Assessing the various GLM lightning products, I found the MFA and FED particularly useful in correlating the more active thunderstorm core / updraft areas. TOE was useful, but maybe a slight change in color scale may help to better identify those very active convective areas. (perhaps the sharper color gradient change could start at a slightly higher value than currently)
The GLM flash point data didn’t show as much ‘clustering’ as I had expected to see, as compared to other surface-based data sources (ENTLN). Is this related to a data display density or a more sensitive surface-based lightning detection? However, due to the parallax correction, the flash point data did line up with fairly well with active convection, although there were a few flash point detections well displaced from any radar reflectivities (see bottom left)–perhaps related to stratiform lightning?
Comparing the Flash Extent Density to base reflectivity in northern Florida, we noticed an area of exceptionally high flash density just north of the best convection. Adding flash points (parallax corrected) to the radar image and they were more where one would expect in the storm. This indicates a parallax issue with the Flash Extent Density. This is a good example of where the flash points can be a good sanity check when interrogating a storm.
Using GLM for an IDSS event allows for great flexibility and confidence in providing warnings for decision makers. In the image below, this shows a complex organization of thunderstorms. In this case, a lightning alert or warning for them. In this case, since the thunderstorm structure is much more complicated, it is unlikely an all clear would be provided for some time. It is likely this would be persistent through the afternoon as the surface front is pushing the sea breeze front back toward the coast and keeping the activity nearly parallel with the coast. In this example, the FED product does provide the forecaster a good idea of the forward extent of the which is to the south and east, despite the fact the upper level winds are pushing the avils to the north. The lightning points are also helpful to get an idea of where the potential return strokes are actually reaching the ground enveing though the GLM “can’t” actually 100% determine this but just based on the probability.
Point flashes also provide the user with an idea of the size and potentially the ability to identify which core the lightning originated from. Not sure how well this will work or if it will work but just a thought. Think more work and or research will be required before we can say one way or the other if this will actually work.
A moderate risk of severe storms capable of producing large hail, damaging wind gusts and tornadoes occurred across the Dakotas. My focus was in the Bismarck, ND CWA where storms were likely initiating in eastern MT and then moving into the very unstable environment across western ND. All of the higher resolution models were a bit late on storm initiation as storms began to fire between 3-4 PM CT. The experiment began around 2 PM CT, which allowed for mesoanalysis of the pre-convective environment.
A NUCAPS CONUS NOAA-20 pass occurred at 19z across ND and then again at 20z where the eastern edge of this pass overlapped with the previous pass across western ND. At 19z, a comparison was made between the NUCAPS profile and a nearby RAP sounding at the same time. Below Image 1 shows the locations of the NUCAPS profile versus the RAP sounding. This area was chosen as it was close to where the satellite was showing some potential for convective initiation and was just east of the dryline in the area where the better instability was to be present.
Using sharppy the two profiles were then compared simultaneously. Both Images 2 and 3 show the two profiles, but image 2 will be highlighting the NUCAPS profile and associated instability values and image 3 will highlight the RAP sounding with associated instability parameters. Looking at the two profiles, there is not much difference in the mid to upper levels between the NUCAPS and RAP. However, the NUCAPS profile struggles more with the boundary layer features and temperature/dewpoint. Looking at observations, the current temperatures near that sounding location at 19z was 86 deg F with a dewpoint of 70 deg F. The RAP seemed to initialize these surface values pretty well and the thermodynamic profile east of the dryline, along with a bit of a capping inversion in place. Meanwhile, the NUCAPS profile struggled with the temperature and dewpoint, thus under doing the moisture and instability parameters. The CAPE values are noticeably different with the NUCAPS profile much lower with the instability due to these surface differences.
After seeing the discrepancies with the observed surface values versus the NUCAPS profile, I decided to grab the modified NUCAPS profile for the same location for comparison. Image 4 shows this modified sounding with a 10 degree difference between the non-modified surface temperature. The modified sounding shows a 82 deg F surface temperature, while the original NUCAPS profile had 91 deg F. With the cooler surface temperature the modified sounding showed a similar inversion to the RAP sounding between 750-800mb. The dewpoint temperature also was better representative of the actual surface dewpoint, which helped increase the instability parameters significantly. NUCAPS profiles tend to be a tad lower on the CAPE values, so the fact that the RAP is still about 1000 J/kg higher is not a surprise. However, with no RAOB sounding available and comparing the RAP with the modified NUCAPS profile there is quite a bit of similarity between the two in terms of the thermodynamic profile. Lastly, as storms begin to fire in the next hour or so and no RAOB profiles closeby, it might be useful to compare and utilize the temperature heights (0, -10, -20, and -30 deg C) for radar interrogation as storms initiate. Knowing the RAP and modified NUCAPS profiles were similar then the heights from the temperature levels could also be compared. The RAP does show higher heights than the modified NUCAPS profile, so this is something to keep in mind and monitor as storms fire along the dryline.
Keeping with the theme of NUCAPS, there was another pass at 20z further west (as mentioned at the beginning) that overlapped the 19z pass in parts of western ND. This included the town of Bismarck, where the office put out a special 20z RAOB sounding. Bismarck was a bit further east than the previous sounding, but was still in the very favorable environment. Images 5 and 6 show the comparison between the NUCAPS sounding at 20z and the RAOB Bismarck special sounding at the same time. Similar results can be seen between the observed sounding and NUCAPS profile where the CAPE values are again lower in the satellite derived sounding. This time the NUCAPS profile did a much better job with the surface temperature and despite the temperature profile being a bit smoother due to lack of detail in the boundary layer, the profile was overall pretty similar to the RAOB temperature profile. The dewpoint profile on the NUCAPS was much drier at the surface and therefore had a bit of a drier boundary layer than the observed sounding, which is likely why the CAPE values are also a bit lower.
Once again the modified NUCAPS profile was compared (Image 7 below). The modified profile did a better job at showing the moisture in the boundary layer and attempted to pick up the dry layer at 650mb, which was actually at 700mb on the RAOB profile. Unfortunately, the temperature was too low and therefore the modified NUCAPS temperature profile shows a very sharp capping inversion that was unrealistic. Overall, the CAPE values did increase with the modified sounding versus the original NUCAPS profile and were closer to the observed sounding. Twice it has been noted that the heights of the temperature levels were closer between the non-modified NUCAPS profiles with the model/observed soundings. There may be some calculation in the modified sounding that is causing the heights to be lower and maybe unrealistic. In scenarios where there is a RAOB sounding, that is the best picture of the atmosphere you can get but it is great to compare the NUCAPS profiles for comparison to future events and potential trends in the satellite derived soundings.
As storms began to initiate across eastern MT, both G16 and G17 GLM were utilized to look for lightning instances in the growing storms. Having both satellites can be super helpful, especially when one viewing angle may not see the strike, while the other does. This happened several times during storm initiation where one satellite would pick up a strike, while the other displayed nothing. Images 8-9 show this occurring twice in two different storms where each satellite picked up a strike that the other did not. As mentioned before, the viewing angle may not be in a good position for the satellite to see the storm’s top and therefore the strike is not bright enough to be detected. Along those lines, the scattering properties in the cloud are also not visible by the angle of the satellite’s view point and could cause the satellite to miss a strike. Lastly, there is a quality assurance that occurs for each product and if the strike wasn’t strong or long enough then the pixel could have been tossed out during this quality assurance. This is why it is so important to utilize both satellites when possible and it is a best practice to err on the side of whichever satellite is showing more lightning is probably more accurate.
ProbSevere version 2 and 3 were compared through the afternoon. The trend continued with version 2 remaining about 20-30% higher in all categories except the tornado probs. Version 3 has leaned towards being slightly higher than version 2 when it comes to tornado probabilities. ProbSevere time series was utilized to track the southernmost storm along the line of storms headed into western ND during the mid afternoon hours. Both radars were pretty far away on either side of the storms, with Glasgow’s radar being slightly closer. The lowest elevation scan was at around 13000-14000 feet when velocity began showing a strong mesocyclone. Image 10 shows the time series of ProbSevere and the readout comparing version 3 with version 2. All four ProbSevere categories were steadily increasing through the last hour with version 2 remaining higher than version 3. Version 2 shows close to 100% probabilities for all but tornado, making this storm look like a slam dunk due to the environmental parameters. Meanwhile version 3 is slightly lower due to the fact that it can pick up on similar storms that occurred in a similar environment with little to no reports (from storm data). This is where version 3 adds in a bit more information to create more realistic probabilities.
Based on the strong rotation in Image 10, the tornado probabilities were close to 30 percent which is relatively high and should give a forecaster confidence on issuance with a lack of lower level radar scans. Chaser footage also helped to back the need for a tornado warning with images of wall clouds, funnels and more being reported from multiple sources. Image 11 shows the time series for ProbSevere along with multiple other parameters. One thing that was interesting to see was the tornado probability drastically dropped in version 2 but remained steady in version 3. Since version 2 is heavily using az shear, you can see the drop in MRMS az shear (red line on second plot down on the far left), which could be correlated with that probability drop in version 2. Also, the MLCIN is slowly increasing (blue line on second plot down on the far right) and could be playing a bit of a role in this drop as well. This is where version 3 might have a leg up on version 2 when it comes to tornado probabilities.
Lastly, the optical winds were utilized to see the winds at the top of the storm. Image 12 shows the optical wind field for 200-100mb. You can see the cooler cloud tops in satellite below the wind field and then the associated diffluence aloft. This is an indication of the very strong supercell that is showing no signs of weakening anytime soon. Also, it is of note that there is another cool cloud top signature a bit further to the northwest associated with another strong supercell with diffluence aloft. The optical wind fields are useful in knowing what is going on aloft and the potential strengthening or even weakening of a storm.
An approaching short wave trough as well as a remnant MCV moving through the area helped trigger thunderstorms over the upper Midwest. Convection developed over Minnesota during the afternoon and evening hours. Storms did not become severe near the Grand Forks, ND CWA until late in the afternoon.
The upper Midwest is near the extent of coverage for both GOES 17 West and GOES 16 East. This made it a good chance to compare how the GLM lightning products were affected by this issue. Research has shown flash densities for both satellites diminish in this area, well away from the nadir of both satellites.
The character and quality of the Flash Extent Density (FED) returns from each of the satellites can be seen, with GOES 17 showing a slightly westward tilt, and an eastward tilt in the returned grids for GOES 16. The strongest storm in north central Minnesota has a higher and possibly better return on 17 than on 16.
Values of Minimum Flash Areas from the satellites were quite different in some cases, and were also skewed as a result of the distance from the nadir of each satellite. Placement of the flash areas also differed from “viewing” angle.
Total Optical Imagery (TOE) was also skewed. It is interesting to note that the pixels of TOE in far southeastern ND change from minute to minute, as seen in Images 3 and 4. At 1958Z, GOES 17 had an area of TOE returns along the ND/SD/MN borders, while GOES 16 had 2 separate areas, one in southern ND and one in west central MN. By 1959Z, both GOES 17 and 16 agreed that there were 2 separate TOE returns in this region.
Scattered strong to severe thunderstorms were popping up along the eastern ND and MN border today. However, these storms were struggling to become severe at times with a majority of the cells pulsing up and down. A NUCAPS pass around 19z provided gridded satellite observations and forecasts for the area to compare with SPC’s mesoanalysis page. The gridded NUCAPS at 19z shows CAPE (Image 1) values ranging from 500-1500 J/kg in the area mentioned above with little to no CIN (Image 2) present based on the gridded NUCAPS data. Image 3 below shows the SPC 19z CAPE/CIN data along with the forecast over the next 6 hours (through 01z). When comparing the 19z SPC mesoanalysis to the gridded NUCAPS, there was not much difference between the two with both showing higher CAPE values further south into Nebraska and Kansas. The gridded NUCAPS for CIN seems a bit erroneous with really no signature for -50 or less of CIN, which is present in the SPC mesoanalysis. This is likely due to the lack of detailed boundary layer features with NUCAPS and the fact that it may likely wipe such smaller inversions.
Looking into the forecasted parameters from NUCAPS, there is a much higher bias in the CAPE values. However, they did a great job at pinpointing an area of higher instability to watch for storms to potentially become more severe with time. The overall CIN forecast looked as if it may start to increase further west near Grand Forks later in the evening, but in central MN where the corridor of CAPE values were higher remained uncapped. As time progressed through the afternoon a few storms did start to intensify and become severe across north central MN with a few severe wind reports. A few lingering surface boundaries were present, along with a weak shortwave at 500mb helped to enhance the storms a bit. I do feel the NUCAPS forecast values for CAPE were a bit too high in comparison to the actual environment and should definitely be compared to model data.
Lastly, the location of storms yesterday provided the chance to compare the GOES-16 and GOES-17 GLM products with one another. However, image 5 shows the extent of the two satellites and GOES-17 was right on the edge of where storms were across the Upper Midwest. As you get further away from the satellite and towards the edge of its coverage, you can start to notice more of a tilt in the gridded data. This may cause some erroneous data as seen in comparison with GOES-16. Comparing the GOES-16 data (Image 6) with the GOES-17 data (Image 7), there is a better display of the minimum flash area and lightning sizes with the GOES-16. You can see GOES-16 shows more of a mixture of shorter and longer flashes (purple and yellow colors), while GOES-17 sees strong shorter flashes (yellow colors). Also, the further away the satellite is to the storms the more likely the flash extent density may be less accurate. This is likely due to the storms being on the edge of the satellite’s reach. Therefore it is important to check out both satellites when possible, but take into account where the storms are in respect to the satellites coverage.
Here are a few more supplemental images of the GLM GOES-16 satellite versus the GOES-17 with similar concerns as mentioned above.
Today we focused on the slight risk across the southeast, specifically WFO Jackson, MS. During the afternoon hours, a small linear complex was coming across northern LA towards Jackson’s CWA. Right before the CWA line, there was a wind report of snapped tree limbs of 3” diameter from Monroe Airport. There was also a measured gust from the airport of 41 mph. The velocity on radar had ~60 knot outbound winds at around 14,000 – 15,000 feet, which easily could have produced a few severe gusts to the surface. The gifs below show the linear line of storms and the associated velocity as the system moved over Monroe Airport in northeast LA with the wind report at 1952z and then continued to enter western MS.
Image 1 shows a loop of radar reflectivity with prob severe overlaid and Image 2 shows the velocity associated with the radar loop.
In this situation, prob severe was not doing as good of a job on picking up on these “stronger” winds. Image 3 below shows the time of the wind damage report and 41 mph gust at the airport in northeast LA, but prob severe and prob wind are both only picking up about 20% probability of this potential. Almost two hours later, the line of storms are a bit weaker on reflectivity but just as strong or even stronger on velocity. Note, the storms were also closer to the radar at Image 4, so the stronger outbound velocities were closer to the surface. So this led to wondering is prob severe a good indicator for straight line winds?
Prob severe utilizes azimuthal shear which as seen in the Images 3 and 4 below are not present with solely outbound velocities and little to no inbound present. This is common for straight line wind scenarios, but not super helpful in terms of how prob wind is calculated. Also, the prob severe is an object oriented product that utilizes reflectivity for these objects. In this scenario, the reflectivity definitely began to weaken but velocity did not. The toughest part was the prob severe began to decrease over the two hour time span shown above, but yet several damaging wind reports of roofs blown off and trees/power lines down led me to believe the probability of prob wind should have remained constant or increased over time.
While investigating the prob severe I also took a look into the lightning characteristics within the line as you can see in the GIF below (Image 5) that there is the formation of some trailing stratiform on reflectivity. A still image was taken (Image 6) to show how the lightning began to extend westward into the light stratiform. The flash area (top right of the four panel) shows the darker purple color extending westward, which indicates the storm mode is more of that light stratiform rain with longer flashes extending through it rather than the intense small flashes within the leading line. This can be helpful in time when you may have a DSS event and the main line has passed through, but lightning is still present in the trailing light rain. Pairing the ground networks with the GLM extent and area allows a forecaster to give DSS on the latest CG stroke within the large area.
Lastly, there was a NUCAPS CONUS NOAA-20 satellite pass at around 19z, which was well before the line of storms made it to the western Jackson CWA line. No special radiosonde launches were made by local offices, so the next best observational guess of the atmospheric profile would be from satellite. Model soundings were also available to compare at the time. A RAP sounding at 19z was taken just east of the western MS border (see Image 7 below for location of this sounding) and a very nearby NUCAPS sounding was also retrieved for comparison (see Image 8 below for location of this sounding).
The soundings (Image 9 and 10 below) looked fairly similar between the model and satellite profiles; however, there were several major differences that played a key role in changing the instability parameters. The NUCAPS sounding was still slightly too low of a surface temperature with 86 deg F versus the RAP’s 89 deg F. Surface observations from 19z at that location showed a temperature of around 91 deg F. Also, the surface dewpoint was far too low on the NUCAPS profile at the surface as it was 5 degrees below the current observation at 19z. Meanwhile, the RAP was only one degree lower than the current surface dewpoint. These subtle differences caused significant variations in the CAPE values.
After realizing the NUCAPS profile was not accurately depicting the surface temperature/dewpoint, I decided to see what the modified sounding might look like through NSHARP. Image 11 below shows the modified NUCAPS sounding through NSHARP with a much cooler surface temperature of near 80 deg F. This was almost 10 deg below the actual surface temperatures and 6 deg below the original NUCAPS profile. The boundary layer was not representative due to this drastic difference and therefore the modified sounding had to be thrown out of the comparison.
Lastly, with knowing the line of storms were headed into the area of interest I decided to see how the forecast products were looking. Unfortunately, I did not get to save the images off in time as the forecast images disappear from AWIPS when the next pass occurs. So I was left with the web-browser version which is only in a gridded format. Unfortunately it is very difficult to depict changes in this format, whereas in AWIPS you can interpolate the image and smooth the results for a more concise display of values. Image 13 shows the comparison of the web-browser gridded format versus the AWIPS smoothed version for the West Coast pass of the NOAA-20 satellite.
Our team, as WFO/JAN, chose the setup for the Mississippi Pickle Fest at 1150 Lakeland Drive Jackson, MS as our IDSS location today (Tue, 08 Jun). Per SPC Outlooks, the Jackson area was on the “edge” of the Marginal Risk Area for severe weather. As operations began for today, a thundershower was noted to the SW of Jackson, moving NE toward the IDSS location of interest:
A modified NUCAPS sounding from near Jackson, MS (which became available later), indicated plenty of instability/CAPE (2000-3500 J kg-1), suggesting that the thundershower would be maintained as it advected toward the Pickle Fest location. This would be a good time for a “heads-up” to the event venue or EM. The unmodified NUCAPS sounding (not shown) still suggested sufficient instability aloft for the storm to maintain itself.
The ProbLightning product on the Web, somewhat surprisingly, still showed only ~25% chance of a GLM lightning flash within the next 60 minutes at 2001 UTC, but this had increased to 75% by 2026 UTC:
By 2029 UTC, the electrical activity was nearly overhead:
Interestingly, the NUCAPS forecast CIN was forecast to increase over the next couple of hours (valid 22UTC, below), after the storm passed, but ahead of another, stronger line further upstream (not shown).
Based on this, and the rapid collapse of electrical activity within the shower around 2110 UTC, a reasonably confident “all-clear” could have been given to the venue at that time…or at least until the upstream line approaches in a couple of hours, assuming it holds together.
Using the minimum flash area to show where the smaller lightning strikes occur but is associated with stronger updraft with cells building faster (Yellow) to generate lightning. Larger lightning strikes occur in the stratiform area of the precipitation field where charge building is slower (Purple). This is also a good way to indicate convective mode as the system translates from individual (SuperCell) to a linear mode.
Using NUCAPS (Modified vs. Unmodified). Why the CAP at mid-levels noted in Arkansas? Is this reasonable or an artifact of the program that isn’t real.
The modified NUCAPS sounding from 1730 UTC revealed in excess of 2,000 J/kg of CAPE (image 2) and by 1900 UTC thunderstorms had developed in an area less shaded by high level cloud cover. Comparing this to RAP mesoanalysis data, it initially seemed too high as RAP mesoanalysis suggested closer to 1000-1500 J/kg (not shown) and based on initial lightning activity. SBCAPE from NUCAPS was extremely high and close to 4,000 J/kg which seems very high (image 3). We did note that for some reason the NUCAPS forecast image showed CAPE being “missing” over central MD/northern VA while the CIN fill was more complete. When sampled over the “missing” data in AWIPS in the Cloud, the readout showed actual values.
Initially, storms appeared to struggle with flash counts on the order of 10 to 20 flashes. Over the course of 30 to 45 minutes, lightning flash counts increased by an order of magnitude (closer to 100-150 flashes)
Closer to 2020-2030 UTC the rigor of convective elements increased and we started seeing transient echo overhang/along with some weak echo regions in tandem with an increase in both lightning activity as well as ProbSevere trends (especially ProbSevere3) as seen in Image 4. This increased our confidence in issuing our own warnings with LWX having a couple of warnings already issued.
Closer to 21:50 UTC things became a bit more interesting from both a radar and lightning standpoint. Imagery from the KDOX radar (Central Delaware) suggested increasing mesocyclogenesis across Baltimore County (image 8) due north of Baltimore.
From a comparison of these images, there was certainly an indication that the updraft was increasing as FED magnitude increased from 51 flashes/5 min to 99 flashes/5 min. MFA also became more concentrated NW of the City of Baltimore and Baltimore County as seen in images 9 and 10. WFO LWX issued a Tornado Warning around 20:51 UTC.
An animation of the TDWR at Baltimore Washington Airport (image 11) had signs of a possible low-level RFD surge (not shown) in the 0.5 degree rapid scan tilt and increasing low level rotation consistent with either a stronger surge of straight line winds or a QLCS mesovortex/tornado. Aloft, (not shown) there did appear to be healthy reflectivity aloft and the concentrated MFA may suggest that a strengthening updraft.
Image 11 from KDOX shows a loose mid-level mesocyclone with a gradual increase in ProbSevere3 with the time series. In this image, the storm was currently tornado warned by WFO LWX. At time we’d likely opt for a tornado possible warning and monitor very, very closely.
By 21:13Z The storm of interest is looking LESS favorable for tornado although prob severe tor increased significantly to closer to 20% perhaps due to an elongated zone of low level shear.
Prob Severe Table Ideas: If the tables could open in a floating tab in Awips that would be very helpful. This way you can manually dock and move the tab around in awips. This way you can quickly view the tab and keep it open until you want to just close it. You can also rename the tab to whatever will help you keep track of the storm that it belongs to. This would get around needing to color code multiple tables etc.
A separate tab will also allow room to show additional information (beyond just prob severe).
Prob Severe jumped to nearly 80% with a fast moving bowing segment through Montgomery County.