Last month at the AMS Severe Local Storms Conference and the CSTAR Virtual Workshop, I presented some preliminary results from a climatology of HSLC mesocyclones and mesovortices. A recording of the SLS presentation can be found here, and the conference paper can be found here. A few highlights from these presentations can be found below.
For this portion of the project, a tracking algorithm was developed and tested in order to study temporal and vertical trends in the strength of HSLC mesocyclones/mesovortices (hereafter more generically referred to as vortices). Azimuthal shear was used to diagnose the strength of these vortices, using radial velocity data from the nearest WSR-88D radar to the vortex. The main emphasis in this portion of the project is to seek discriminating factors in azimuthal shear between tornadic and non-tornadic vortices, by studying a relatively large number of cases. This part of the project also seeks to determine what the potential lead time is for HSLC tornadoes, and what radar velocity information is most important for forecasters to focus on. 83 unique tornadic vortices and 84 non-tornadic vortices (determined using false alarm warnings) have been successfully tracked so far (Figure 1), using HSLC cases that occurred after the WSR-88D radars were upgraded to produce “super-resolution” velocity products in mid-2008.
Since azimuthal shear values can be dependent on range from the radar, due to radar beam spreading and the increase in height of the radar beam with range, the vortices were sorted into three separate bins based on the distance of the tornado/false alarm warning from the closest radar. Figure 2 shows composite timeseries of azimuthal shear at the 0.5 degree elevation scan for tornadic and non-tornadic vortices within 50 km of the nearest radar in a tornado/false alarm warning issuance-relative time-coordinate system. The center of the t axis (the horizontal axis) is the volume scan immediately prior to tornado touchdown for each tornadic vortex, or the volume scan immediately prior to false alarm warning issuance for each non-tornadic vortex. Values of t greater than or less than zero correspond to each volume scan after or before the tornado touchdown/false alarm warning issuance time, respectively. Since different vortices could be tracked for different amounts of time, the number of samples for each t point varies, as not every vortex existed or could be tracked forward and backward to every t point. The number of vortices that could be tracked to volume scans farther backward and forward from the tornado touchdown/false alarm warning issuance time decreases over time.
For the five volume scans leading up to the tornado in Figure 2, an upward trend in median azimuthal shear for the tornadic vortices is found at the base scan, indicating a tendency for vortex strengthening leading up to the tornado as would be expected. A downward trend is found after the tornado. Median azimuthal shear values are above 0.01 s-1 for a long period of time, and at one volume scan prior to tornado touchdown even the 25th percentile is above 0.01 s-1. Overall, these results indicate that meaningful rotation could be detected for most tornadic vortices that were close to a radar before the tornado occurred.
For the non-tornadic vortices, there was a sudden increase in median azimuthal shear values in the volume scan immediately prior to the issuance time of the false alarm warning compared to the previous volume scan, but then the median azimuthal shear values level off and decrease somewhat. A comparison with the tornadic vortices indicates that the azimuthal shear values for the non-tornadic vortices tended to be lower than for the tornadic vortices. The median azimuthal shear value for the non-tornadic vortices is usually well below the 25th percentile for the tornadic vortices, and the median azimuthal shear value for the tornadic vortices is typically close to or above the 75th percentile for the non-tornadic vortices.
Figure 3 shows similar timeseries for tornadic and non-tornadic vortices that were between 50 and 100 km from the nearest radar. A slight upward trend may exist prior to the tornado for the tornadic vortices, but this is less clear. Overall median azimuthal shear values appear relatively flat. Median azimuthal shear values are greater than 0.01 s-1 prior to the tornado for the tornadic vortices, but are weaker than for the tornadic vortices less than 50 km from the nearest radar. There is a slight upward trend in median azimuthal shear leading up to the false alarm warning issuance time for the non-tornadic vortices, when the median briefly reached 0.01 s-1, and then a slight downward trend. Comparing the non-tornadic with the tornadic vortices indicates that median azimuthal shear is greater for the tornadic vortices, but there is a lot of overlap as the median of each respective population is between the 25th and 75th percentiles of the other.
Finally, Figure 4 shows similar timeseries for tornadic and non-tornadic vortices greater than 100 km from the nearest radar. The timeseries for the tornadic vortices shows a downward trend in median azimuthal shear leading up to tornado touchdown. This combined with azimuthal shear values below 0.01 s-1 at the volume scan immediately prior to tornado touchdown suggests that these tornadic vortices are probably not being well-sampled by the radar. A comparison between tornadic and non-tornadic vortices at this range indicates little difference in median azimuthal shear values between the two populations.
Overall, the preliminary results discussed above show that azimuthal shear can discriminate well between HSLC tornadic and non-tornadic vortices when they are close to the radar, but not very well when they are farther away from the radar. Additional details can be found in the conference paper. Ongoing work is being done to refine the methods used in this portion of the project, add some additional cases, and further explore the results. Calculations of vortex lifetime and depth will be done in order to look for further differences between tornadic and non-tornadic vortices. Also, the results will be stratified using the convective modes from the SPC database. Study of radar reflectivity signatures associated with HSLC tornadoes is also planned.