BMX Severe Thunderstorms

Overall, I used OCTANE, PHS, ProbSevere 3 and LtgCast today. NUCAPS wasn’t really accessible. Worked the DSS event, an Air Show, which was canceled due to severe thunderstorms all afternoon producing tornadoes, large hail and damaging winds. DSS for this event would have been done days ago.

Below is a shot of LtgCast on a radar background and ELN measured lightning, the +/- are positive and negative ground strokes, and the cyan dots are in-cloud. It is interesting how the 75% probs lead out into southwestern Georgia though the showers there are more stratified and lightning isn’t expected, yet it gave about 45 minute notice of lightning strikes; that’s a good thing. But how useful is this? It predicted a few single lightning strikes tens of miles apart scattered across 100 miles which isn’t really useful; would you stop all outdoor activities across ¼ of Georgia for a few stray strikes? Would you clear the baseball field because a lightning strike will hit in the next hour somewhere within 50 miles? Not likely, but knowing there is some chance is valuable information for an event coordinator for risk analysis. If they can make minor changes to activities with little or no impacts, it helps, especially if it’s an area where lightning isn’t expected. What would be a big plus would be an estimate of flash density/frequency expected to go with the probs. That gets back to tracking the convective cells to predict areas of dense lightning. We have radar and ELN’s for that.

PHS composite reflectivity vs radar at 21Z… I find little value in the PHS composite reflectivity product. Below you see PHS composite reflectivity compared to the radar returns at 21Z. It’s not doing too well and I haven’t seen a time when it has done well predicting where the storms will be. The HRRR, NSSL WARF, HRef, NAM Nest and other high res models do much better.

PHS Bulk Shear 0-1 km below on the left and 0-3 km below on the right both show a line between areas of lower and higher shear along the boundary where the severe storms were tracking, but this occurred after the convection started. I don’t see a pre convection signal pointing to where the training storms formed.

The Bulk Shear 0-6 km below shows more promise with the 19Z  frame showing a boundary where the training severe storms formed/tracked (what did it look like at 16Z or 17Z?). I would need to see more of this pre convection to really make a judgment, and would need to see positive validation/verification to have any confidence in it as a tool.

– Super Bolt

PHS Reflectivity Forecast Helping Out with DSS

The Panda Ceremony was held at the Jackson, MS Zoo on the evening of Wednesday, Jun 14, 2023.  Strong to severe storms were forecast throughout the afternoon and evening hours and the event coordinator requested DSS for lightning and any severe weather with as much lead time as possible.  Using the 14.16z initialization of the PHS reflectivity was useful in providing some timing details to the event coordinator (See Figure 1). It suggested a fairly robust storm to roll through the Jackson area by around 00z with the forecast reflectivity ranging from 50-65 dBz directly over the event site.  

The reflectivity forecast did well with depicting a fairly large storm to move across southern Mississippi but was a bit too north on the location.  Overall, the product was very useful in boosting the forecaster’s confidence in the convection timing to impact the event. See Figure 4 for a look a the verification.

Figure 1: Loop of the 14.16z PHS layer reflectivity had a large storm over Jackson, MS, and impact the event (black range ring) by 00z.

Figure 2: A DSS Graphic was issued shortly after 3 p.m. to highlight the timing and potential impacts of the Panda Ceremony.  The original image had the animated GIF above with the PHS reflectivity forecast.

Figure 3: This was the first of 3 graphics created for the event and this graphic highlighted the severe weather threat and timing.  Overlaid is the MRMS reflectivity greater than 35 dBz and OCTANE Winds.

Figure 4:  A loop of MRMS with ProbSevere and LightningCast shows some verification of the PHS reflectivity forecast shown above. This loop ends at 2232z.   

– Podium

Day 3 Review of Products & Operational Applications

I took on the warning forecaster role during the operational period today. A supercell was underway at the start of the period, so I began by issuing a Severe Thunderstorm Warning. ProbSevere was (unsurprisingly) indicating a slam dunk for severe wind and large hail (Figure 1) but I noticed when sampling the ProbSevere data that MESH was slow to pick up on just how large the hail was with this storm. A broadcast meteorologist passed along a report of 5-inch hail at 2000z, but MESH did not indicate 4+ inch hail until 2020z.

Figure 1

I also noticed that PHS composite reflectivity was much too slow with the progression of convection across the SHV CWA. Below are a screenshot of PHS progged composite reflectivity at 21z and a screenshot of the radar 0.3 degree reflectivity at 21z. PHS expected the supercell of interest to still be located over the northeast portion of SHV (Figure 2) while in reality the supercell was about to exit the easternmost fringes of the CWA (Figure 3).

Figure 2

Figure 3

Once the aforementioned supercell exited the SHV CWA, we pivoted to the LZK CWA to monitor the ongoing convection there. I decided to interrogate the OCTANE products because I hadn’t looked at much satellite imagery thus far.

OCTANE was especially useful in two ways today:

  • OCTANE Speed indicated a storm split underway as two speed maxima can be seen in the red-orange hues (Figure 4, top left panel)
  • I tweaked the AWIPS colormap for OCTANE Direction to better depict the storm-top divergence occurring (Figure 4, top right panel)

Figure 4

– Vort Max

Verification of NUCAPS Soundings

Using the 1930z pass of the NUCAPS soundings during an SPC Moderate Risk was quite exciting.  The environment was primed for significant severe weather and having NUCAPS soundings available at the time of convection out of ahead of storms was beneficial. It provided additional confidence in the severity of the environment. 

As you can see from the satellite, soundings were unusable further north due to the ongoing convection and cloud cover. Thankfully clear skies prevailed from Jackson and southeastward providing some impressive sounding data to compare to SPC Mesoanalysis.  Sounding A was well in the warm sector and sunny skies and thus had the highest CAPE values and lapse rates. Sounding B was just to the southwest of the DSS event and was beginning to see some anvil over that location, but it still provided reasonable data and compared well with SPC mesoanalysis.    

Figure 1: Location of the NUCAP Soundings chosen to view and compare to SPC mesoanalysis. For reference, the range rings outline the 5 and 10-mile radius around a DSS event in Jackson, MS. 

Figure 2: Sounding A with a surface-based CAPE of 6004!

Figure 3: Sounding B with a modest 4079 surface-based CAPE.

Figure 4: SPC mesoanalysis Surface-Based CAPE and CIN at 20z on June 14, 2023.  The star is the DSS event location and the two dots are the estimated location of the soundings analyzed.

Figure 5: SPC Mesoanalysis 0-3km lapse rate (C/km) depicted lapse rates from 7.0-8.0. Once again, both soundings matched up well with SPS data.

– Podium

BMX Warning Met (West Sector) on June 14, 2023

Our group role played as BMX during an anomalously conducive environment for severe weather consisting of very high instability and very high shear. My position was warning met, sectorized for the western half of the CWA.

ProbSevere helped considerably in triaging which storms deserved attention, and which storms were trending in such a way that warranted a warning decision and/or adjustment. A combination of several supercells that would merge with other cells and evolve into clusters and bowing quasi-linear structures made it difficult to have a comfortable handle on storm behavior and associated hazards. This was compounded by the fact that radar data from the favored radar site, KBMX, was dropping out at times. While other datasets were referenced (satellite, surface observations, storm reports, objective analysis, etc.), ProbSevere felt like it gave the biggest helping hand in warning decisions during this complex scenario.

Looking in the southwest portion of the CWA around 20:30 – 22:00 UTC, ProbSevere signaled the diminishing probabilities associated with a bowing segment that earlier resulted in an 82 mph wind gust within eastern MS as convection was decaying rapidly. This helped grow confidence in holding off on a warning as it entered BMX’s CWA.

ProbSevere also pointed at a cluster of storms increasing in potential for severe hazards. This was helpful given the noisy velocity data, higher than desired sample height (closest radar sampled ~5-8kft), and likely complications regarding wind vectors having some orthogonal component to the radar radial. Essentially, ProbSevere provided clarity and continuity in a noisy radar situation increasing confidence in storm behavior.

Having discussion with a developer, I expressed how useful it was to not only have a time series to view trends in overall probabilities of severe hazards, as well as of each hazard (i.e. hail, wind, tornado), but also time series of the data that drives the probabilities like those offered by CIMSS.

The above time series corresponds to the cluster of storms in southwest BMX around 20:30 – 22:00 UTC (link). Notice in the time series the local minima in probabilities within ProbSevere v3 around 21:00 UTC (~47%)  followed by a notable increase in probabilities, exceeding 80% around 20 minutes later. This trend in local minima followed by increasing probabilities was more obvious than ProbSevere v2. Looking at the individual inputs, this was likely driven by an increase in things like Azimuthal Shear, VIL, MESH, Lightning attributes, along with other environmental attributes.


6/14/23 HWT – SHV

Operational window encompassed ongoing severe storms at the start of the period which moved eastward out of the CWA by mid afternoon. The resultant outflow boundary / cold front intersection to the west of the CWA become the focus regions for potential for renewed storm development.

NUCAPS 6/14/23 19z sounding south of the outflow boundary of interest compared to nearby surface observation showed a large discrepancy in observed surface conditions (90/72 at nearby ASOS). This raises considerable questions on how these soundings are to be best utilized in operations.

OCTANE was useful in monitoring the attempted updrafts along the outflow boundary. It seems to show updraft growth more clearly than the day cloud phase product. I would like to see these data incorporated into the LightningCast model as this product is also very good at highlighting potential areas for CI.

OCTANE output was also viewed  in a region of a splitting supercell across western AR. Good conversation with developers on further development of the wind retrievals with observational data constructed hodographs being an operational request.

Utilized LightningCast to highlight the region of concern for new storm development. The storms did develop but it was a more gradual evolution. The highlighted region did verify as the region of development.

ProbSevere did a great  job as a safety net for radar warning operations. The element trend window is a helpful addition.  Talked with the developer on potential marking times along the time series where the element changes ID or grows in area to show when ProbSevere has merged or separated elements as this impacts resultant probabilities.

PHS model output from the 16z run was viewed online and compared to the corresponding HRRR. At the time of the exercise conclusion its solution had verified better on composite reflectivity than did the HRRR. A product in AWIPS highlighted regions where the fusion data is different than a zero hour model output would be beneficial for situational awareness and potentially for model output utilization.

– jbm

Probsevere and Octane

Widespread severe storms have been ongoing over portions of the southeast CONUS nearly all afternoon. ProbTor was doing a decent job with picking up on a couplet in Montgomery county Alabama around 1950Z. Overall it seemed that ProbTor V3 was giving higher tornado probabilities than V2. It was noted that V3 is based off of the HRRR whereas V2 is based off of the RAP. So that was rather interesting to me that the HRRR was potentially picking up on better areas of rotation.

Image of ProbSevere with ProbTor in the bottom left

Loop of tight couplet that ProbSevere V3 was picking up on

The other really interesting feature I noticed on the part of Octane. The Octane vertical divergence fields which is noted by a sharp gradient of blue in the coloring. This was occurring around the same times as cooling cloud tops that were seen on IR imagery. This also seemed to be a precursor to when storms began to intensify on radar. Along with that intensification was when a better couplet began appearing on radar and an increase in ProbTor V3. Overall I thought that those three correlations were rather neat to see in real time. If this correlation does continue to show up then this has the potential to be another important feature of the Octane product.

Loop of Probsevere with intensifying cluster of storms

Still image of Octane showing the “Blue” Divergence aloft features

Loop of cooling cloud tops via IR

– Tor Nader    

Assessing LightningCast in Ongoing Convection

In the Tallahassee CWA, convection was ongoing at about 21z over the western portions of the Florida Panhandle.  In order to get a quick assessment of the convective trends, loading up LightningCast with ProbSevere and MRMS radar is a good start. A loop is provided in Figure 1 below. 

Figure 1: Loop of MRMS reflectivity overlaid with LightningCast and ProbSevere. 

I noticed in southeast Alabama into southwest Georgia that LightningCast was teetering back and forth between the 50 and 75 percentile contours.  LightningCast even went down to the 25% contour for a brief stint at 2041z.  Looking at NLTN data, there was certainly plenty of lightning left within that storm, so I couldn’t quite understand why it went down but then pulsed back up. The other issue I ran into is the contours suggesting lightning, but convection is nowhere near the 75% contour (See Figure 2). 

Figure 2: 2041z LightningCast with ProbSevere and NTLN lightning. 


Strong to Severe Thunderstorms This Evening

Based on the latest satellite and radar data, a supercell thunderstorm has moved into the very northwestern corner of our cwa (northwest of Boise City). A Severe Thunderstorm Warning has been issued for the Boise City area with the greatest threat being large hail, damaging winds, and maybe an isolated tornado. Echo (cloud) tops are ranging between 40-50 kft. LightningCast data is showing lightning activity increasing over that area as CAPE values have increased to 2,000 J/kg.Based off the OCTANE Motion and Speed Sandwich product, storm top divergence is evident near Boise City with the tight color gradient with winds to the north of Boise City at 234 degrees vs 188 degrees south of Boise City. This activity is being driven by a surface low near the TX/OK state line. As this feature continues to shift to the east-northeast there is the potential for this storm to strengthen and/or for additional showers and thunderstorms to develop across the area.


LtgCast Precursor to Storms Diverging

LtgCast gave advanced indication (about an hour) in the FWD region that the northern storms were diverging to the north away from the original track that was more easterly. Below you can see the LtgCast extending to the north into clear skies in the first two images. The next four pairs of images are the LtgCast paired with the radar at about the same time. The last two images are the next 20 minutes of radar.

In the TAE area, the LtgCast wasn’t as effective. Behind the storms, LtgCast kept >75% probs for lightning in clear air as seen in satellite imagery, possibly as much as 50 miles or more. This was odd in that it usually quick to pull in the probs behind the storms. It wasn’t that the storms were slow moving or inconsistent in their motions; they were moving uniformly to the southeast at a constant rate (didn’t measure the speed, but would estimate the line speed 20 kts or more).

– Super Bolt