HSLC Environmental Climatology Update

Over the last several months, the HSLC environmental climatology has been gaining steam. With the null data now in hand, I have begun to compare the environments of HSLC events and nulls. However, prior to the acquisition of the null data, I focused on two primary investigations: the role of boundary-relative winds and shear in discriminating between tornado events and straight-line wind events and the breakdown of environments associated with differing convective modes. This blog will serve as an update on my progress so far.


Boundary-Relative Winds and Shear
As discussed in a previous conference call, all work with boundary-relative winds and shear was completed using high impact wind and high impact tornado events. All high impact events were those with at least 80 severe reports of any kind, while high impact wind events (HIWND) had at least 90% wind reports and high impact tornado events (HITOR) had at least 20 tornado reports. This subset of data represented 9 total events (5 HIWND and 4 HITOR) and 1511 total reports (1019 HIWND and 492 HITOR).

In general, deep layer shear magnitudes between the two event types were similar. However, mean shear orientations varied by as much as 30 degrees, with HITOR having distinctly more zonal shear vectors than HIWND, especially in deeper layers (0-3 km, 0-6 km, and 0-8 km). Using the subjectively-determined average orientation of synoptic boundaries (i.e., cold fronts) over the course of the events, I was able to determine the mean along boundary component (ABC) and mean cross boundary component (CBC) for shear vectors over numerous layers. In each of the layers (0-1 km, 0-3 km, 0-6 km, 0-8 km, cloud-bearing layer, and effective layer), the ABC was larger in HIWND, while the CBC was larger in the HITOR.

The orientation of shear along boundaries and associated linear forcing has been shown in previous non-HSLC studies to be a primary discriminating factor in storm morphology and subsequent hazards. These results imply that the same thinking can be used in HSLC events.


Convective Modes
Using the convective modes database described in a previous HSLC presentation, I created mean soundings for the following environments: supercells embedded in clusters (CLUS; 61 total reports), discrete supercells (DISC; 57), supercells embedded in lines (LINE; 57), and linear non-supercells (NONS; 126). These were created through archived RUC analysis at a 25 mb interval from 1000 mb to 100 mb. The only reports included in the convective modes database are those which were associated with tornadoes, significant (65+ kt) wind, or significant (2″+) hail.

Figure 1. This slide shows a comparison between mean soundings for the discrete supercell and non-supercell environments. The bottom dashed lines are the LFCs, while the top dashed lines are the ELs for the respective environments.

Overall, there were subtle differences in the thermodynamic and kinematic profiles of the four convective modes. Figure 1 shows a comparison between DISC and NONS mean environments. Note the differences in the LFCs and ELs for the surface-based parcels in each environment and the overall compressed unstable layer (UDEPTH) for non-supercells. Also, note the meridional enhancement of the winds aloft in the non-supercell hodograph. These two features were the main differences across all modes. Figure 2 shows a table reflecting the thermodynamic and kinematic characteristics of the four modes. “Straight aloft” indicates a hodograph similar to that found in the non-supercell case, whereas “curved throughout” indicates a hodograph similar to that in the discrete case.

Figure 2. Contingency table for the four convective modes showing the differences in thermodynamic and kinematic profiles. UDEPTH refers to the depth between the LFC and EL.

Currently, skill scores are being calculated for parameters across the convective modes to determine which paramaters are the best discriminators between the modes. Ultimately, the differences are likely more substantial than can be identified by a simple visual comparison between averaged soundings.


Hits vs. Nulls

The list of nulls was compiled through guidance from the FFC nulls spreadsheet and through an automated verification process. For our purposes, a null was considered a severe thunderstorm or tornado warning issued by a WFO during a convective day (1200 UTC-1200 UTC) in which no severe reports were gathered for that respective CWA. Following quality control, 114 null points remained over 62 unique null days.

Preliminary comparisons between the events and nulls focused on all tornado events against all nulls. This was explored as a first comparison because these two environments likely represented the largest discrepancies in parameter values. The parameters with the highest skill scores in discriminating between tornadoes and nulls included composite parameters (significant tornado parameter and vorticity generation parameter), effective shear magnitude, and the low-level lapse rate. Figure 3 shows box-and-whisker plots for distributions of the significant tornado parameter over various data subsets. Note the very low distribution for nulls but generally low distributions across the board compared to climatological norms.

Figure 3. Box-and-whisker plots for the significant tornado parameter across various subsets of HSLC data. The red horizontal line indicates the median value; the blue boxes encompass the 25th to 75th percentile of data; the black lines encompass the 10th to 90th percentile of data, and the red crosses represent outliers.

Composite parameters, in general, seem to perform well in discriminating between high profile HSLC events and nulls; however, the thresholds for HSLC events appear to be lower than previously determined due to composite parameters’ reliance on CAPE. Thus, it may be worthwhile to increase the number of contours on the SPC mesoanalysis to account for lower values (e.g., perhaps contour every tenth from 0 to 0.5 on the significant tornado parameter). As with the convective mode comparisons, I am currently calculating skill scores in order to determine what parameters—and what values of those parameters—are most useful in discriminating between events and nulls.


Google Earth Report Maps

Figure 4. Example of a Google Earth reports map. Description is in the text.

Figure 4 shows an example of a side project I have been working on. These maps, viewed in Google Earth, show all of the reports for a given event as placemarks. The placemarks are different colors based on the type of report (red for tornadoes, blue for wind, and green for hail; significant reports include a “dot” in the placemark), and when one clicks on the placemark, further information about the report—including date and time, number of associated injuries and fatalities, and associated convective mode, when available—is displayed. The reports can also be animated by time in order to show the sequence of reports over a given event.

If you would like the Google Earth KML for any of the HSLC events in our database, e-mail me the date(s) at kdsherbu@ncsu.edu.

Advertisements
This entry was posted in High Shear Low Cape Severe Wx. Bookmark the permalink.

2 Responses to HSLC Environmental Climatology Update

  1. nwscwamsley says:

    Great insight !!

  2. Matt Parker says:

    Non-comment so I receive emails notifications 🙂

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s