Category: News

Prob Tor Prob

Posted in News on by Michael.

At times, there is some unnatural jumpiness in the Prob Tor Product. See the time series of 0.5 base velocity and Prob Severe/Tor below.

The first image is 2310Z. The storm is not overly impressive, with 5 minute FED max of 3 (not shown). Accordingly, the Prob Hail/Wind/Tor is 4/2/11 % respectively. The base velocity shows some decent winds of 45+ knots at about 3500 ft. above ground.

When moving to 2312Z however, the Prob Hail/Wind/Tor  are 4/12/61 % respectively. That is a huge jump in Tor Prob, but the velocity does not indicate any substantive change in organization, and the 5 minute FED has not changed.

Going ahead to 2316Z, the base velocity looks even less impressive, but the Prob Hail/Wind/Tor  are virtually unchanged at 3/12/61 % respectively.

Finally, at 2318Z, the Prob Hail/Wind/Tor  drop to 3/2/10 % respectively. These kinds of ups and downs despite any other evidence pointing to a change in organization made the products difficult to use in this case.

Thorcaster

NMDA slow to identify St Louis supercell

Posted in News on by Michael.

The NMDA initially identified a mesocyclone with the embedded supercell at 2253Z, however the storm was exhibiting supercell characteristics for at least 15 to 20 minutes before this time, including a very well defined bounded weak echo region capped off by a 70+ dBz echo above 20 kft. Below is a loop showing the line approaching St Louis with NMDA overlaid.

 

Below are 4 panel radar images from klsx at approximately 2235z (nearly 20 minutes before the NMDA first identified a mesocyclone), showing the 8.0 (top) 4.0 (middle) and 0.5 (bottom) elevation cuts. A very well defined bounded weak echo region is evident in the reflectivity images in the top left.

-64BoggsLites

MRMS AzShear and SAILS scans

Posted in News on by Michael.

The MRMS AzShear product produces multiple areas of rotation as it tries to ingest 0.5 degree SAILS scans. The animated gif above is from WFO TOP late Tuesday afternoon as a tornadic circulation moved north of I-70. The AzShear algorithm easily identifies the area of rotation, however during SAILS cuts it breaks the rotation into multiple areas rather than moving the entire circulation forward with each time step. This may be due to only the lowest scan being available for AzShear during SAILS cuts, but it makes the product look disjointed and unrealistic. While this may be easy to ignore in discrete supercell mode, the AzShear output could be confusing if SAILS were deployed in situations that may have numerous areas of weaker rotation and/or shear zones, such as a QLCS/MCS.

Dave Grohl

Total Optical Energy vs. Flash Extent Density…the steel cage match.

Posted in News on by Michael.

On the afternoon of May 21, portions of the Wichita (ICT) and Topeka (TOP) CWAs were under a Severe Thunderstorm Watch. To give one of the new GLM products a run through, I wanted to compare two cells of differing strength to see if GLM trends were helpful in nowcasting the cells. See the loop below.

The first cell of note was in Dickinson County (TOP CWA), and the second  was in Rice County (ICT CWA).  At the initial time of 2131Z, the storm in Dickinson County was already severe, and the Prob Severe Product supported this assertion with 68% prob wind and 76% prob tor. Meanwhile, the cell in Rice County was much weaker, with 50DbZ topping out below the -20C level (not shown), and prob hail/tor values of 2/1 % respectively.

At that time, the 5 minute flash extent density showed a little enhancement (max of 32), but the total optical energy was actually less (63 fJ) than several other, weaker storms in the area. Meanwhile, the storm in Rice county showed very modest flash density values (max of 9), but more significant TOE (144 fJ).

As the loop continues through the next 40 minutes, the intensity of the Rice County storm steadily increases as it moves northeastward, reaching Prob Hail values of 34% by the end of the loop (prob tor was still very low).  meanwhile, the storm originally in Dickinson County remained fairly steady state, with Prob Wind/Tor both at 68%.  It’s obviously conjecture at this point, but maybe the TOE product on the Rice County storm was signaling some type of intensification. hard to tell from just this one case, but interesting to note nonetheless.

Thorcaster

 

NUCAPS Data from Yesterday’s (May 20th) Event…

Posted in News, NUCAPS on by Michael.

So, at the map briefing this morning there was a LOT of discussion about yesterday’s event, and the fact that despite the tornadoes that did occur, this was a “bust” of sorts.   That is, there wasn’t near the widespread development of supercells that was anticipated.  There may be multiple reasons for this, but I thought I’d take an initial look at some data and soundings again from yesterday to see if anything stands out that I happened to miss.

Here’s a sounding (radiosonde) taken from Norman (KOUN) yesterday at 18 UTC 20 May 2019.

KOUN Sounding (radiosonde) from 18 UTC 20 May 2019

Notice that lapse rates were not particularly steep through the sounding, with 850-500 lapse rates around 6 C/km.  Also, notice the deep moist layer from the surface up through ~700 mb, which I’m told that this is rather unusual for this area.

Ok, let’s take a look at a NUCAPS sounding in this proximity taken from the ~20 UTC swath.

NUCAPS Sounding from ~20 UTC 20 May 2019 (“green” sounding south of Norman, 34.98N 98.26W)

Immediately, I notice that the NUCAPS sounding doesn’t have the deep moist layer that the KOUN sounding showed just about 2 hours earlier.  If anything, this layer did not disappear during the 2 hours, it remained, and perhaps even got a bit deeper.  The 00 UTC KOUN sounding (not shown) indicated a moist depth up through about 680 mb or so.  Ok fine…there’s the limitations with the accuracy of observations in the boundary layer in NUCAPS, but this is still important to capture since many severe weather parameters are derived from the surface or near-surface layers.  However, it was noticed that lapse rates aloft were fairly similar, showing lapse rates that were not very steep, about 6.6 C/km in the 850-500 mb layer.  The modified sounding (shown next) also didn’t capture the moist boundary layer.

Modified NUCAPS Sounding from S of Norman, 34.98N, 98.26W

I did want to point out again the good characterization of the mid-layers.  Granted, while the NUCAPS soundings didn’t capture every squiggle in the dew point temperature aloft, it did capture the overall character of the moisture profile.

Well, I ended up having to give up my workstation, but perhaps this case should be investigated some more.  Although not shown here, the lapse rates and theta-e from my memory did appear to show a relatively large W-E gradient across the area from north Texas into S-Central Ok.

Kris

Comparing GLMIR to regional radar

Posted in News on by Michael.

The GLMIR imagery is doing a nice job this afternoon pinpointing the most active portion of the line of convection moving into eastern MO, highlighting this area in cyan. It also picked up on a region of lightning which developed within the stratiform region to the north (shaded in purple).

-64BoggsLites

NMDA vs MDA observations

Posted in News on by Michael.

Looking at the line moving into the LSX CWA and comparing the legacy MDA to the NMDA, a couple things are evident. The legacy MDA (not shown, couldn’t get the image to save properly) only identifies a couple mesocyclines further south along the line, but identifies them both as being in the lowest scan. The NMDA on the other hand identifies numerous circulations along the line, however none of these are identified as being at the lowest elevation slice.  Granted the algorithms typically struggle with linear features, but the MDA seems to be filtering out a lot of the noise and keying on the two stronger areas of rotation while the NMDA seems to be overly sensitive to shear along the line.

 

update…. as the line gets closer to the RDA the NMDA has now detected several areas of rotation which are marked with cross hairs. Again the algorithm does seem to be too sensitive with identification of mesocyclones.

 

-64BoggsLites

Azimuthal Shear Noise Behind Convective Line/ProbSevere Analysis

Posted in News on by Michael.

Dealiasing problems in northwest Arkansas and southwest Missouri carried over to the operational version of the azimuthal shear rotation tracks.  The azimuthal shear in the lower left panel uses a new dealiasing suppression algorithm, which effectively eliminated the noise. Fortunately, the noise with the current operational version of azimuthal shear was in a stratiform area of precipitation well behind the convective line, making it easy to ignore this data.  Sampled winds in the region of the erroneous rotation tracks were very high — up to 80+ kts.

By focusing my attention on the azimuthal shear associated with the leading edge of the QLCS, and analyzing Probsevere data and trends, I was able to anticipate a strengthening of the line.  The Probwind component was only 32% at 1934Z, then steadily increased, reaching 89% at 2012Z.  At 2030Z there was a LSR of trees and power lines down in Mountain Home,  Arkansas. Probtor increased to 65% at 2014Z, and as this storm shifted north into southern Missouri, there was a report of a blown out store front in Hartville, MO.  It has not been confirmed whether or not there was a tornado. However, the increasing Probtor values increased my confidence for severe straight line winds that are often associated with strong mesovorticies.  Depending on the event, Probtor trends and values could be used as a confidence builder to increase my warning wind speeds from 60 to 70 mph or 70 to 80 mph. Roy