Mapping mangrove forest degradation using lidar and Landsat data

• This paper proposes a novel approach for mapping mangrove forest degradation. • Degradation indicators were selected from mangrove forest structure and function. • Development of a new method to establish a spatial–temporal reference frame. • Improvement of hierarchical models to estimate large-area mangrove biomass. • The overall mangrove degradation in 2022 on Hainan Island was 30.78%. Monitoring mangrove forest degradation is challenging compared to deforestation because degradation involves the loss of ecosystem structure and function without changing forest cover, and selecting appropriate degradation references is difficult. Existing remote sensing time-series approaches offer a practical way to monitor forest degradation, but they mainly monitor degradation that occurred after images became available and focus on the forest’s horizontal dimension. This study proposes a novel approach for quantifying mangrove forest degradation. The approach evaluates the degree of mangrove degradation by comparing degradation indicators (canopy cover, canopy height, and aboveground biomass) with benchmark values inside environmentally homogeneous land capacity sub-regions. The benchmark values are determined by combining the LandTrendr algorithm and land capacity knowledge, accounting for temporal and spatial reference types. Large-area mangrove degradation indicators were estimated using improved hierarchical models with a “feed-forward” strategy. We applied this approach to mangrove forests on Hainan Island, China. The results showed that the overall accuracy of degradation identification was 95%, with a kappa coefficient of 0.90. The overall mangrove degradation degree in 2022 was 30.78% on Hainan Island, with reductions of 38.01% in canopy height, 28.04% in canopy cover, and 26.27% in aboveground biomass, indicating that mangrove dwarfing was the most serious degradation issue. Protected areas exhibited lower degradation (29.89%) than unprotected areas (35.31%). The findings also showed that the approach successfully detected degradation signals earlier than those from traditional satellite remote sensing data could to a certain extent (e.g., Landsat imagery dating back to ∼1985). The R 2 values (RMSE) for the canopy height, canopy cover, and biomass models were 0.72 (1.03 m), 0.85 (0.11), and 0.69 (43.99 Mg/ha), respectively. The proposed approach can be an alternative to time series-based methods when mangrove degradation has a long-term history and can provide critical information to support mangrove conservation and restoration efforts.