A Reanalysis of the Saturday September 8 Severe Weather Event

In some ways, the Saturday, September 8 Severe Weather Event exceeded expectations regarding the 70mph EF0 tornado in Breezy Point, NY and the 110mph EF1 (one MPH away from being an EF2) tornado in Canarsie, NY. In other ways, though, the severe weather event was underwhelming since the squall line weakened well before reaching the Metro region.

SPC Filtered Severe Storm Reports from Saturday, September 8th

Here are the SPC filtered severe weather reports from the event. There is a decent amount of coverage. However, notice the relative minimums in Pennsylvania, New Jersey, Long Island, and most of central and eastern New England. Additionally, notice that there were no high wind reports (65 knots or greater) and zero hail reports.

RAP 500mb Analysis from 15z, or 11am Saturday, September 8th

However, let’s start with the two tornadoes in Brooklyn. They occurred around 11am, which is 15z.



RAP analysis at 500mb showed a robust shortwave in the Great Lakes in association with a potent longwave trough. The shortwave was at our latitude at this point, which is important because this helped the synoptic scale forcing for lift stay at our latitude, at the time. Of course, the trough is still well west of the region, so the best forcing was closer to the base of the trough. But considering how potent the trough is, synoptic scale forcing can run out well ahead of the trough base, and combine that with surface-based instability, and lift is generated.



SPC Mesoanalysis of Surface-Based CAPE, Valid 15z, or 11am Saturday, September 8th.

In early September, the ocean waters are in the mid 70s, which is quite warm. The strong trough and storm system to our west led to strong synoptic wind flow from the ocean to the coastal Metro regions. Considering how warm and moist that airmass is, it helps to generate a lot of surface-based instability and CAPE, with values near 2000 J/KG. Also notice the lack of CIN, which certainly helps for convective initiation. Further west, there was much less surface instability, which is why there were no discrete cells forming in those regions.

SPC Mesoanalysis of 0-1km Storm-Relative Helicity, Valid 15z, or 11am Saturday, September 8th.

Not only did we have instability, but we also had low-level shear and helicity, helping to rotate the updrafts. We had surface winds out of the SE with veering towards the SSW and a great increase in magnitude at 850mb– caused by the increased low-level jet from the closed 850mb low to our northwest. This led to helicity values around 150 meters squared per second squared, and in combination with the low-level moisture and instability, low-topped supercells were able to form. The reason why the supercells remained low-topped was because the mid-level lapse rates and mixed layer CAPE was quite low, so updrafts in the mid-levels were weak and were not able to sustain themselves. This became a problem for the squall line later in the day.

TJFK TDWR Velocity Scan of the Canarsie Tornado, Valid for Around 11am Saturday, September 8th.

Thus, as you can see here, a very strong couplet was detected by the TJFK TDWR radar site crossing the Canarsie, NY area. This indicates an area of very tight, strong low-level rotation, supporting the EF1 tornado that the Canarsie area had.


Further connecting the shear and surface-based CAPE, the rising air parcels that generated the storms were able to be surface-based to begin with; greatly helped by the lack of CIN, synoptic-scale ascent, and relatively steep low-level lapse rates. This led to surface-based thunderstorms, instead of ones that were based higher in the atmosphere. Thus, these thunderstorms were fully able to tap the strong surface-based CAPE, instability, and shear that were in the low-levels to begin with, and were able to become tornadic. Since the storms were surface-based, the weak mixed layer CAPE and mid-level lapse rates essentially became irrelevant as a hindrance to tornadic development.


00z 500mb NAM Initialization from 00z Sunday, September 9th, or 8pm Saturday, September 8th.

00z NAM 500mb 24 Hour Forecast from September 8th, Valid 00z Sunday September 9th, or 8pm Saturday, September 8th (The two images are valid for the same time, for comparison purposes).




Now, why did the squall line weaken so much? Here is the NAM analysis at 500mb, valid 00z, Sunday or 8pm, Saturday. Notice how much further north the shortwave is! It is now well into Canada, as opposed to pressing well down into Michigan and Upstate NY. This means that the forcing for lift also shifted well to the north, meaning that there was less lift in Pennsylvania, New Jersey, and the immediate Metro region in the late afternoon and evening hours than what was originally thought. (The 500mb image on the bottom is the 24-hour NAM forecast made from the prior night’s run, meaning they are valid for the same time. The 24-hour forecast was much further south than what actually resulted). Additionally, since it lifted northward faster than originally thought, that means the squall line outran the best forcing for lift faster than originally thought, thus the squall line also weakened before reaching most of New England, as well.

RAP 300mb Analysis From 22z or 6pm Saturday, September 8th.








The 22z RAP analysis shows a strong upper-level jet-streak, but well off to the north and west of the Metro region, placing the area well south of the right entrance region. The right entrance region of a jet-streak is where divergence occurs, which favors ascent beneath it. Thus, the Metro region was south of the best ascent/lift.

18z KOKX (Upton, NY) Observed Sounding.





So, why was this so important? Of course, more lift means stronger, sustained updrafts, helping to grow and maintain thunderstorm strength. But we were relying on this synoptic-scale ascent to counter the extremely poor mixed layer CAPE and mid-level lapse rates. The observed 18z sounding at KOKX shows the problems quite well. In the morning, we had surface-based parcels, so the poor mixed layer CAPE and mid-level lapse rates were not an issue. However, with a squall line, the lift is generated from the gust front, which is much more mixed layer based, as opposed to surface based. This means that the high surface-based CAPE and instability essentially became irrelevant. Since the gust front was trying to create lift in an atmosphere that had very low mixed layer CAPE and poor mid-level lapse rates, the updrafts were not able to be robust, and the strong shear tore the updrafts apart. Had we had the synoptic scale ascent from the shortwave, that would have been able to force the air-parcels upward, creating lift and updrafts strong enough to compensate for the poor mid-level lapse rates and mixed layer CAPE, and create a taller cloud. It also would have helped to create lift from the surface, thus air parcels would have actually been able to use the high surface-based CAPE and instability. This also explains the lack of lighting and hail – the cloud tops were not tall enough to form ice crystals, which are necessary for the charge differentials for lightning, and necessary for hail formation. Also, there were no strong updrafts to even “push” any ice crystals further up into a cloud to begin with…an overall terrible atmosphere for hail formation.

18z KIAD (Dulles International Airport) Observed Sounding.

There was a secondary maximum of storms in the Washington, DC area. Why was this? One could argue that they were even further south, and intuitively, further away from the lift from a latitudinal perspective. However, they were also further west. Thus, they were a bit closer to the lift from a longitudinal perspective than the Metro area was, so from a lifting perspective, I’m not quite sure they had any less lift than we did. Also, they had better low-level instability, as the 18z KIAD (Dulles International Airport, which is near DC) observed sounding shows. This helped to create surface-based updrafts just strong enough to rise through the weak mixed layer CAPE and mid-level lapse rates. Because they had surface-based updrafts, which were able to tap into the low-level directional shear, they actually had a long-tracked tornado warning (though a tornado did not actually touch down), despite the fact that they were initially well south of the tornado threat area.

Considering that Upton issued 8 tornado warnings and there were a relatively high amount of low-topped supercells, it was a pretty good convective day. However, it certainly had the potential to be a better convective day. It just goes to show you how hard it is to truly get a serial-derecho type of event. It was still a pretty potent squall line – most squall lines are not going to produce widespread 50 knot winds in every location they hit, and this one certainly produced more wind reports than most other squall lines.




3 replies
  1. Mat
    Mat says:

    Nice writeup. However, I’m a bit confused by this statement: “…with a squall line, the lift is generated from the gust front, which is much more mid and upper level based, as opposed to surface based”. Gust fronts are low-level phenomena.
    I would think that the gust front would be acting to lift the higher-CAPE low-level parcels (or layers, more accurately), not mid/upper level air as suggested here. Am I misunderstanding?

  2. Doug Simonian
    Doug Simonian says:

    Thanks for the compliment and question!

    Yes, gust fronts/cold pools are at the surface. But they extend well above that. The ML (mixed layer) CAPE (which was terribly low) uses the lowest 100mb of the atmosphere, which is about 1km.

    The gust front/cold pool might not have been exactly 1km deep, but it’s much closer to that than just a few meters above the surface. Thus, gust front lift is much more ML CAPE driven than surface-based CAPE driven. When you combine the awful ML CAPE with the fact that the mid-level lapse rates were very weak, the gust front was not able to provide sufficient enough lift to maintain the strength of the squall line.

    I should probably edit my statement to mixed layer driven, which thus means that the SB CAPE is unimportant and mid and upper level based instability and lifting potential becomes more important, as opposed to just saying that gust front is directly mid and upper level based.

  3. Doug Simonian
    Doug Simonian says:

    Also, I typed “mid-level” CAPE instead of “mixed layer” CAPE on the post several times since I was so used to thinking about “mid-level” from the weak mid-level lapse rates. Sorry if that caused any confusion, and the post has now been edited accordingly.

Comments are closed.