Comparison of Tornado Damage Characteristics to Low-Altitude WSR-88D Radar Observations

Recent studies have indicated that the strongest horizontal wind speeds in tornadoes are often concentrated very close to the surface (e.g., Kosiba and Wurman 2023) and that many tornadoes are stronger than they are rated, with as much as 20% of tornadoes produced by supercells capable of producing EF4–EF5 damage (Wurman et al. 2021). A significant fraction of tornado intensity underestimation is often attributed to a lack of damage indicators from which higher tornado intensities can be estimated (e.g., Dahl et al. 2017; Lombardo et al. 2023). The mobile radar datasets that underpin these studies are almost exclusively from tornadoes observed on the Great Plains, a region marked by vast swathes of open, relatively flat land, which is ideal for deploying mobile radars but challenging for estimating near-surface tornado intensity from damage. Over the past decade, the National Weather Service (NWS) has begun utilizing the Damage Assessment Toolkit (DAT) to gather more detailed characteristic information along tornado tracks. Furthermore, the NWS has begun utilizing the Supplemental Adaptive Low-Level Intra-volume Scan (SAILS) and Multiple Elevation Scan Option for SAILS (MESO-SAILS) scanning techniques across the WSR-88D network to increase low-level sampling of severe convective storms. The combination of the DAT and increased low-level sampling has led to numerous low-level radar samples of tornadoes at close range to WSR-88D radars that can be coupled with detailed damage assessment information for intercomparison across regions of varying population density, land cover/land use, and topography.

This study analyzes 195 observations from 105 different tornadoes collected from WSR-88D radars at elevations at or below 150 m AGL and coupled with DAT damage information points during the period from 2011–2023. The median tornado peak wind speed profile from Kosiba and Wurman (2023) is then applied to the radar observations to estimate the peak tornado wind at 15 m AGL, and those estimates are compared to the peak damage intensities associated with each radar observation. Consistent with mobile-radar-based studies, radar predictions of near-surface wind speeds tend to be higher than damage-based wind speed estimates from the Enhanced Fujita (EF) scale, particularly for stronger tornadoes. Intriguingly, errors between radar-estimated and damage-estimated wind speeds are only weakly modulated by the density of quality damage indicators, as estimated through U.S. Census housing unit density data. Implications for EF-scale and radar-based wind speed estimates of tornadoes are discussed.