Tracking lightning-induced mangrove canopy gaps in Panama using multi-temporal remote sensing

Lightning-induced canopy gaps are an important form of fine-scale disturbance in mangrove forests, with implications for regeneration, canopy structure, and forest heterogeneity. Reconstructing their long-term dynamics remains challenging because recovery is prolonged and historical high-spatial-resolution imagery is temporally sparse and heterogeneous among sensors. This study developed a multi-temporal remote-sensing framework to map and track lightning-induced mangrove canopy gaps in Panama from 2000 to 2020. Field observations were integrated with high-spatial-resolution IKONOS, QuickBird, WorldView-2, and SPOT 6/7 imagery. Relative radiometric normalization and pan-sharpening were used to construct a harmonized image sequence, followed by change-aware U-Net segmentation and object-based temporal registration. The change-aware model achieved a mean test Intersection over Union (IoU) of 0.527 and supported reconstruction of 2530 linked gap tracks comprising 6957 observations. Among 26 field-observed gaps evaluated for continuous persistence, 34.6% remained continuously detectable after eight years, while median Normalized Difference Vegetation Index (NDVI) increased from 0.388 to 0.550. Over the same period, median image-derived gap area decreased from approximately 286 m2 at the first available satellite observation to 118 m2 after eight years, although temporary increases at intermediate intervals indicated a non-monotonic trajectory. These patterns indicate concurrent changes in vegetation condition, spatial extent, and optical detectability during gap recovery rather than simple progressive contraction of gap boundaries. Overall, heterogeneous high-spatial-resolution satellite imagery can support reconstruction of long-term disturbance and recovery trajectories across irregular observation intervals, enabling fine-scale mangrove canopy dynamics to be examined over multi-decadal periods.