Post-DORIAN forest damage assessment in the Prince Edward Island National Park, Canada employing multi-sensor satellite data

Abstract

Cyclones have been considered one of the major natural disasters for decades. Particularly, tropical cyclones are more devastating in terms of damage (e.g., physical, environmental, economic). In the present study, we have investigated the forest damage that occurred due to the intense category 5 catastrophic hurricane Dorian that struck one of the Atlantic Provinces, Prince Edward Island (PEI), from August 24 to September 10, 2019. We have employed multi-sensor satellite remote sensing datasets, including Sentinel-1 Synthetic Aperture Radar (SAR) and Sentinel-2 Multispectral Instrument (MSI), as well as very high-resolution commercial satellite imagery, ground reference points, secondary references, and local knowledge. We have utilized four widely used spectral indices (SIs), including normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), normalized difference infrared index (NDII), red-edge spectral indices (RESI), and RADAR Vegetation Index (RVI). First, we assessed the forest damage using a conventional simple method, where the damage area was estimated by subtracting the post-Dorian and pre-Dorian imagery. Second, we have developed an algorithm based on the statistical analysis of the ground reference points and associated vegetation indices and RVI values, as named author derived decision tree (ADDT) method. Finally, random points were generated in the ArcGIS platform, and an accuracy assessment was performed. The height accuracy has been found for the RVI (94.74 %) using the ADDT method, which is comparatively promising. The proposed algorithm will help researchers/scientists to estimate forest damage in varied geographical settings.