Observations of tornadogenesis from high-resolution reflectivity data using a W-band mobile radar: The Cordell storm of 5 May 2001

On 5 May 2001, a tornadic supercell near Cordell Oklahoma was scanned by a mobile, W-band (3 mm wavelength) Doppler radar. Tornadogenesis occurred within 2 km of the radar, allowing for the collection of a unique, close range, high-resolution dataset of the tornadogenesis process. The high quality of the reflectivity field allowed for some small-scale structures to be observed during the evolution from hook echo to tornado on 10 s time scales, at low elevation angle. Initially, small-scale spin-ups on the order of 200 m were apparent as the hook began to coil up. These small spiral features disappeared in the next scan as the hook continued to coil eventually producing a tornado roughly two minutes after the initial evidence of rotation in the reflectivity field. The tornado produced had a pronounced weak echo feature, otherwise known as a “doughnut hole” within 45 s after tornadogenesis occurred. This feature had an initial diameter of approximately 500 m with an annulus of higher reflectivity spanning about 1.2 km in diameter. Within the annulus, small scale banded features are noted, as well as misovorticies on the southern-most edge (the side closest to the radar) of the higher reflectivity. The far edge of the tornado undergoes some attenuation and therefore there is a period of time where the doughnut hole feature is less pronounced, however it reappears within two minutes. Banded structures are noted periodically throughout the collection time, but are most prominent just prior to tornadogenesis and about five minutes later, and they typically exist within one to two scans rather than being continuous correlating with a time frame of 45 s to a minute and a half.

Unfortunately, accurate radial velocity measurements were not collected, owing to a faulty cable. A re-creation of the horizontal wind field was accomplished from the W-band radar data using TREC (Tracking Radar Echoes by Correlation), which makes use of tracking reflectivity features from one scan to the next and determining velocities from this information. This method was attempted, but failed, primarily owing to the limited azimuthal extent of the data collected over the necessary time period.