<![CDATA[Mangrove restoration is a global priority for enhancing coastal resilience, conserving biodiversity, and increasing blue carbon sequestration. However, identifying suitable restoration sites remains challenging due to the complex interactions among environmental, ecological, and anthropogenic factors. This study proposes an Explainable Artificial Intelligence (XAI)-based framework that integrates remote sensing and geospatial data to support transparent and evidence-based mangrove restoration planning. Multi-source spatial datasets, including Sentinel-2 imagery, digital elevation models, land use and land cover, tidal inundation, hydrological connectivity, soil characteristics, coastline dynamics, proximity to rivers and settlements, and historical mangrove distribution, were analyzed using Random Forest and Extreme Gradient Boosting (XGBoost) models. Model predictions were interpreted using SHapley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) to identify the relative contribution of each predictor. Simulation results for 15 candidate restoration zones in Kapetakan, Pangenan, Mundu, and Losari districts allocated a budget of IDR 902.13 million for restoring 46.86 ha. Three zones were recommended as optimal restoration sites with high success probabilities (0.629–0.954), six zones were excluded due to budget limitations, and six were rejected because of ecological constraints or predicted success probabilities below 0.40. The proposed framework enhances model transparency, strengthens stakeholder confidence, and supports informed coastal planning by providing interpretable, data-driven recommendations for sustainable mangrove restoration and climate adaptation]]>
Copyrights © 2026
