An ML-Based Ensemble Approach for the Precision Classification of Mangroves, Trend Analysis, and Priority Reforestation Areas in Asir, Saudi Arabia

In the recent past, mangrove ecosystems have undergone significant transformation, necessitating precise classification, the assessment of ecological changes, and the identification of suitable sites for urgent replantation. Therefore, this study aims to address three key objectives: first, to map the current extent of mangroves; second, to assess the ecological changes within these ecosystems; and third, to identify suitable areas for replantation, ensuring their sustainability across coastal Asir. The mangrove classification was conducted using an ensemble of machine learning models, utilizing the key spectral indices from Landsat 8 data for 2023. To analyze the ecological trends and to assess the changes over time, Landsat 5–8 data from 1991 to 2023 were used. Finally, a generalized additive model (GAM) identified the areas suitable for reforestation. The EC identified the mangrove area as 14.69 sq. km, with a 95.6% F1 score, 91.3% OA, and a KC of 0.83. The trends in the NDVI and LST increased (p = 0.029, 0.049), whereas the NDWI showed no significant change (p = 0.186). The GAM model demonstrated a strong fit (with an adjusted R2 of 0.89) and high predictive accuracy (R2 = 0.91) for mangrove priority reforestation suitability, confirmed by a 10-fold cross-validation and minimal bias in the residual diagnostics. The suitability varied across groups, with Group (e) showing the highest suitability at 77%. Moran’s I analysis revealed significant spatial clustering. This study provides actionable insights for mangrove reforestation, supporting the for sustainable development through targeted efforts that enhance ecological resilience in coastal regions.