Category: GREMLIN

Developing and Splitting Cells in Sterling County TX

Posted in GLM, GREMLIN, SPG2024 on by Kevin Thiel.

This lengthy post will cover a number of interesting observations with developing and splitting cells near Sterling County TX. This post will mainly focus on GREMLIN, but also a few other products.

Initially, we were focused on new updraft development Glasscock County TX. GOES-E GREMLIN (left panel) depicted Z values of about 57dBZ. KMAF comp reflectivity (right panel) was also picking up some 50+ dBZ. The 0.5 slice on KMAF (middle panel) was yet to really show anything of interest. This is evidence that GREMLIN is picking up on situations where echoes aloft have not yet started reaching near the ground. This has a degree of predictive value.

 

The loop from here shows that this storm then proceeds to split as it moves into Sterling County TX. The storm split is apparent on the KMAF lowest slice and comp reflectivity, as early as 2020Z. On GOES-E GREMLIN, the split does not start to show up until 2040Z, and is not readily apparent and clear until 2100Z.

 

So, let’s look a little closer into why GOES-E GREMLIN may have struggled with picking up the split. The loop below compares GOES-W satellite imagery (left panel) with GOES-E satellite imagery (right panel). While GOES-E data masks the updraft of the left-mover under the anvil from the right-mover, GOES-W has a better view of the left-mover updraft.

 

 

 

 

Comparing GOES-W GREMLIN and GOES-E GREMLIN, it’s clear that GOES-W had the better view of the left-mover updraft, and picked up on the split much more accurately (though it was low on dBZ values). On the contrary, GOES-W GREMLIN did a much less consistent job in handling the right-mover.

–Insolation

From a GLM perspective, it’s clear that GOES-West GLM sensor had a better view of the left split updraft and lightning activity than GOES-East. The flash extent density product shown below detected an increase in activity much quicker on GOES-West than its east counterpart. This may be due to the less inhibited view GOES-West had of the updraft looking at it from an angle compared to a more top-down view GOES-East had, which may have been blocked by water droplets or hail near the top of the thunderstorm.
-Joaq

Comparing GREMLIN Across Two Storms

Posted in GREMLIN, SPG2024 on by Kevin Thiel.

 

The GREMLIN radar product is shown above in the top left while MRMS composite reflectivity is on the top right. The two storms in question have severe warnings on them in this loop. The thunderstorm in the southwest is more isolated from other convective cloud debris, while also displaying a more intense thermal couplet and cloud top divergence signal than the thunderstorm to the northeast. It seems likely that this is the reason why the northeast storm has a lower ceiling for GREMLIN reflectivity than the southwest storm, even though the MRMS composite includes a 60 dBZ core in the northeast storm.

 

-Joaq

Using GREMLIN for Time of Arrival

Posted in DSS, GREMLIN, LightningCast, SPG2024 on by Kevin Thiel.

LightningCast has been running high near the PGA Tournament today near Fort Worth as a result of several thunderstorms around the region, but not over the region. As a result, decided to swap over towards trying to use GREMLIN to find the time of arrival of the boundary noted earlier.

So an attempt was made to highlight this in the DSS image to provide an estimate of when these hazards would arrive. Perhaps another proofread would’ve caught the initial typo from copying it over. However, a value of about 40 percent was noted. Partners still had information that storms would become increasingly likely over a narrower time frame in addition to the probabilistic information from LightningCast.

But nature had other plans with a strike and a few pulses of lightning over the event at the time this was “sent out”. An area of weak convection was in the area, and so I would’ve thought that this would not produce lightning.
Sneaking in information on the RGB evaluation, you can see how this was not a young storm and likely something in decay, and that may track with a bit of the surprise element.

Kadic

Warning vs DSS – “WFO DMX”

Posted in DSS, GLM, GREMLIN, LightningCast, OCTANE Speed Sandwich, SPG2024 on by Kevin Thiel.

It was interesting when comparing the Warning side of the house today versus the DSS side. When we (Cumulus and Kadic) were discussing this and picked two of the tools we used the most, there were similarities and differences:

For Warning Ops: OCTANE and LightningCast

OCTANE proved very useful in interrogating convection. LightningCast was also a helpful diagnostic tool in highlighting the potential for and track of intense convection when used with ProbSevere.

OCTANE:

Here’s a look at 2 particular instances from OCTANE:

A combination of cloud top cooling in OCTANE and subsequent divergence aloft was a helpful clue in assessing the potential of a storm that was distant from the radar. It was caught a little later in analysis, but OCTANE proved helpful in diagnosing the storm and deciding to pull the trigger.

 

 

This next instance was a warning that was issued solely using OCTANE and seeing how well it lined up with radar. The warning targeted the center of where the maximum storm top divergence was taking place, and then stretched down towards the south to account for parallax. The warning decision was made for the impressive cloud top cooling and pronounced divergence that appeared in the scans leading up to the warning. The panel on the top left shows the OCTANE speed, and it transitioned to a blue color leading up to the event.

 

 

LightningCast and Radar:

As the line shifted east into the area, ProbSevere stood out, while the LightningCast steadily increased. The left hand panel depicts GREMLIN, and it properly highlights the southernmost storm as being the most intense. Unfortunately, not every storm that we issued warnings for got a specific screenshot, but when looking at LightningCast, areas that were likely to experience 10 or more flashes with a 70% probability seemed to correspond well with ProbSevere values would support issuing warnings.

Below is the example of what MRMS looked like the moment DMXSVR005 was issued solely based on OCTANE. Much of the SVR encompassed the highest LightningCast values with a probability of 10 flashes of 70% in yellow and the various ProbSevere contours. Again, this highlights how useful these tools can be in performing storm interrogation. However, when thunderstorms are numerous, this may be a lot to run through. They are definitely useful tools in the tool belt, though.

For DSS: LightingCast (especially the Dashboard) and GREMLIN/GLM.

– LightingCast: I REALLY like the form and Dashboard. It helps focus on the DSS site specifically and organizes the data really well to where I would feel comfortable explaining/showing an EM the graph of   lightning probabilities. Honestly, I could bring this back to my home WFO right now and use this for DSS events this summer. A couple things that could be added to make it even more awesome: adding more options for ranges (right now there is only 10 miles, perhaps adding 15 and/or 20 miles). Folks could then choose which to display in the graph. The other thing (fairly minor), perhaps reversing the size of the bubbles for the GLM data (smaller range, smaller bubble). But, this is personal preference – maybe if this could be customized by the user like the colors?

– GREMLIN/GLM: GREMLIN followed the storms a lot better today (seems to do better with more intense storms versus run of the mill/sub-severe ones). I used a two panel display with GREMLIN on the left and MRMS on the right with GLM and LightningCast and compared the two. I used time of arrival for the storms to 10 miles outside the DSS event and also at the site itself. GREMLIN was able to keep up with MRMS really well! I am becoming more and more convinced that this could be a really great product to help if a radar goes down or there is a radar hole (in data).

Overall, it seems as though OCTANE was used more for warning ops versus DSS, but LightningCast was used by both the warning operator and DSS forecaster.

Forecasters Cumulus and Kadic

Using GREMLIN and LightningCast for Warning Ops and DSS

Posted in DSS, GREMLIN, LightningCast, SPG2024 on by Kevin Thiel.

GREMLIN continues to perform well with regards to the overall picture of precipitation and convection. In fact, in the example below, GREMLIN seems to be catching on to the northern extent of the line breaking apart and the southern portion becoming more intense. However, it seems to be a bit slower than the MRMS data.

LightningCast could be more useful in this case if the thresholds were modified (25 flashes versus 10) to better identify and focus on the more intense convection.

LightningCast overlayed on GREMLIN emulated radar (left) and MRMS Composite Reflectivity (right)

Looking at the LightningCast dashboard for our DSS event, probabilities of lightning are increasing. I found that this is actually an easier way of being able to communicate lightning probabilities for a site (or a range around the site) versus using the map (seen above). This would allow me to let an EM or site official know that probability of lightning is near 80% for 10 miles form the site (within the next hour) and 50% at the site itself.

 

LightningCast Dashboard for Belin Quartet Summer Concert Series (Des Moines, IA)

We provided valuable support. A message was sent nearly 45 minutes ahead of time following the issuance of a Severe Thunderstorm Warning to the west.

If our partners are simply looking to delay, LightningCast may also prove helpful in giving the all clear if you don’t see anything upstream on radar. The values react well as convection clears the site. This is also great in helping you know for sure when the last lightning flash took place, and it can help us give better information.

 

Forecasters Cumulus and Kadic

Collection of Day Two Thoughts

Posted in GLM DQP, GREMLIN, OCTANE Speed Sandwich, PHSnMWnABI, SPG2024 on by Kevin Thiel.

Day 2 has featured more convection, and has been a helpful day testing these products and how they help in warning operations. Although I might not feel confident making warning decisions solely based on any of these tools, I think that each tool provides a valuable piece of information.

Pre-Storm

To keep things short here with all the observations, PHS was very helpful today in showing how the QLCS situation would evolved with several areas of embedded rotation. Having CAPE with SRH together showed how these came together, and in conjunction with velocity highlighted rotation updrafts within PHS. This proved to be a helpful pre-storm evaluation. A few storms began rotating, and then everything began rotating as the PHS model indicated.

 

Observations Related To Warning

The developing squall had a linear appearance at first. As time progressed with more embedded areas of rotation, this became a lot less neatly organized.

Here is a look pre-warning for a tornado warned cell with ProbTor increasing up to almost 40 before moving off the point.

 

 

A zoom in on an impressive overshooting top. Sorry for the reverse loop.

 

Here is a V-notch like structure. Though it doesn’t correspond with a radar V-notch, it does indicate how strong an updraft this was.

 

 

And here’s the radar look of that, which appears to somewhat match the configuration seen aloft.
Interesting Signals

 

One thing to note early was that the PHS forecast had a lot of convective debris lingering in Iowa that was not present in reality. This does not appear to have impacted the instability parameters very much.

We’d mentioned looking at the dewpoints for the tendency for aggressive convection. But it only seemed slightly high compared to reality.

We did have a blob near Sioux City on Gremlin that didn’t really correspond with any signal on radar, and it didn’t seem to have satellite signal to go with it. Not sure where it came from, but we were able to see it was erroneous.

 

Here’s a look at GREMLIN with waves and wobbles following the GLM lightning.

 

 

Here’s another fun look at where it seemed the convection on the northern flank may have affected GLM quality with values decreasing on the north side. Note the reversed image loops.

Here’s an instance where GREMLIN’s max intensity happened before a lightning jump. Unfortunately this is reversed, but GREMLIN struggled to resolve an intensifying storm in the middle of the line.
Here is an example of GREMLIN losing a cell in 3 surrounding cells.

Kadic

End of Day 1 Thoughts

Posted in GLM DQP, GREMLIN, LightningCast, OCTANE Speed Sandwich, PHSnMWnABI, SPG2024 on by Kevin Thiel.

Thoughts at the end of Day 1…

The LightningCast product I think would be VERY useful for DSS. Overall, when seeing it perform in real-time, the increasing LC probabilities seem to eventually correlate well with GLM flash density. I look forward to using the DSS form this week and seeing how that works for specific sites.

The GREMLIN product seems to be a great way to see the overall picture of precipitation (say, for a region). I think it struggles with precipitation intensity a bit (>45 dbZ) both for storm cells and for heavy stratiform precipitation. At the “storm” level, I have seen instances of the model not following the evolution well (either too intense or not enough).

For OCTANE, it was easier to pick out an example of CI and divergence with the IR versus the Visible products. I could use the direction product on its own in operations, but I really like having the speed, direction, and cloud top divergence all together in a 3 panel to identify convection.

PHS did a great job today identifying convective initiation when overlayed on visible satellite imagery. I look forward to seeing how this performs in other areas of the country this week.

Still learning how best to utilize the GLM DQP; but, when looking over Cuba, I was able to better understand how it locates areas where the data might not be the best. I hope to learn more about this product through the week and see more examples of its application.

Forecaster Cumulus

GREMLIN versus Radar

Posted in GREMLIN, SPG2024 on by Kevin Thiel.

Overall, GREMLIN performs fairly well with the zoomed out shape of the precipitation returns compared to radar, but perhaps a bit larger extent. Picks up on a few heavier cells moreso versus heavier stratiform rain. Since the overall radar mosaic of GREMLIN looks so close to current radar, it would increase my confidence in using it for the big picture (especially if the radar data were to go out for some reason).

GREMLIN vs radar reflectivity 2010Z through 2058Z 20 May 2024
Focusing on the two cells in the west… it’s interesting because it looks like GREMLIN resolves both cells fairly well initially but follows the evolution of the southern cell for a time a bit better than the northern cell. In this instance, I couldn’t say much about confidence (sort of cancels out). But, it would be important to make sure to take into account the environment and also maybe use multiple tools in addition to GREMLIN to help with confidence.
Zoomed in GREMLIN vs radar reflectivity 2010Z through 2058Z 20 May 2024

Forecaster Cumulus

Waiting For Convection To Go

Posted in GREMLIN, LightningCast, OCTANE Speed Sandwich, PHSnMWnABI, SPG2024 on by Kevin Thiel.

While looking for thunderstorm initiation, my eyes are turned towards anything that informs me whether convection will develop further or decay. There have been several things to note while waiting for instability to move into the region.

The main forecast challenge is the favorable ingredients to produce severe convection has to be advected into the region. The WRF with PHS data demonstrates that initial convection intensifies later in the day as better instability arrives across the region on the bottom right panel. This also corresponds with better dynamics noted across the others entering eastern Colorado. Model reflectivity on the bottom right increases as a result of better forcing over time. Until then, it’s monitoring at what point things actually start turning the corner and using the new tools available to find that point.

 

And so far, the things found have been what it’s not. So here are some things being observed in this period of waiting. At the beginning of the day, one of the things I noticed was related to the OCTANE detecting warming and cooling. As clouds moved off snowy foothills, it was apparent on the viewer where water clouds appeared warmer than the snow surface, and caused a pocket of cooling to appear on the eastern foothills once satellite could see the frozen snow, and warming whenever a cloud layer shifted overhead obscuring the snow. In the middle of convection, this would probably be irrelevant, but just a thing to note while we wait.

 

 

I like the idea of mashing together several products that we’re testing at once. So, I’ve applied the LightningCast, WRF with PHS CAPE, and GLM. The idea will be to monitor how the storm is pulsing compared to with what information is provided from PHS. It’ll also help track in what area LightningCast is lighting up and whether it is heading towards a favorable or unfavorable environment. With LightningCast aiming for detection within an hour, it began highlight a cell that corresponded with favorable instability. The combination of these two helped me hone in on this cell as being more likely to produce lightning than a similar LightningCast to the northwest. The 10% contour formed just before 20z, and steadily increased leading up to the first flashes on GLM roughly 30-40 minutes later. Nicely done!
While waiting, a small cell caught my eye. The area was almost completely clear, and showed very dark on visible imagery, to being cloud covered near Palmer. This made it appear this was about to blow up, but then you can see the OCTANE tool quickly reverse course once it becomes clear it will not develop and it begins to come down on visible.
Off to our west, there were a couple cells. Analyzing the tool on GREMLIN, the southern cell was less intense on GREMLIN compared to MRMS, and reversed for the storm to the north. However, neither are particularly intense, but it does indicate to keep a watchful eye and use other products like GLM to assess intensity.
As we move past 22Z and how the WRF with PHS data, it has done an excellent job forming the convection near the Denver Airport, but by Shamrock/Leader/Adena, that cell has not formed. This is creating a region of spurious data due to convective feedback. Some of the model appears to drive convection by the cold pool from this storm meeting instability advecting in from the east. It then focuses on this cell over the others, but this appears unlikely to verify at this point given how it is performing so far. By 00z, the 0-3km SRH bullseye creeps above 1600 m2/s2 moving towards Fremont. I won’t put the image of the new cycle that just came in, but the bullseye got more dramatic over 2000 m2/s2. Not sure if how much those magnitudes are in the realm of possibility.
One of the interesting behaviors lately has been a few storms forming in the cold pool as we approach 23Z. There has been convection developing on the western side of decaying cells. This has me thinking about how this would look for backbuilding precipitation. Would it have this look of the cool purples remain anchored in place while reds for new convection continuously appear to upstream that rides atop the areas of divergence? The signals may not appear robust, since there may not be fast storm motions.
Looking back at LightningCast, I have noted the known limitation of the forecast trying to bridge separate pieces of convection. The gap seems quite large though, and I wonder if there may be other means to QC the LightningCast with existing radar without making it slow to process. Or we can trust that the human eye is capable of noting that radar will confirm the lack of reflectivity at -10 C or higher.

That’s all I have today!

Kadic

Using GREMLIN/OCTANE alone to Assess Mesoscale Conditions and Monitor Storm Intensity

Posted in GREMLIN, OCTANE Speed Sandwich, SPG2024 on by Kevin Thiel.

Quick MesoAnalysis Comparison

Using the OCTANE speed sandwich we were able to quickly assess the SPC mesoanalysis of upper level winds entering our CWA. The SPC mesoanalysis indicated upper level winds of around 100 kt spreading through the Big Bend part of TX ahead of an upper level trough. This analysis was quickly confirmed by the OCTANE speed sandwich data. High cirrus was streaming into the region ahead of ongoing convection and was sampled at 105 kt in the OCTANE data. This real time analysis supports the objective analysis by SPC and yields higher confidence in the data.
Image 1: The OCTANE speed sandwich showing velocities >100 kt across the Big Bend region of TX spreading into the southern parts of the SJT CWA.
 
Image 2: The SPC mesoanalysis of 300 mb winds indicating a 100 kt upper jet spreading into the Big Bend region of TX.
 
 
GREMLIN/OCTANE to Monitor Convection without Radar
 
We were tasked with monitoring convection across the SJT CWA with an ongoing DSS event in San Angelo. While convection remained to our west and east, there were a few thunderstorms that developed across the northern part of our CWA. We did not have radar data, but were able to use OCTANE speed/direction products to assess storm strengthening trends though the afternoon. In combination with the GREMLIN radar emulation, we were confident in our assessment of storm strength even without supporting radar/MRMS data.

Image 3: The two storms of concern are to the north of San Angelo and are noticeable in the speed/direction imagery. There were a couple notable periods of cloud top cooling and increased divergence, but the uniformity in the mid/upper level wind speeds and direction increases confidence that these storms are either maintaining intensity or weakening.

 

Image 4: The GREMLIN data showing two distinct cells across the northern part of the CWA. These storms maintained intensity for about 30 minutes without any noticeable change in the OCTANE data. As the OCTANE wind speeds decreased, the GREMLIN data showed that the storms diminished overall.

 
These two products (OCTANE speed/direction and GREMLIN) used together can increase confidence in warning/decision making, particularly in the absence of radar/MRMS data. While there are some latency concerns with GREMLIN, further testing of these data should be continued.