Synergistic use of ICESat-2 lidar data and Sentinel-2 imagery for assessing hurricane-driven forest changes

Coastal forests worldwide are vulnerable to hurricanes, which cause significant canopy loss and long-term disruption of ecosystem services. Scalable methods for assessing hurricane-driven forest damage are critical for ecosystem recovery, yet conventional field-based approaches are time-consuming, costly and challenging to acquire over large areas. The Ice Cloud and land Elevation Satellite-2 land and vegetation height product (ATL08) provides three-dimensional information, and its capability of measuring forest structure is already well demonstrated. However, its integration with satellite imagery for mapping hurricane-induced canopy height changes remains limited. In this study, we analyzed coastal forest canopy losses and thematic transitions using ICESat-2, Sentinel-2, and ancillary predictors. We used Random Forest (RF) and Extreme Gradient Boosted (XGB) regression models to extrapolate ATL08 pre-hurricane canopy heights and applied the better-performing model for post-hurricane mapping. The resulting canopy height maps were combined with existing land cover products to assess structural and thematic transitions across the impacted landscape. RF outperformed XGB (R² = 0.44, RMSE = 4.30 m vs. R² = 0.41, RMSE = 4.76 m). Landcover shifts included transitions from evergreen to herbaceous, scrub, and barren classes, and from woody wetlands to emergent herbaceous wetlands, with mean canopy height losses of 2.3-5.2 m. Canopy cover analysis showed dense (> 60%) and sparse (< 30%) cover experienced greatest losses (up to 8.3 m), while moderate covers (30-60%) were resilient. This study demonstrates the potential of integrating ICESat-2 and Sentinel-2 for assessing structural and thematic changes and informing adaptive strategies in hurricane-prone coastal ecosystems.