Synthetic Aperture Radar as a Tool for Tornado Classification in Forested Areas

Canada’s tornado climatology has historically been challenging to accurately assess. Many Canadian tornadoes are currently rated on the Enhanced Fujita scale using treefall as a damage indicator, though this approach may occasionally underestimate tornado intensity. In this work, we propose that the use of synthetic aperture radar (SAR) images of tornado tracks may lead to more accurate tornado ratings in tornadoes that have at least 25% of their track consisting of forested areas. We create a replicable methodology by which SAR data can be used to classify tornadoes by generating difference images using Sentinel-1 C-Band radar data acquired before and after the tornado event. We apply this methodology to a large number of recent Canadian tornadoes. The results showed that only higher intensity tornadoes (EF3+) were detectable in the SAR change detection images. Critically, the results revealed possible misclassifications, where tornadoes originally assigned a rating of EF2 based on treefall damage indicators were visible in the SAR imagery. Our study provides a foundation for future work, providing an improved methodology for classifying tornadoes in remote forested areas.