Current tornado detection algorithms search for vortex signatures in the field of mean Doppler velocities by identifying strong localized azimuthal shears. However, such shears become difficult to detect when a tornado is located at far ranges or when the size of the tornado is small compared to the radar beam. This work is motivated by the hypothesis that tornado spectral signatures will deteriorate at a slower rate with range than azimuthal shear signatures in radial velocity data. Therefore, tornado detection can be improved by identifying vortex signatures in Doppler spectra.

A Doppler spectrum reveals the reflectivity and radar weighting distribution of velocities within the radar resolution volume, while a Doppler velocity is a weighted average of such a distribution. Therefore, a characteristic spectrum associated with tornadoes may facilitate their identification. In this work, tornado signatures in Doppler spectrum are investigated. Using a Rankine vortex model, it is shown that tornadic vortices have a distinct signature in the Doppler spectrum that deviates from ordinary weather signatures. The spectral signature depends on the interplay between the radius of maximum wind, the size of radar resolution volume, and the relative location between the center of vortex and radar volume. Doppler spectra from May 3, 1999 tornado are presented and discussed.