Sehr Schweres Unwetter in NRW

Not having full command of the German language, “sehr schweres Unwetter” seems like an understatement. It translates as “very bad thunderstorm,” which in this case is like calling the Titanic a “very big boat”. Of course, if you live in the Great Plains, you probably refer to a supercell thunderstorm as “a little bit of rain and wind” but the storms that hit Nordrhein-Westfalen (NRW) on 9-10 June 2014 rival anything the toughest Oklahoman has experienced (minus the tornadoes). Also, keep in mind that Germany and the Low Countries have nowhere near the wide-open spaces the U.S. Great Plains are known for. Take 5 times the population of Oklahoma and cram them into a land area the size of Maryland. (Or, if you’re from Maryland, multiply your state’s population by three to approximate the population density of the area we’re talking about. Then ponder how anyone in that part of Germany is able to spend less than 18 hours per day stuck in traffic like you would be if you were suddenly surrounded by three times as many people.)

Let me set the scene for you. (If you’ve ever lived in the Midwest, you know the drill.) The air is hot and unbelievably humid. The sky is overcast. There is no wind to speak of, but there is a certain “electricity” in the air that tells you that a violent end to the heatwave is coming. Off in the distance, clouds lower and darken. A gentle rumbling of thunder slowly builds as the storm approaches. Lightning appears and becomes ever more frequent. Right before the storm hits, the winds pick up out of nowhere and… Wait! I don’t need to describe it. I can show it to you:

EDIT: I did need to describe it, because the videos are no longer available. If you weren’t able to see the videos before they were removed, they showed scary looking clouds and nearly constant lightning approaching Bochum. In fact, there were an estimated 113,000 lightning strikes across Germany from the storm.

Germany is, apparently, a land of iPhones and GoPros and all sorts of video recording equipment, and there is no shortage of video of the storm. There are videos of the storm approaching from different perspectives (here, here and here), the strong winds and heavy rains that are more reminiscent of a tropical storm (here, here and here), footage of the lightning in slow-motion and, because this is the Internet, a 30 min. montage of storm footage set to salsa music (although one commenter says the first footage is from a storm in 2010).

The aftermath is pretty impressive also – trees and large branches down everywhere blocking roads, crushing cars and stopping the never-late German train system. In fact, 6 people were killed – mostly by falling trees. Winds were observed in the 140-150 km h-1 range (approximately 85-90 miles per hour), which puts it just below a Category 2 hurricane according to the Saffir-Simpson scale. There were even reports of baseball sized hail, something that’s not unusual in Oklahoma, but is very rare in Europe. (Here is some pretty big hail in the town of Zülpich from earlier in the day.)

Now that you’ve used up the last 90 minutes looking at YouTube videos, let’s get down to business. What do satellites tell us about this storm?

EUMETSAT put together this animation of images from the geostationary satellite Meteosat-10:

Watch that video again, preferably in fullscreen mode. First, the white boxes highlight the supercell thunderstorms over Europe between 01:00 UTC 9 June 2014 and 08:15 UTC 10 June 2014. Right before sunset on 9 June, you can see a storm moving north out of France into Belgium that seems to explode as it heads towards the Netherlands and western Germany. This is our “schweres Unwetter”. The second thing to notice is where that storm is at 02:00 UTC on the 10th. That was the time that VIIRS passed overhead.

So, without any more bloviating, here’s the high-resolution infrared (I-5) image from VIIRS:

VIIRS I-5 image from 02:07 UTC 10 June 2014

VIIRS I-5 image of severe thunderstorms over Europe from 02:07 UTC 10 June 2014

The storm that caused all the damage over Nordrhein-Westfalen has weakened and is now over northeastern Germany on its way to Poland. But, a second impressive supercell complex is pounding Belgium and the Netherlands, and taking aim at western Germany once again.

The coldest pixels are 196.5 K (-76.7 °C or -106 °F) in the storm over Benelux and 198.7 K (-74.5 °C or -102.1 °F) in the storm over northeast Germany. Another impressive thing about these storms is their size relative to the size of these countries. That Benelux storm looks like it’s at least five times the size of Luxembourg and as big as Belgium! (And I’m not counting the area of the anvil, which is even larger. I’m only counting the area containing overshooting tops.)

Since it’s nighttime, what did the Day/Night Band see? Well, the answer depends on how you display the data. You see, we’re approaching the Summer Solstice in the Northern Hemisphere, where the days are long and twilight encroaches the nighttime overpasses at these latitudes. If you try to scale the radiances from lowest = black to highest = white, you get something like this:

VIIRS Day/Night Band image, taken 02:07 UTC 10 June 2014

VIIRS Day/Night Band image, taken 02:07 UTC 10 June 2014. Radiance values are displayed and scaled according to text above.

That’s not very helpful because the radiance values vary by 6 orders of magnitude across the scene and we only have 256 colors to work with to relay that information. But, we can take advantage of the fact that the Day/Night Band radiance values are, to the first order, a function of the solar and lunar zenith angles, and use this as the basis for a “dynamic scaling” that compares the observed radiance with an expected maximum and minimum radiance value that is a function of those angles. (In case you’re interested, the dynamic scaling algorithm used here is based around the error function.) This allows you to produce something like this:

VIIRS Day/Night Band image, taken 02:07 UTC 10 June 2014

VIIRS Day/Night Band image, taken 02:07 UTC 10 June 2014. This image uses dynamic scaling as described in the text.

Here, we’ve lost some quantitative information (colors no longer represent specific radiance values) but we’ve gained valuable qualitative information.  Now we can see where the storms are! Notice the shadows in the overshooting tops of our Benelux storm – right where the coldest pixels are in the infrared image. We can see some of the city lights, but not others, because the twilight encroaching from the northeast is brighter than the cities in that part of the image. (It is easy to pick out London and Paris, though.) If you read the previous post, you might be wondering why there are no mesospheric waves with these storms. That’s because there is too much twilight (and moonlight) to see the airglow. (There’s also the possibility that the stratosphere and mesosphere weren’t conducive for vertically propagating waves, but you wouldn’t be able to tell that under these lighting conditions.)

Some people like to combine the infrared with the Day/Night Band into a single image. This is done by changing the opacity of one of the images and overlaying it on the other. Here’s an example of what that looks like using the dynamically scaled Day/Night Band image:

VIIRS combined IR/DNB image from 02:07 UTC 10 June 2014

VIIRS combined IR/DNB image from 02:07 UTC 10 June 2014

The light/shadow effect of the visible information adds a sort-of 3-D effect to the infrared images and, since this is the Day/Night Band, it can show where the storms are in relation to the urban areas. Here, it seems to work better for the Benelux storm than it does for the other one. (Of course, it would be better without the twilight. And, it works best with a full moon, which occurred three days later.)

Of course, if you have access to the Near Constant Contrast imagery, you don’t have to worry about scaling. The imagery is useful as-is:

VIIRS NCC image, taken at 02:07 UTC 10 June 2014

VIIRS NCC image, taken at 02:07 UTC 10 June 2014

And the combined IR/NCC image looks like this:

Combined IR/NCC image from 02:07 UTC 10 June 2014

Combined IR/NCC image from 02:07 UTC 10 June 2014

In case you’re interested, there are additional videos, animations and images of these storms from the Meteosat High Resolution Visible (HRV) channel at the EUMETSAT Image Library.


Severe Weather in the Mesosphere

So far (*knock on wood*), it’s been a pretty quiet year for severe weather. If you only count tornadoes, there have been 81 tornado reports from 1 January to 4 April this year. (11 of those have come just this week.) This is a lot fewer than the previous three year average of 192 tornadoes by the end of March. For that, you can thank the dreaded, terrifying “Polar Vortex” you’ve heard so much about over the winter. Tornadoes don’t like to come out when it’s cold everywhere. (Although, there was a notable exception on 31 March 2014, when a tornado hit a farm in Minnesota when the area was under a blizzard warning.)

I just said that there have been 11 tornado reports this week. Eight of those came in the past 24 hours. At the southern end of the line that brought the tornadoes to Illinois, Missouri and Texas, the severe weather included golf ball-size hail and this:


That report came from the National Weather Service in Corpus Christi, TX and it was caused by non-tornadic straight-line winds in Orange Grove. Winds capable of ripping a shed out of the ground, combined with golf ball-sized hail – that’s one recipe for broken windows. And it’s not a pleasant way to be awakened at 4:30 in the morning.

A couple of hours earlier, VIIRS caught this severe storm as it was rapidly growing. Here’s what the storm looked like in the high-resolution infrared channel (I-5, 11.45 µm):

VIIRS high-resolution IR image (channel I-5), taken at 08:13 UTC 4 April 2013.

VIIRS high-resolution IR image (channel I-5), taken at 08:13 UTC 4 April 2013.

Make sure you click on the image, then on the “2999×2985” link below the banner to see the full resolution image, which, for some reason, is the only version where the colors display correctly.

The storm that hit Orange Grove is the southern-most storm, with what looks like a letter “C” imprinted on the top. (That kind of feature typically looks more like a “V” and makes this an “Enhanced-V” storm, which you can learn more about here. Enhanced-V storms are noted for their tendency to produce severe weather.) For those of you keeping score at home, the coldest pixel in this storm is 184.7 K (-88.5 °C).

Compare the image above with the Day/Night Band image below (from the same time):

VIIRS Day/Night Band image, taken at 18:13 UTC 4 April 2014

VIIRS Day/Night Band image, taken at 08:13 UTC 4 April 2014

There are a few interesting features in this image. For one, there’s a lot of lightning over Louisiana, Arkansas and Mississippi. (Look for the rectangular streaks.) There’s even some lighting visible where our “Enhanced-V” is. Two, it takes a lot of cloudiness to actually obscure city lights: only the thickest storm clouds appear to be capable of blocking out light from the surface. Three: there are a lot of boats out in the Gulf of Mexico at 3 o’clock in the morning (and a few oil rigs as well). And four: notice what appear to be concentric rings circling the location where our severe storm is with its enhanced-V.

In this image, there is no moonlight (we’re before first quarter, so the moon isn’t up when VIIRS passes over at night). The light we’re seeing in those ripples is caused by “airglow”, which we’ve seen before. And the ripples themselves may be similar to what is called a “mesospheric bore.” If you don’t want to get too technical, a mesospheric bore is when this happens in the mesosphere. They are related to – but not exactly analogous to – undular bores, which you can read more about here.

Unlike the situation described for the undular bore in that last link, the waves here are caused by our severe storm. To put it simply, we have convection that has formed in unstable air in the troposphere. This convection rises until it hits the tropopause, above which the air is stable. This puts a halt to the rising motion of the convection but, some of the air has enough momentum to make it in to the stratosphere. This is called the “overshooting top“, and is where our -88°C pixels are located. (Look for the pinkish pixels in the middle of the “C” in the full-resolution infrared image.) The force of this overshooting top creates waves in the stable layer of air above (the stratosphere) that propagate all the way up into the mesosphere. The mesosphere is where airglow takes place, and these waves impact the optical path length through the layer where light is emitted. This of course, impacts the amount of light we see. The end result: a group of concentric rings of airglow light surrounding our storm.

You could make the argument that the waves we see in the Day/Night Band image are not an example of a bore. Bores tend to be more linear and propagate in one direction. These waves are circular and appear to propagate in all directions out from a central point. It may be better to describe them as “internal buoyancy waves“, which are similar to what happens when you drop a pebble into a pond. Only, in this case the pebble is a parcel of air traveling upwards, and the surface of the water is a stable layer of air. Compare the pebble drop scenario with this video of a bore traveling upstream in a river to see the difference.

In fact, if you look closer at the Day/Night Band image, in the lower-right corner (over the Gulf of Mexico) there is another group of more linear waves and ripples in the airglow that may actually be from a bore. It’s hard to say for sure, though, without additional information such as temperature, local air density, pressure and wind speeds way up in that part of the mesosphere.

By the way, you can see mesospheric bores and other waves in the airglow if you have sensitive-enough camera, like the one that took this image:

Photograph of a mesospheric bore. Image courtesy T. Ashcraft and W. Lyons (WeatherVideoHD.TV)

Photograph of a mesospheric bore. Image courtesy T. Ashcraft and W. Lyons (WeatherVideoHD.TV)

And, if you’re interested, the Arecibo Observatory has a radar and optical equipment set up to look at these upper-atmosphere waves (scroll down to Panel 2 on this page). The effect of these waves on atmospheric energy transport is a hot topic of research.

Golf ball-sized hail at the Earth’s surface is related to energy transport 100 km up in the atmosphere!


NOTE: This post has been updated since it was first written to clarify that the circular waves are likely not evidence of a bore, as was originally implied. They are more likely internal buoyancy waves, which are also known as gravity waves. For more information, consult your local library.

The Last Line of Storms from the 14 April 2012 Tornado Outbreak

The second major tornado outbreak of the year took place on 14 April 2012 (after the 2 March outbreak that slammed Indiana and Kentucky). At last count, 115 tornadoes were reported from Oklahoma to Iowa. Credit must be given to the Storm Prediction Center, National Weather Service offices, and local TV and other media outlets for accurately predicting the severe weather event and keeping people informed as it happened, and the people of the area for paying attention to the weather. It must be counted as a success on many levels that 115 tornadoes over 4 states only resulted in 6 deaths (and those deaths occurred in the toughest situation to warn people – a rain-wrapped tornado in the middle of the night where the tornado sirens were disabled due to a lightning strike earlier in the day).

The last bout of severe weather occurred with a squall line that formed in the late evening (~02:30 UTC 15 April 2012) along the dry line in western Texas and quickly expanded into Oklahoma and Kansas. This line produced the deadly tornado in Woodward, OK, along with many reports of 1-2″ diameter hail. Suomi-NPP passed over this line of storms between 07:45 and 07:50 UTC (15 April). The high resolution infrared window band, I-5 (11.45 µm), shows the immense scale of this storm system stretching from Wisconsin and Minnesota to Texas, in great detail. Be sure to click on the image, then on the “1497×1953” link below the banner to see it in full resolution. (The full resolution image is ~2MB in size.)

View of a squall line over the Central Plains from VIIRS channel I-5, 7:45 UTC 15 April 2012

View of the squall line over the Central Plains from VIIRS channel I-5, 7:45 UTC 15 April 2012

The color scale here is the same one used for the 2 March 2012 tornado outbreak image and the 25 January squall line over southeast Texas. The darkest blue pixels visible amongst the white overshooting tops (more easily visible on the southern end of the squall line) have a brightness temperature below -77 C, indicative of very strong convection.

VIIRS View of March 2 Tornadic Storms

NPP/VIIRS passed over Southern Indiana on March 2 about thirty minutes before the most devastating tornadoes struck the towns of New Pekin and Henryville (among others).  At 1935 UTC, a pair of rotating thunderstorms, also known as supercells, were advancing eastward across Indiana.  The easternmost storm spawned the most damaging tornadoes.  Below is a VIIRS true color image from the NPP pass at 1935 UTC.

VIIRS True Color image of the severe storms on 2 March 2012 at 1935 UTC.

A zoomed-in visible view of the storms is below.

VIIRS I-band 1 (375-m resolution) from 2 March 2012 at 1935 UTC

The infrared (I-band 5) image is below, along with some annotations pointing out the two active supercells discussed above.  Note that the brightness temperatures associated with the overshooting top (OST) of the westernmost storm are colder than the easternmost storm, although both storms were quite strong at the time and the eastern storm ended up producing the deadlier tornadoes.  OSTs are transitory, so it’s possible that a new cold OST formed with the eastern storm shortly after the NPP pass.  These very high resolution infrared views of tornadic storms are among the first documented, given the recent launch of NPP.

VIIRS I-band 5 Infrared view from 2 March 2012 at 1935 UTC

To illustrate the effect of high resolution in the IR, below is a GOES-13 10.7 micrometer IR image from 1932 UTC, which has 4-km resolution at nadir.  The coldest brightness temperature in the westernmost storm in southern Indiana from GOES is 206.6 K, but with VIIRS it’s 195 K.

GOES-13 4-km IR Image from 1932 UTC on 2 March. Compare this image to the 375-m VIIRS image above to see the improvement provided by VIIRS over GOES.

The day after the tornadoes, relatively cloud-free skies in eastern Kentucky allowed VIIRS to see some of the tornado tracks.  In the image below, the faint white lines circled in red in Kentucky and West Virginia denote the new tornado damage paths.  When green vegetation is disrupted/destroyed, the result is typically a brighter scene at visible wavelengths.

VIIRS I-band 1 from 3 March 2012 over eastern KY and western WV. The tornado tracks are circled and show up as faint white lines


A squall line over Texas as seen by VIIRS

VIIRS RGB "true color" composite

A severe squall line formed over eastern Texas on 25 January 2012. There were 19 tornado reports and 48 reports of wind damage, including “a house destroyed by a possible downburst”, according to the Storm Prediction Center. The high resolution imager on VIIRS captured this squall line as it was rapidly intensifying. Shown below are images collected from channel I-5, the high-resolution infrared window channel (11.45 μm). (Click on images for full resolution.)

VIIRS Channel I05

A squall line over eastern Texas observed by VIIRS channel I05 (11.45 um) at 19:24 UTC on 25 January 2012.

This squall line had several overshooting tops over the Gulf of Mexico that reached a temperature of -77 C. A zoomed-in view of these tops are shown below.

VIIRS Channel I05

A squall line over eastern Texas observed by VIIRS channel I05 (11.45 um) at 19:24 UTC on 25 January 2012.

The dark blue pixels near the center of the image indicate an overshooting top approximately 5 km in diameter where temperatures were less than -77 C. Several pixels in a storm top at the bottom center of the image and in a storm top at the top center of the image (near Galveston, TX) also reached that temperature.

A sounding was taken at 18:00 UTC at the Lake Charles, LA, National Weather Service (NWS) office, which observed a minimum temperature of -74 C at 17.9 km above sea level, indicating that these are some tall thunderstorms. Image courtesy the University of Wyoming.

Radiosonde sounding

NWS sounding taken at 18:00 UTC from the Lake Charles, LA office.

The VIIRS imagery was collected right as the squall line was intensifying. Shown below is the radar loop from the Houston/Galveston radar between 18:00 UTC and 21:00 UTC. Note, at the beginning of the loop, the southern end of system consists of two rather disorganized lines of cells. These lines of cells merge at around 19:25 UTC (the time of the Suomi NPP overpass), and a much stronger and more organized squall line develops.

Radar loop

Radar loop from the Houston/Galveston NWS WSR-88D radar beginning at 18:00 UTC, 25 January 2012.

At roughly 375-m resolution at nadir, the I-5 channel on VIIRS is providing some of the highest resolution infrared imagery available to the atmospheric science community. We are just beginning to see the capabilities of this powerful instrument.