Thursday, August 13, 2020

The August 10, 2020 Midwest Derecho - how did it develop?

Winds from the derecho on Monday, August 10 (pronounced 'day-ray-cho', a widespread convective system with high winds over a large area that can affect several states) killed two people, one in Iowa and one in Indiana, while leaving over a million people without electricity.  The images above show the derecho moving into Chicago at mid-afternoon, the derecho's shelf cloud approaching Sioux City, Iowa on Monday morning, and major damage in Marshalltown, Iowa at mid to late morning.   

The derecho capped off a very active (for August) 7-day period of severe weather.  August 3-4 saw numerous tornadoes associated with the remnants of Hurricane Isaias from North Carolina (where 2 people died in a nighttime EF3 tornado) to Virginia, Maryland, Delaware, Pennsylvania, and New Jersey.  Then an EF2 tornado on the evening of August 7 killed two people in southwest Manitoba, Canada.

Monday's derecho started in the early morning hours over South Dakota, crossed Iowa during the morning, and hit Chicago and northern Indiana in the afternoon.  Here's an hour by hour composite radar image from NWS Chicago that shows the derecho's rapid progress:

And here's the associated storm reports showing the wide axis of damage across several states, including some embedded "bow echo" or QLCS-type EF0-EF1 tornadoes in the Chicago area:

How does a derecho like this form?  Large convective systems are pretty common across the Midwest in summertime.  So, what causes one of these systems to become a dangerous and deadly squall line or cluster of storms?  Answers have to do with the origin area of the derecho (south-central/southeast South Dakota in this case), as well as the synoptic setting out ahead of it.

Below is the surface map at 5:00 am CDT on 8/10/20, along with a radar inset focused on south-central South Dakota (SD).  I've indicated the genesis region of the derecho, where an area of thunderstorms had formed during the early a.m. hours within the thick black oval on the surface map, north of a frontal boundary over Nebraska:

A RAP model forecast sounding at Lake Andes, SD, just east of these thunderstorms, is shown below:

Notice that there was a significant layer of dry air in the lowest 3 km, indicated by the broad distance between the red temperature and blue dew point curves in lower levels.  This was _below_ a relatively moist layer beginning at roughly 3 km above ground where significant CAPE was present from a lifted parcel at that elevated level.  With little or no convective inhibition (CIN) at this level, elevated storms could initiate well north of the surface front over Nebraska (see 5:00 am surface map earlier), and rain into this dry layer below.

This would produce strong evaporative cooling, creating dense cool air accelerating downward and outward and generating strong surface winds beneath and ahead of the expanding elevated storms over southern SD.  Here's a diagram illustrating that process for many summertime derechos that initiate in the northern Plains of the U.S.:

This may look similar to diagrams of downbursts and microbursts, which are much more localized.  But what is different and important in Monday's case is that the evaporative cooling and downward air acceleration was taking place over a larger area with the _cluster_ of developing storms, rather than a single thunderstorm.  This then spread out into the broad squall line and convective system seen in the composite progressive radar image earlier.

I should emphasize that soundings like the one at Lake Andes up above, with a moist layer of sizable elevated CAPE above a significant layer of dry air (a type of  'overrunning' situation, north of the Nebraska frontal boundary seen on the earlier surface map) are not that common.  With a developing thunderstorm cluster, it's a little unusual to see such an elevated moist layer located that far above a depth of much drier air beneath.

Below is the 500 mb NAM model forecast for 4:00 am CDT in the mid-levels of the atmosphere on 8/10/20, showing a significant shortwave disturbance (thick dashed black line) moving across the northern Plains. This disturbance provided the upward forcing that helped fire up the elevated thunderstorms over southern SD that evolved into the derecho:

The inset on the graphic above is a forecast of mlCAPE valid at 7:00 am CDT, showing a long axis of instability extending eastward to Chicago, a corridor that would feed and maintain the convective complex and derecho as it evolved and moved eastward across several states during the morning and afternoon.

A similar situation accompanied a developing derecho that my wife Shawna and I experienced at Pierre SD in the early morning hours of 6/22/15, five years ago.  Below is the surface map with the genesis region of the derecho indicated around Pierre at 1:00 AM CDT on 6/22/15:

Here is the RAP model sounding at Pierre SD at about the same time:

Notice that the environment was similar to the 8/10/20 "genesis region" sounding shown earlier, with an elevated moist layer and sizable CAPE located _above_ a dry layer in the lowest 3 km.  As new storms formed west of Pierre around midnight, evaporative cooling of rain through this dry layer resulted in the generation of very strong winds (120+ mph measured at Hayes SD!) that caused widespread damage in Pierre (including our motel) and locations eastward on 8/22/15 as the storms morphed into a derecho event.

Many derechos that initiate over the northern Plains and move eastward through the Midwest (the most common area for derechos in the U.S.) probably have this type of atmospheric setting in their genesis region.  The drier air in the lowest 2-3 km below an elevated moist layer can result in rapidly-developing strong convective surface winds that spread eastward as a derecho along a west to east corridor of instability.  

An important factor is the presence of a stationary or quasi-stationary west-east front (check out the surface maps earlier) with this axis of unstable air along it, helping to provide a corridor along which the recently-intiated derecho can intensify, spread, and maintain itself moving eastward.  This was a key feature of Robert John's seminal work on derechos at SPC in the 1980's (see this paper).  Johns deserves much credit for making forecasters and meteorologists more aware of these dangerous convective systems.

Derechos are more complicated than this brief analysis would suggest.  But I've touched on a few factors that are important for many warm season derechos across the Midwest.

- Jon Davies 8/13/20 

Sunday, August 2, 2020

A 'surprise' tornado on July 29, 2020 north of Kansas City - another subtle, marginal summer tornado setting

I've had several people ask me what was going on with the unexpected but well-photographed tornado (above, EF0) last Wednesday 7/29 at early afternoon near Smithville, MO, north of Kansas City, only a few mile north of where I live.  I was quite surprised when I heard about the warning associated with it, and wondered myself how there could be a tornado on this particular day.

Although not something one could really anticipate, a careful look back at the setting revealed that it was not 'random' or 'from out of the blue'.

There was a well-defined boundary (dashed red-blue line below; a weak stationary front?) on the surface map at noon on 7/29/20 across the Kansas City metro area, delineating south winds to the south of the boundary from light easterly winds on the north side of it:
As many forecasters and researchers have pointed out in recent years, tornadoes like boundaries, often because favorable wind shear (a change in wind direction and speed as winds veer with height) that might help generate a 'spin up' can be present along and across such boundaries.  The black-circled 'S' on the map above shows that the location of the small supercell that later spawned the weak tornado was indeed very close to the boundary.

Here's satellite and zoomed-in radar reflectivity near the time of the tornado (1:45 pm CDT, or 1845 UTC), with a white circle showing the location of the supercell mesocyclone and rotation in the storm near Smithville:
It is also interesting that a broad, weak low was present at 700 mb (roughly 10,000 ft above sea level) nearby over southeast Nebraska at midday on the SPC mesoanalysis:
This weak low aloft and a short wave trough (heavy black dashed line) moving around it helped provide lift for generating thunderstorms over northeast Kansas and northwest Missouri.  In a way, this looked like a weak 'cold-core' tornado setting, and coupled with the east-west surface boundary, was another factor that may have helped set the stage for a tornado.

The storm environment, though subtle, also offered some ingredients that were marginally supportive of supercell tornadoes.  The first panel below from the SPC mesoanalysis at 1800 UTC (noon CDT) showed that low-level wind shear, though not large, was enhanced some just north of the east-west boundary (0-1 km storm-relative helicity / SRH > 50 m2/s2), as one might expect with easterly surface winds just north of Kansas City on the surface map earlier:

The 2nd panel above showed that instability in low-levels close to the ground was large (0-3 km mlCAPE > 150 J/kg) near the boundary across the Kansas City area.  Although subtle, combinations of both low-level shear and instability together were largest in that area (see black oval in 1st and 2nd panels above), just south of Smithville.

One last ingredient was deep-layer wind shear (surface to 6 km above ground, 3rd panel above) that was near the lower limit (around 25 kt) of what is considered supportive for supercell storms.  But on this day it was enough to do the trick.

Here's a RAP model analysis sounding estimate of the environment near Smithville at midday along the east-west boundary:

Although not impressive, the instability (which I've manually shaded in red above) and vertical wind shear were apparently just enough to help generate and support a tornado as the supercell crossed the surface boundary moving north-northeast.  

In particular, the instability in lowest levels close to the ground probably helped with stretching of air in the small storm's updraft as it interacted with the local boundary.  The photo below shows a cloud ridge flowing into the storm with 'vapory-looking' scud close to the ground, a typical visual 'look' when large amounts of low-level CAPE are present not far above ground, and low-level stretching in updrafts may be enhanced.

This weak event wasn't of much importance (brief tornado, no damage), but is another illustration of how subtle ingredients can come together to produce a mesoscale 'accident'.  Given how marginal the setup was, an experienced nowcaster and meteorologist would not expect anything more than a weak tornado or two from such a setting.

- Jon Davies  8/2/20