More on the 5/29 Albany Tornadoes

As expected, the National Weather Service has confirmed the 2 tornadoes that radar confirmed on Wednesday, May 29th and one additional tornado in southern Saratoga County. The Schoharie and Saratoga County tornadoes were rated EF-1 while the Montgomery and Schenectady County tornado was rated an EF-2.

The tornadoes occurred in a modestly unstable environment. The 18z 4km NAM by 22 UTC had 1283 j/kg of MLCAPE at KALB but also had an impressively long hodograph that veered with height particularly in the 0-1km layer. A backed boundary layer flow underneath 700mb winds of 50 knots produced impressive shear that aided in storm organization and lead to tornadogenesis.


The 2 tornadoes that produced tornado debris signatures (TDS) on doppler radar give us a useful learning lesson in just how powerful dual polarization technology can be. In the northeast, the presence of heavily forested areas will likely make TDS relatively easy to spot and there may be a larger percentage of tornadoes that produce TDS here than in the plains. I also expect dual polarization to result in an increase in tornado reports since, in many cases, we’ll know exactly where to look for tornado damage even in the case of a tornado moving through a very rural area.

Schoharie County Tornado – EF-1

The tornado that touched down in Schoharie County was on the ground for 2 miles and a total time of 5 minutes according to the National Weather Service storm survey.

The tornado touched down at 6:57 p.m. At that time KENX radar showed a gate-to-gate LLΔV of about 75 knots at about 1500 ft AGL. The TVS was only present for one volume scan at 0.5º base velocity and then the rotation broadened with 75 knots of LLΔV separated by about a mile . By 7:02 p.m. there was a significant debris signature with very low CC, co-located with ZDR near zero, and the remnant circulation.

Within 5 minutes the tornado lofted enough debris to produce a significant TDS.  The TDS was still present, though much less impressive, by the 7:07 volume scan.


The debris was picked up by radar within 5 minutes of touchdown and lingered for 5 minutes after the tornado lifted up. The signal was probably even stronger given the close proximity to the radar site. The radar was sampling the storm at only about 1500 ft AGL on its lowest slice.

The TDS was also visible on the 1.4º slice at approximately 3500 ft AGL though it was weaker. The TDS was not visible on the 2.4º slice or ~5500 ft AGL. If this storm was more than about 40 miles from the radar it is certainly possible that the TDS wouldn’t be picked up at all and if it was a bit farther than that the low level mesocyclone on the QLCS would have been overshot as well.

Montgomery/Schenectady County EF-2

The tornado in Montgomery and Schenectady Counties was the only significant tornado of the day with an EF-2 rating. It was a mile wide for most of its life and was on the ground for many miles ending just past the New York State Thruway in Rotterdam.  While the Public Information Statement from the NWS says the tornado was on the ground for 17 miles based on the coordinates of its beginning and end location I’m coming up with only 12 miles. I’m not sure what the discrepancy is here.

The National Weather Service storm survey puts the touchdown at 6:47 p.m. which seems too late by about 4 minutes per KENX radar. Radar shows the tornadic circulation over the beginning part of the track at 6:43 p.m. (the warning came out at 6:45 p.m.).

Around the time of touchdown the rotation was fairly broad, though quite strong, with LLΔV of nearly 95 knots spread over 7 radials (about 3/4 mile). By the time the circulation was over Mariaville the gate-to-gate LLΔV was 95 knots. Very impressive.


What’s interesting in this case is that the debris signature doesn’t appear until 6:57 p.m. – more than 10 minutes after touchdown. It took twice as long for the TDS to appear with this tornado than in the weaker and shorter-lived Schoharie County tornado. The 0.5º slice was also sampling this storm at a lower elevation than the Schoharie County storm (900 ft AGL versus 1500 ft  AGL). The debris signature was present for 2 volume scans – 6:57 p.m. and 7:02 p.m. before disappearing at 7:07 p.m. Though there is an area of depressed CC and ZDR near the circulation at 7:07 p.m. the values aren’t low enough to classify as a TDS though since the radar isn’t detecting hail in that location chances are there was some small detritus from the tornado being sampled.


After the storm went through Mariaville, NY there was a debris signature as high as 5100 ft AGL on the 3.4º slice.

One possible reason by the debris signature took twice as long to appear on the Schenectady County storm is that in the beginning the tornado was moving through mainly farm land. The Schoharie County storm was in a heavily wooded area and approached a state forest.


Once the storm moved through Mariaville Lake it encountered homes, buildings, and a more wooded landscape. The TDS appears after the tornado moved east of Mariaville Lake.

The ability to detect tornado debris will be a huge help in severe weather coverage. Being able to confirm a tornado around here in real time is so difficult due to hilly terrain and the fact many of our tornadoes are rain wrapped.

Moore Struck Again


The first reports of damage, deaths and injuries are beginning to trickle in from today’s monster tornado that moved through Moore, Oklahoma. The staggering numbers and images are hard to comprehend.

Just like after Katrina, Joplin, and Tuscaloosa we’re reminded of nature’s power, fury, and cruel indifference. Throughout my life the weather has fascinated me. Whether it’s a Tornado,  blizzard,  or nor’easter for as long as I can remember I’ve found myself glued to a TV or window or standing outside in any kind of storm. It’s days like today that we’re reminded that nature’s incredible power can change lives and towns in an instant. Forecasts are better. Warnings are better. Sometimes no matter how good the warning or forecast or preparation or communication some storms are impossible to survive.

Moore, Oklahoma is no stranger to violent tornadoes. The F5 Bridge Creek-Moore tornado on May 3, 1999 was one of the most documented, photographed, and well forecast tornadoes in history. Parts of Moore were simply swept away. On May 20, 2013 history found a way of repeating itself.

The 1999 tornado was tracked live by local TV stations via helicopter much like this storm. The National Weather Service in Norman issued the first ever “Tornado Emergency” for Moore in 1999 just like they did today.

May 3, 1999 tornado emergency
May 3, 1999 tornado emergency from the NWS in Norman, OK


When a tornado become violent (EF-4/EF-5) surviving a direct hit, even if you take proper precautions, becomes difficult.  This is how the radar looked through the storm’s evolution from Newcastle to Southwest Oklahoma City to Moore.






On radar you can see the monstrously large debris ball that gets larger and larger  as more homes are chewed up and lofted along the tornado’s path.

This tornado will undoubtedly join the list of historic U.S. tornadoes – a list that has been growing too quickly in the last few years.

Pennsylvania Tornado Debris Signature

A severe weather event on Friday resulted in a 16 mile long EF-1 tornado in Lancaster County, Pennsylvania around 8:00 p.m. The synoptic setup is rather typical for late-season severe weather events with a deep trough/cut-off low to the west.

500mb Heights and Isotachs / Courtesy: SPC

While synoptic-scale quasi-geostrophic forcing was relatively weak, a small area of modest instability developed in the Mid Atlantic from southeastern Pennsylvania south toward the D.C. area. There was little CIN and a rather potent cold front was moving east through the Commonwealth of Pennsylvania with dew points in the mid 60s near Philadelphia (60 at KLNS prior to tornadogenesis) with mid 40s in central Pennsylvania. That was enough to fire convection and strong deep layer shear helped organize the convection into a powerful QLCS.

Here’s the 23z RAP initialization for KLNS (Lancaster) at 23Z.

The RAP analysis shows just shy of 1000 j/kg of surface based CAPE here in Lancaster about 50 minutes prior to tornadogenesis. While mid and upper level winds were screaming out of the southwest, the winds in the lower troposphere were somewhat unimpressive. Take a look at this 23z RAP hodograph from BUFKIT and you can see the unimpressive low level environmental shear.

Certainly nothing in the low level kinematic field that screams tornado to me! The  QLCS developed in a region of low CAPE/high Shear (70 knots bulk shear 0-6km) which is typical of late season northeast U.S. severe weather events.

Here’s the radar from KDIX at 2345 UTC and you can see the QLCS with an embedded supercell in eastern York County. The environment was supportive of supercellular structures within the QLCS  but what made this storm tornadic once it crossed the Susquehanna?

2345 UTC KDIX 0.5 degree Base Reflectivity (top) and 0.5 degree Storm Relative Velocity (bottom)

From the above hodograph it’s unclear what pushed this supercell to tornadic. There must have been some maximum in low level horizontal vorticity over Lancaster County to do it. A localized modification in the near-storm environment may have increased 0-1km shear enough? Any guesses blog readers?

As the storm’s mesocyclone tightened tornadogenesis occurred shortly after 2350 UTC based on radar (I think the PNS times from CTP are incorrect  by about 15 minutes) and hopscotched to the northeast for 16 miles with occasional touchdowns. What made this storm unique is that it provides the clearest example of a tornado debris signature that I have seen in the northeast U.S. since the dual polarization upgrades began.

Let’s review some of the dual polarization products before we dive into the details. For detecting a TDS you want to look at both correlation coefficient (CC) and differential reflectivity (ZDR).

Correlation coefficient looks at how similar the size and shape of a target is. Uniform drop size or flake size will result in high CC while irregularly shaped and different objects will result in lower CC (hail, biological scatters, or tornado debris).

Differential reflectivity detects a target’s shape. Horizontally enlongated targets like big rain drops have a high ZDR while randomly shaped targets (snow flakes or tornadic debris) has ZDR near zero.

To detect a Tornado Debris Signature (TDS) you need to look for a strong velocity couplet on SRV, very low CC, and near zero ZDR – all co-located  The Lancaster County tornado featured a surprisingly long lasting TDS.

Here’s a 2354 UTC radar grab from KDIX and you can see a lowering of CC crossing the Susquehanna River. That’s not a debris signature! It appears that there is a bit of small hail that is being detected by the radar. The radar beam is around 9500 ft AGL which is near the 0C level.

2354 UTC KDIX 0.5 Base Reflectivity, 0.5 Correlation Coefficient, 0.5 Differential Reflectivity, 0.5 Storm Relative Velocity (clockwise from top left).

No TDS here since the CC lowering is displaced from the velocity couplet. As we move on in time by 2358 UTC you can see a TDS. That small area of small hail or graupel is evident just southwest of the TDS but CC dropped dramatically between 2354 UTC to 2358 UTC with values as low as 50% next to low ZDR (near 0) and 65 knots of delta-V with that SRV couplet moving through Lancaster County.

2358 UTC KDIX 0.5 Base Reflectivity, 0.5 Correlation Coefficient, 0.5 Differential Reflectivity, 0.5 Storm Relative Velocity (clockwise from top left).

The tornado was an high-end EF-1 with winds approaching 110 mph. The radar beam here is more than 9000 feet above the ground so I was a bit surprised to see the TDS here! Unfortunately, there was no tornado warning on this storm in Pennsylvania (there was earlier in Maryland) even though dual pol was indicating airborne tornado debris.

By 0002 UTC KDIX shows the TDS likely embedded within an area of small hail or graupel. The signature persists, though becomes more muted, by 0015 UTC. It’s not a huge surrpise that the hail signal on CC was present as the updraft did strengthen through Lancaster County. Reflectivity over 50 dbz occasionally tickled -20C after the tornado touched down.

0002 UTC KDIX 0.5 Base Reflectivity, 0.5 Correlation Coefficient, 0.5 Differential Reflectivity, 0.5 Storm Relative Velocity (clockwise from top left).
0015 UTC KDIX 0.5 Base Reflectivity, 0.5 Correlation Coefficient, 0.5 Differential Reflectivity, 0.5 Storm Relative Velocity (clockwise from top left).

From what I could gather from news reports and the NWS CTP PNS the worst damage was near the Susquehanna River in Drumore Township, PA. It makes sense that the TDS was most visible at 2358 UTC and became more ambiguous as time went on. Lancaster County is also rural farm country! This tornado was not moving through heavily wooded area so there’s not necessarily a lot of debris for the tornado to pick up.

The radar showed a classic QLCS tornado. The S-shape pattern on reflectivity with an embedded mesocyclone right at the “kink” in the line.

While dual polarization cannot increase the lead time of tornado warnings it can play a role when one is on the fence with a tornado warning. In this case, use of dual pol could have been used to issue a torado warning after the initial touchdown. While there would have been negative lead time the warning the word would have been out for people farther northeast in Lancaster County.

Dual polarization most likely would have helped the NWS in Gray issue warnings for the 52 mile-long New Hampshire tornado that was only occasionally covered by a warning because of no ground truth reports at the time. If it happened today I bet we would have seen an impressive TDS on radar.

New Video Unearthed of 1989 Tornado Outbreak


The 1989 tornado outbreak was the most significant in the state during the 20th century. I recently was able to unearth some new video from the day after the storm. I also found a package from Cornwall showing the damage at Mohawk Mountain and the Cathedral Pines (fast forward to about 13:30)! I had feared this video was lost forever but I found it stashed in our archives this past weekend!

Several years ago I posted this video from our 11 p.m. newscast the night of the storm and also raw video of the damage in Bantam that Governor O’Neil toured on July 11, 1989.



How does the 1989 Cornwall to Hamden tornado stack up next to other big Connecticut tornado days? The 1979 Windsor Locks tornado was more intense but was relatively short lived only tracking through 2 towns prior to lifting. The 1878 Wallingford tornado likely rivaled the 1989 outbreak in terms of duration, length, and intensity.

I’ve written  about the infamous 1989 tornado outbreak previously here and here – though a more thorough discussion is warranted and will be done eventually.

Looking at some of the video I’ve uploaded and some other video clips and pictures I’ve seen I’m becoming more and more confident that the tornado was not F-4 intensity in Hamden. Here’s a BIG caveat: I’m basing the following statements on video being viewed 23 years after the fact! These are just observations and it is certainly possible I’m missing damage indicators that were available on storm surveys at the time.

Based on video there are some indications of F-3 damage (similar to what I saw in Monson and Brimfield, MA) most of the damage appears to have been F-1 and F-2. For example in Monson and Brimfield there were several cases of homes destroyed, swept off their foundations, and nearly unrecognizable. Most of the damage in Hamden featured roofs being blown off, second stories of homes destroyed, but in general the frames of homes remaining in tact.

The worst damage I can find is from this YouTube video uploaded by the Hamden Fire Department. At approximately 2:30 in you can see the damage at the corner of Augur and Newhall streets across from the Whitneyville Condominiums.


It’s not clear to me if that pile of debris is from the second floor of the white house on the right or if it’s a separate house that was leveled. While severe, I just don’t see F-4 damage indicators.

Furthermore, there are indications from Litchfield County that the tornado that tracked from Cornwall to Bantam may have been stronger than F-2. While we’ll never know the storm’s true intensity from what we know now about damage indicators today and the approximately half hour of damage video from July 1989 my best estimate is F-1/F-2 damage in Cornwall, F-2/F-3 damage in Bantam/Litchfield, and F-3 damage in Hamden.

Regardless of exactly how strong the tornado was the long track and duration of the tornadoes that touched down that day in 1989 is exceptionally impressive.

Long Island Tornado

OKX 0.5 degree SRV

While the storms in Connecticut didn’t produce a tornado there was a tornado that touched down on Long Island. The 4.5 mile EF0 tracked from Great River to Ronkonkama producing tree damage across its path.  You can see the mesocyclone about 1000 feet above the ground here on the OKX storm relative velocity image. Not terribly impressive but it was enough to drop a weak tornado.

The developing MCV (MCV within an MCV?) tightened up once over the Connecticut shoreline. The added frictional drag over land once past Long Island Sound may have helped tighten up the small vortex (and associated mesolow) in the boundary layer. While the vortex began tightening this is what the attendant mesocyclone looked like over Guilford.

OKX SRV 0.5 degree 1931Z

That’s an impressive couplet (the best it looked all day) with 60 knots of gate-to-gate shear. I can’t ever recall seeing such a signature over my hometown since the radar was installed in Brookhaven in ’94.

Here’s a loop of OKX base reflectivity showing how the unusual storm evolved across the state (click to animate).