Observational evidence for exponential tornado intensity distributions over specific kinetic energy

Abstract

[1] Observational evidence supports the recent analytical prediction that tornado intensities are exponentially distributed over peak wind speed squared (v 2 ), or equivalently, Rayleigh-distributed over v. For large USA data samples, exponential tails are found in the tornado intensity distributions over v 2 from about F2 intensity on. Similar results follow for smaller worldwide data samples. For the 1990s data from the USA and Oklahoma, deviations from the Rayleigh distribution for weak tornadoes can be explained by the emergence of a separate, likely nonmesocyclonic tornado mode. These bimodal datasets can be modeled by superposition of two Rayleigh distributions. The change in modal dominance occurs at about the F2 threshold (v % 50 m s À1 ). In France, likely mainly the mesocyclonic tornado mode has been recorded, while in the UK, only a non-mesocyclonic mode seems to be present. Citation: Dotzek, N., M. V. Kurgansky, J. Grieser, B. Feuerstein, and P. Névir (2005), Observational evidence for exponential tornado intensity distributions over specific kinetic energy, Geophys. Res. Lett., 32, L24813,

The spatial distribution of severe convective storms and an analysis of their secular changes

“…substituting v ≡ x-a, and v 0 ≡ b, Dotzek et al (2005) were able to corroborate and generalize results by Kurgansky (2000) who had analytically predicted a Rayleigh distribution for tornado intensity distributions over wind speed. Comparing the evolution of parameter c from general Weibull distribution fits worldwide, Dotzek et al (2005) showed that for increasing quality and sample size of tornado intensity datasets, a trend towards c = 2 can indeed be concluded.…”

“…Brooks and Doswell (2001b) noted that the distribution of tornadoes by intensity appeared similar in many countries, implying that there might be an underlying physical process (or set of processes) that leads to the intensity of a tornado. Since that time, Dotzek et al (2003Dotzek et al ( , 2005 and Feuerstein et al (2005) have worked to put that speculation on a firmer statistical footing. Dotzek et al (2003) and Feuerstein et al (2005) showed that Weibull distributions could be fit to the data from a variety of countries worldwide and applied a two-parameter leastsquare fit to observed worldwide tornado intensity distributions both in F-scale and wind speed v. With x denoting either of these, the Weibull distribution is given in three-parameter form for probability density p(x):…”

Steady‐state properties and statistical distribution of atmospheric dust devils

“…To make such a comparison, hDi-value in equation (7) has been calculated from the observational data; it provides an example of the moment estimation technique [cf. Dotzek et al, 2005]. A rough a posteriori L 0 -estimate for typical desert conditions in Arizona and southern California [Kurgansky, 2006] confirmed plausibility of the conjecture (B).…”

“…A rough a posteriori L 0 -estimate for typical desert conditions in Arizona and southern California [Kurgansky, 2006] confirmed plausibility of the conjecture (B). To independently demonstrate a good match of equation (7) and Sinclair's [1966] data (for the Tucson Basin and the Avra Valley in Arizona) we fit these data by a two-parameter Weibull distribution Dotzek et al, 2005], written in terms of the probability P for the vortex diameter to exceed a given D-value. Two parameters x and D 0 are specified from the best fit to observations; equation (7) corresponds to x = 1.…”

On the implementation of the enhanced Fujita scale in the USA

“…Some of these changes were intentional, while others were not. The implications for continued applicability of comparisons of ratings across time and space that we have been involved in the past are troubling (e.g., Brooks and Doswell, 2001;Dotzek et al, 2003Dotzek et al, , 2005Feuerstein et al, 2005).…”