Multi-sensor geospatial modelling to address complex mangrove dieback: misattribution of chemical stressors versus physical impact.

Accurately identifying the causes of mangrove dieback is essential for global coastal management, yet visually similar dieback events can result in incorrect attribution of causes. A significant mangrove dieback in Boambee Creek Estuary (2021) was initially linked to chemical contamination from nearby Coffs Harbour Airport. This study uses a forensic geospatial reconstruction to test the validity of this toxicity hypothesis against a physical-hydrological alternative. Scaling up from localised, site-specific field observations, we used multi-sensor satellite telemetry (Sentinel-2), LiDAR-based geomorphic modelling and ERA5 climatological reanalysis to assess the ecosystem at the landscape scale. The investigation pinpoints a statistically extreme hailstorm on 20 October 2021 as the trigger event, representing a > 1-in-100-year anomaly (Z = 4.10σ). Time-series diagnostics confirm an immediate structural collapse (p < 0.001) coinciding with the storm, ruling out the signature of gradual chemical ageing. Mortality followed a Death Curve, where the likelihood of death neared 100% at elevations below 1.5 m AHD (p < 10-8) within stagnant topographic basins (< 2° slope). Hydrological routing shows that the main runoff from the airport flows directly into the remaining forest, creating a Runoff Paradox that statistically discredits the chemical vector hypothesis. We conclude that the dieback is best explained by a compound disturbance framework, primarily triggered by a hydrological trap. Sudden physical defoliation halted canopy transpiration, causing rapid soil anoxia and root drowning in geomorphically unstable basins. Secondary chemical stressors may inhibit long-term recovery. Future management should focus on restoring hydrological connectivity rather than chemical remediation. This study highlights the vital need to incorporate landscape-scale multi-sensor remote sensing (Optical, Radar and LiDAR) for validating localised field sampling and accurately diagnosing heterogeneous dieback events worldwide.