The Utility of InSAR Coherence for Tornado Damage Track Identification from the 2–3 March 2020 Tennessee Tornado Outbreak

Abstract Damage surveys after suspected tornadoes can be time-consuming for NWS forecasters to conduct. Suspected damage tracks can cover a large area with limited road networks and rugged terrain, making traditional surveying methods challenging. Satellite remote sensing imagery and derived products have frequently been leveraged to assist forecasters with damage surveys. Optical imagery can be obstructed by clouds and shadows, limiting when collections can be used for analysis. Synthetic aperture radar (SAR) instruments collect in all weather and are not currently incorporated into the damage survey process. This study uses interferometric SAR (InSAR) coherence to identify tornado damage tracks from the 2–3 March 2020 Tennessee tornado outbreak. Coherence damage paths (CDPs) were qualitatively identified from a single coherence pair. Physical characteristics of the identified CDPs were compared with survey data from the NOAA/NWS Damage Assessment Toolkit (DAT). Portions of five of the nine tornadoes analyzed with varying EF-scale ratings were identified. Over 70% of the total surveyed length was captured in CDPs for tornadoes rated ≥ EF-1. The coherence-derived lengths from each CDP showed strong agreement, but the width showed little to no agreement when compared to the offical survey. Lower-rated damage (≤EF-2) was associated with higher coherence values from 0.4 to 0.6, while higher-rated damage (≥EF-3) had lower coherence values closer to 0.3. The results from this study demonstrate the ability of coherence to provide supplemental information to NWS forecasters during the storm survey process, especially after large outbreaks in hard-to-access or inaccessible locations. Significance Statement Optical satellite observations have demonstrated the ability to identify tornado tracks and supplement ground surveys, but their utility is often limited by weather and solar conditions. This study investigates the use of synthetic aperture radar sensors, which collect independently of these conditions, to qualitatively identify tornado damage tracks from the 2–3 March 2020 Tennessee tornado outbreak. Five of the nine tornadoes analyzed from this event were identified, and over 70% of the surveyed track length ≥ EF-1 was identified. Future work will involve applying different change detection techniques to tornado outbreaks in the southeastern United States to better understand when this dataset can be beneficial to incorporate into ground surveys.