Mangrove canopy-height estimation using LiDAR and multispectral remote sensing: linking spatiotemporal species composition to restoration baselines

Abstract

Canopy height is an important indicator of mangrove forest structure, ecological functioning, and restoration development; however, field measurements are spatially limited by cost, accessibility, and survey effort. This study mapped mangrove canopy height in Trat Province, Thailand, by integrating ICESat-2 photon-counting LiDAR with Sentinel-1 synthetic aperture radar (SAR), Sentinel-2 multispectral imagery (MSI), and field measurements. Previously reported mangrove species-community and stand-history layers, derived from observations spanning 1987–2022, were used to stratify model sampling and interpret canopy-height patterns rather than as predictors. A random forest model was used to relate ICESat-2 relative-height metrics (RH10–RH98) to multisensor predictor variables, and its predictions were evaluated against field-plot measurements. Among the tested relative-height metrics, RH55 showed the best agreement with field measurements, with a root mean square error (RMSE) of 2.80 m, coefficient of determination (R2) of 36%, and mean absolute percentage error (MAPE) of 30%. Comparison with four global canopy-height products showed broadly comparable performance, with the proposed approach achieving the lowest RMSE (2.80 m). Canopy height varied substantially among species communities and stand histories. Although restoration increased mangrove extent, rehabilitated stands generally exhibited limited structural development. Additionally, the site-calibrated approach captured fine-scale height heterogeneity in scrub and rehabilitated zones. Integrating spaceborne LiDAR with SAR, MSI, and independently derived ecological baselines provides a scalable framework for characterizing spatial variation in mangrove canopy structure and evaluating restoration development.