NSSL SWAT Case Study - 12 July 1996 Hurricane Bertha-spawned Tornadoes


 
Hurricane Bertha made landfall on the North Caroline coast. In the outer rain bands of the storm, a series of tornadic tropical-cyclone mesocyclones (TC-mesos) developed.  Several F0 and F1 tornadoes struck the coastal areas of Virginia and Maryland.

Tropical mesocyclones present a challenge to algorithms because they are especially shallow.  Circulations are often detectable only on the lowest elevation scan of the radar, and some appear deceivingly weak.  Some of the Hurricane Bertha mesocyclones appeared relatively weak but still managed to produce tornadoes. In addition, tropical mesocyclones can develop very quickly, allowing little room for error. These kinds of mesocyclones are also rare, only occurring with landfalling tropical systems. We simply do not have a large enough data base of these types of vortices to fully understand them.

Presented here are various WSR-88D radar images from the Wakefield, VA radar (KAKQ) and the Sterling, VA radar (KLWX). Included in some of the WSR-88D images is output from NSSL's Mesocyclone Detection Algorithm (MDA).  A yellow circle depicts a mesocyclone.  The red-in-yellow circle depicts a mesocyclone whose base is at the lowest radar scan (where it is a more likely tornado threat).

TROPICAL MESOCYCLONES (KAKQ Loops)

RADAR COMPARISON (Both KLWX and KAKQ radars viewed this storm)
 

Description:  This is a superb example of two radars seeing the same storm in different ways.  Some things to note: 1) KAKQ (right side) is closer to the storm, thus sees lower levels of the storm.  2) KAKQ sees the circulation as NE winds adjacent to SW winds because of the radar's line of sight.  KLWX sees the circulation as SE winds adjacent to NW winds.  Between the two, we could see the entire  structure of the circulation if the radars were able to view the same level of the storm.

The applied research community is increasingly becoming aware that the integration of radar data from more than one radar (WSR-88D and media radars as well!), will be necessary to reduce some of the uncertainties inherent in viewing storms from only one angle and distance. There are a variety of limitations of Doppler radar, and if two or more radars are viewing a storm, you decrease the chance that the multiple radars are each limiting the view of a single point in space. Watch for more examples on these pages in the future!

(KAKQ) IMAGES

Description:  The storm at Smithfield did show good vertical continuity, as illustrated in the first Panel.  This is mostly due to the Smithfield storm having been much closer to the WSR-88D than the rest of the TC-mesos that day, allowing for more scans of the storm's lower levels.

The second Panel follows the mesocyclone from pre-tornado to tornado; the frames are 10 minutes apart.  One can clearly see the mesocyclone spin up tighter at the time of the tornado.

The third Panel shows reflectivity alongside the corresponding velocity data for the time of the tornado.  The tornado is at the comma head where convergence and storm updrafts are taking place.  The comma head is nicely reflected in the velocity data.

The storms are moving to the northwest, so the RFD, which is typically aligned northeast to southwest in a northeastward moving supercell, is instead aligned northwest to southeast. However, the same principles for tornado formation apply. The reference frame is simply rotated 90 degrees clockwise. The supercells assocaited with Hurricane Opal had a similar orientation.

Description:  Similar to the Smithfield tornado, the Yorktown tornado formed at the comma head of a convergence line, as evidenced by the Panel.

The Loop image follows two mesocyclones, the first of which is the Yorktown mesocyclone.  It moves across the center of the screen.  The second mesocyclone appears offshore in the Atlantic and is actually quite strong for a time.  Fortunately, the mesocyclone weakens before moving onshore near Severn, but it still produced minor tornado damage.

Description:  The Panel illustrates frame by frame how quickly the tornado formed and then dissipated.  The first three frames represent elevation scans which are five minutes apart.  The circulation went from weak to very strong to nearly nonexistent in this fifteen minute span.  The fourth frame is the second elevation scan at the time of the tornado.

The Cross Section images, when viewed sequentially, again illustrate the rapid development of this tornadic circulation .  Note that the circulation originated at storm base.

(KLWX) IMAGES
 

Description:  The California tornado also developed very quickly from what had been a less
than impressive low-top mesocyclone.  The first Panel speaks for itself as the circulation transitions from weak to tornadic.  The Six Panel Examination really sums up the difficulties which tropical mesocyclones present to detection algorithms.  Note how the rotation is only clearly evident on Sweep 1, and by Sweep 3 the radar is nearly viewing the storm top (only 18 thousand feet) where rotation is nonexistent in this case.  This makes it impossible for the algorithm to recognize a vertically continuous circulation.  One remedy is to include in the algorithm a "Weak Low-top Mesocyclone" classification. This was coded into the NSSL MDA, and is now being tested on all of our cases. On this case, it increased the Probability of Detection (POD).

The Loop presents an interesting side by side look at reflectivity and storm relative velocity in motion.  The California, MD, storm went on to produce another brief tornado near Hughesville.  The storm was tracking just to the northeast of Bertha's center of circulation which is evidenced by the curved line of storms on the left side of the loop images.
 

Description:  This storm, much like the Smithfield, VA storm, took place relatively near the
radar site, allowing for a better look at the vertical structure.  Notice that the low-level circulation was quite strong, but by the third and fourth elevation scans the radar was already viewing a divergent rotation signature, evidence that the beam was nearing a divergent storm top.
 

Back to NSSL SWAT Case Study Table of Contents Page.