Simulating Tornado Probability and Tornado Wind Speed Based on Statistical Models

Abstract

This study presents the development and testing of two statistical models that simulate tornado potential and wind speed. This study reports on the first-ever development of two multiple regression–based models to assist warning forecasters in statistically simulating tornado probability and tornado wind speed in a diagnostic manner based on radar-observed tornado signature attributes and one environmental parameter. Based on a robust database, the radar-based storm-scale circulation attributes (strength, height above ground, clarity) combine with the effective-layer significant tornado parameter to establish a tornado probability. The second model adds the categorical presence (absence) of a tornadic debris signature to derive the tornado wind speed. While the fits of these models are considered somewhat modest, their regression coefficients generally offer physical consistency, based on findings from previous research. Furthermore, simulating these models on an independent dataset and other past cases featured in previous research reveals encouraging signals for accurately identifying higher potential for tornadoes. This statistical application using large-sample-size datasets can serve as a first step to streamlining the process of reproducibly quantifying tornado threats by service-providing organizations in a diagnostic manner, encouraging consistency in messaging scientifically sound information for the protection of life and property.

Tornado damage ratings estimated with cumulative logistic regression

“…This study makes the case that cumulative logistic regression is the right tool for quantifying the combined role environmental factors play on the distribution of tornadoes by EF rating. It might be tempting to fit a simpler model to these data as was done in Cohen et al (2018) who suggested that simulated tornado wind speeds from their model can be scaled within the context of the damage ratings. But it is unclear how this can be done while preserving the relative frequency of ratings given that the model residuals are assumed to be described by a normal distribution centered about the conditional mean wind speed.…”

“…These advances come from a better understanding of relationships between near-storm regional-scale environmental conditions and the resulting mode of convection [see Smith et al (2012) and Thompson et al (2012) for a review of the literature on this topic] and from careful statistical analysis of relationships between radar-based rotational signals at the storm scale and the probability of specific damage rating categories (Smith et al, 2015;Thompson et al, 2017). Cohen et al (2018) investigated multivariate models as a way to combine various environmental and storm-scale factors influencing the probability of specific enhanced Fujita (EF) ratings. While this approach is an improvement over earlier bi-variate methods (e.g., box-and-whisker plots), more work is required to put the research onto a solid statistical foundation.…”

“…Mathematically the approach we take is similar to that outlined in Cohen et al (2018), who fit a linear regression to wind speeds corresponding to midpoints of the EF rating intervals. But our approach differs in that we fit a cumulative logistic regression model to the recorded highest EF rating directly.…”

Techniques and Thresholds of Significance for Using WSR-88D Velocity Data to Anticipate Significant Tornadoes

“…Thompson et al (2017, hereafter T17) followed up on this work by increasing the dataset to include a robust sample of nontornadic severe thunderstorms, with an increased focus on reducing estimation errors as a result of WSR-88D data quality issues like vertical sidelobe contamination, which has proven to be a significant challenge when trying to use the radar as precisely as tornado intensity estimation requires. Cohen et al (2018) summarized many of these findings into statistical models designed to diagnose tornado intensity in real-time.…”

Early Operational Successes of the University of Louisiana Monroe’s Polarimetric S-band Doppler Radar

“…Improvements in coverage, warning performance, and comparisons with the nearest WSR-88D are shown for three high-impact events that occurred in the ArkLaMiss during 2017 and 2018. To provide an objective measure of improvement, simulated tornado probabilities and intensities (Thompson et al 2017;Cohen et al 2018) were calculated and compared for each radar. Although KULM has led to significant improvements in radar coverage, a gap-filling, non-NWS radar is not without limitations and challenges within the operational NWS environment.…”