The Relationship between Automated Low-Level Velocity Calculations from the WSR-88D and Maximum Tornado Intensity Determined from Damage Surveys

Abstract

The relationship between automated low-level velocity derived from WSR-88D severe storm algorithms and two groups of tornado intensity were evaluated using a 4-yr climatology of 1975 tornado events spawned from 1655 supercells and 320 quasi-linear convective systems (QLCSs). A comparison of peak velocity from groups of detections from the Mesocyclone Detection Algorithm and Tornado Detection Algorithm for each tornado track found overlapping distributions when discriminating between weak [rated as category 0 or 1 on the enhanced Fujita scale (EF0 and EF1)] and strong (EF2–5) events for both rotational and delta velocities. Dataset thresholding by estimated affected population lowered the range of observed velocities, particularly for weak tornadoes while retaining a greater frequency of events for strong tornadoes. Heidke skill scores for strength discrimination were dependent on algorithm, velocity parameter, population threshold, and convective mode, and varied from 0.23 and 0.66. Bootstrapping the skill scores for each algorithm showed a wide range of low-level velocities (at least 7 m s−1 in width) providing an equivalent optimal skill at discriminating between weak and strong tornadoes. This ultimately limits identification of a single threshold for optimal strength discrimination but the results match closely with larger prior manual studies of low-level velocities.

Dissipation Characteristics of Tornadic Vortex Signatures Associated with Long-Duration Tornadoes

“…1a) prior to tornado dissipation (e.g., Wurman and Gill 2000;Wurman et al 2007a,b;Marquis et al 2008;Kosiba et al 2013;French et al 2014;Houser et al 2015;Bluestein et al 2016). However, TVS intensity in coarse-resolution WSR-88D data is unlikely to reflect the strength of tornadic winds (e.g., Brown et al 1978;Burgess et al 1993;Wood and Brown 1997;Trapp et al 1999;Kingfield and LaDue 2015). In addition, tornado weakening is sometimes observed prior to reintensification and followed by prolonged tornado persistence (e.g., Kurdzo et al 2015) or the weakening is unsteady and/or abrupt (e.g., Davies-Jones et al 1978;Burgess et al 2002;.…”

Simulating Tornado Probability and Tornado Wind Speed Based on Statistical Models

“…Such flaws include an underrating bias, which has direct implications on the opportunity for this regression to represent ''truth.'' Kingfield and LaDue (2015), Strader et al (2015), and Alexander and Wurman (2008) all collectively highlight some of the flaws of the tornado database. While this database is the official documented source of tornado information, corresponding results from the statistical models need to be considered within the context of the flaws of the tornado database.…”

A skill assessment of techniques for real-time diagnosis and short-term prediction of tornado intensity using the WSR-88D

“…The IBW system differentiates tornado events by weak and strong/violent events. This guidance is built off of decades of science, but focuses on recent work quantifying the relationship between tornado damage intensity and subjectively analyzed WSR-88D volumetric products-specifically V rot (Kingfield and LaDue 2015;Smith et al 2015) and TDS height (Entremont and Lamb 2015;Van Den Broeke and Jauernic 2014). This study, evaluating 432 surveyed damage points from 84 different tornadoes, finds that the guidance is skillful.…”

“…In an attempt specifically to address needs related to IBW implementation, Kingfield and LaDue (2015) quantified the skill that automated low-level radar rotational velocity [V rot = (V max -V min )/2] calculations have in discriminating between weak and strong tornadoes. The authors found that skill scores for V rot peaked between 18 and 25.72 m s -1 (35 and 50 kt) in quality controlled data, but noted that considerable overlap still existed between strong and weak tornado events.…”