serve as critical bio-shields protecting shorelines from environmental hazards, making species-level assessment vital for conservation. This study aims to quantitatively assess and compare the resilience of mangrove species in Henry Island, West Bengal, India, by integrating Sentinel-2 multispectral imagery and spectral unmixing with Fractal Dimension, Entropy, and Moran’s I to quantify species-specific changes in structural complexity, spatial fragmentation, and spatial clustering associated with environmental disturbances and subsequent recovery. Starting with GPS-based ground surveys for species identification and abundance estimation, imagery was processed and endmembers were extracted to capture temporal spectral variations. Spectral Unmixing was then performed using Non-Negative Least Squares (NNLS) and Simulated Annealing (SA) sub-pixel mapping algorithms. The resulting satellite-derived abundance maps were validated directly against the ground-truth plots, yielding high accuracy confirmed via Root Mean Square Error analysis. To evaluate species-specific cyclone vulnerability and post-disturbance recovery, spatial distribution patterns were quantified using spatial metrics, Fractal Dimension (FD), Entropy, and Moran’s I, which mathematically translate mapping data into indicators of ecological stability and revealed distinct disparities in resilience. Calculating FD for boundary complexity, Entropy for fragmentation, and Moran’s I for spatial clustering is critical because spectral abundance alone cannot capture structural vulnerability. Synthesizing these metrics enables an empirical, comparative evaluation of how different species physically resist and recover from environmental stress. The Avicennia community and Excoecaria agallocha exhibited the highest resilience, maintaining maximum structural complexity and minimal fragmentation. Conversely, the Ceriops community suffered severe initial canopy fragmentation but demonstrated robust long-term natural recovery. Casuarina equisetifolia and Phoenix paludosa displayed intermediate to low-moderate resilience with slower, site-specific regeneration. The proposed framework bridges ground observations with advanced sub-pixel analysis to provide a quantitative, scalable, and transferable approach for assessing mangrove resilience. By integrating advanced analysis techniques and validation, it offers a reliable decision-support tool for long-term ecosystem monitoring, biodiversity conservation, and climate-resilient coastal management worldwide.