On August 24th, 2016, over 20 tornadoes tracked across the states of Indiana and Ohio in what was one of the largest tornado outbreaks on record in the month of August. What made this event particularly notable was how unforeseen the outbreak was by professional meteorologists and forecasters. Tornadoes require a certain set of ingredients to manifest, namely: vertical wind-shear, atmospheric instability, and a triggering mechanism. The superposition of these ingredients is generally well forecast at least a day in advance for most major events. On August 23rd, the day before, the Storm Prediction Center (SPC) in Norman, Oklahoma forecasted no tornadic storms to occur in or near Indiana on the following day. It wasn’t until the event had essentially begun that the threat was truly assessed. The outbreak has since been referred to as the Surprise Indiana Outbreak.
An analysis at US Tornadoes by meteorologist and UIUC Professor Jeff Frame goes into great detail as to how this particular event occurred and why it was so unexpected. The author concludes that a remnant MCV, or mesoscale convective vortex, strongly contributed to the tornado threat over Indiana that day. More so, he concludes that it was the unpredictability of such a feature, along with concerns regarding convective mode, that allowed the event to go unnoticed by the nation’s leading experts at the SPC and elsewhere.
A mesoscale convective vortex is an area of low pressure in the middle levels of the atmosphere (2-3 miles high). The vortex typically spans a width of 30-60 miles across and can persist for many hours. Particularly strong MCVs can take on a shape and spin reminiscent of tropical cyclones.
These features do not form on their own but rather form within clusters of thunderstorms called mesoscale convective systems (MCSs). MCSs form most often in the Midwest during the late spring and summer months. They are the result of rich moisture flowing north from the Gulf of Mexico overriding fronts that typically reside over the upper Midwest during the warm season. This triggers powerful thunderstorms that then coalesce into these larger systems — an MCS (shown below). This process usually begins in the afternoon with the initiation of discrete thunderstorms and then the system coalesces during the overnight hours.
The vast quantities of cloud droplets and rain that are produced within these systems release latent heat (the energy that is released when water changes phase from gas to liquid) in the middle levels of the atmosphere. This heat and subsequent expansion of layers of air aloft induces rotation through a redistribution of what meteorologists call potential vorticity. Once this process occurs, an MCV develops and it can even outlast the thunderstorm cluster that spawned it, possibly traveling several hundred miles over 24 hours. An MCV increases low-level wind shear (the change in wind speed or direction between the surface an altitude of 1 km) particularly within the feature’s southeast quadrant. Low-level shear has been identified as a vital ingredient for the formation of tornadoes (tornadogenesis).
Beyond increasing the local tornado potential, MCVs also provide ascent ahead of them and can initiate new thunderstorms themselves. In some cases, a single vortex can regenerate thunderstorms for 48 hours or more.
The video loop below shows the visible satellite imagery over the Midwest on 8/24/16, the day of the Indiana tornado outbreak. You can see the remnant MCV spinning over northern Illinois and Indiana as it drifts east. The tornado warnings are overlaid in red as they were issued. Notice how virtually all the warnings are straddling the feature’s southeast side. The connection between the MCV and the tornadoes can might be inferred from this video animation alone.
August 24th was a prime example of how MCVs can drastically change a severe weather forecast. MCVs have been attributed to other tornadic events in recent months as well. Some two weeks after the Indiana outbreak, on September 9th, a smaller but notable event occurred just a hundred miles west in central Illinois. Again, a remnant MCV tracked into the Ohio Valley from the west and sparked 3 separate tornadoes, one of which was classified as significant (EF2 strength or greater).
These storms also fired within the SE quadrant of a remnant MCV and produced tornadoes in an otherwise low-risk environment (i.e. one without an obvious abundance of parameters favorable for tornadoes). Below is a reanalysis from 9/9/16 with the visible satellite imagery overlaid just as storms began to initiate in south central Illinois. I have drawn the approximate location of the MCV (black), warm front (red), and tornado reports (blue).
— Andrew Pritchard (@skydrama) September 10, 2016
This event was better forecasted by the SPC as the risk of tornadoes was accounted for in their morning outlook and more demonstrably so just a few hours before storms initiated. Forecasters made note of the developing MCV moving from Missouri into Illinois and forecast an increasing risk of tornadoes accordingly.
Regardless of these recent events, MCVs remain an elusive forecasting challenge as they can be difficult to predict well in advance, as their present is completely dependent on antecedent thunderstorms. The features can also be poorly resolved or misplaced by numerical weather prediction models used by forecasters.
Regardless of forecast challenges associated withh these events, the local National Weather Service Forecast Offices in Indianapolis, Indiana, and Lincoln, Illinois (KIND and KILX respectively), issued timely and accurate warnings in both cases. Despite homes and properties suffering significant damage, there were no serious injuries reported between either day — a credit to the local forecasters.
If these latest storms have a lesson to teach us, it’s that tracking and understanding MCVs can be particularly important when forecasting late-season (late June – September) corn belt tornado potential, especially considering climatological data. As mentioned earlier, MCSs form most often during the warm season in the Midwest, and these complexes of thunderstorms frequently spawn MCVs.
An MCS can persist through the night, without the sun, as long as the atmosphere remains unstable. Instability can be thought of as the fuel for thunderstorm development. What makes these nighttime storms so interesting is that they are able to persist despite a lack of solar heating which is an important source of instability, and in the presence of longwave radiative cooling, a stabilizing effect. In fact, it has been observed that MCSs can become more intense after the sun goes down.
There are a few theories as to why this occurs. The first accounts for a classic feature of the Great Plains called the nocturnal low-level jet (LLJ). This is a fast corridor of winds about a mile above the surface that blows from south to north carrying rich Gulf moisture with it, helping to refuel nighttime thunderstorms. The LLJ increases at night due to the cessation of vertical mixing and the loss of surface friction at these elevations.
Another contribution to this peculiar spike in intensity is theorized to be the effects of nighttime long-wave radiation at the top of a MCS’s thunderstorm clouds. As the top of the thunderclouds release radiation (heat) upward at night, it cools the top layer of the storms. This increase leads to a more rapid decrease in temperature with height and greater instability, which we have prescribed as the fuel for thunderstorms. This allows the MCV to form and continue overnight and even into the next day. This strange ability to live on overnight and impact the next day’s weather and tornado potential makes these features all the more ominous. It is also this seemingly undying that inspired the title of this review: Zombie MCVs.
Below is an infrared (longwave) satellite image of the Midwest at 6am (EST) 8/24/16 before sunrise in Indiana. You can clearly see a large cloud shield over NW Illinois, reflective of the MCS ongoing at this time. This overnight convection led to to the MCV which contributed to the Indiana outbreak that afternoon.
An important distinction to make before going further is between any tornadoes that might form within the overnight storms and the ones that might form from new storms the next day triggered by the MCV. Tornadoes that form within an MCS are often embedded in a line of storms and are thus often weak and brief (discrete supercell thunderstorms are known to produce more violent long-track tornadoes). If the storms decay the next morning, it allows the sun to reheat the surface and increase surface instability while the environment retains the low-level spin from the overnight MCV. It was this transition that made the Surprise Indiana Outbreak both hard to forecast and conducive for tornado development as the storms formed during the afternoon were generally discrete.
Another feature of a MCS that is vital to its forecast is its cold pool. When storm complexes rain, they create cold air beneath them through evaporative cooling (when water changes from liquid to gas it cools the air instead of heating it — this is why you often feel cold when you are wet). This cold air then spreads away from the storms and acts as an effective cold front, helping to trigger new thunderstorms. If the MCS dissipates the next morning or afternoon, one would expect a remnant cold pool in the vicinity, where temperatures are cooler than the surrounding environment. Since MCSs often form near temperature gradients (fronts), their own cold pools can reinfornce these fronts, sometimes making them a better focus for thunderstorms the next afternoon. A common forecasting conundrum, even without a remnant MCV, is predicting where a cold pool situates after the passing of its parent MCS.
In both recent events discussed above, the boundary between the cold pool and the warmer environmental air (the gust front or outflow boundary) was east of the MCV and acted as an effective warm front as the day went on. This warm front then moved northward on both days, helping to initiate supercell thunderstorms (the relative contributions of low-level shear from the warm front versus the MCV are of interest to the author).
Below are the surface analyses from the SPC mesoscale discussions from both 8/24 (left) and 9/9 (right). In both images you can see the warm front (black scallops) east of the MCV (marked as an X within a circle).
The similarities between the two events are evident in the images alone. Interestingly, the environment was similar in both cases yet the SPC used much stronger language and detail regarding the tornado threat on 9/9/16 which could be at least partially in response to the surprise 8/24/16 outbreak.
Accounting for the outflow boundary and its ability to transition into and/or merge with a warm front the next day gives the environment near an MCV another ingredient for tornadic storms: a triggering mechanism. Warm fronts can initiate storms and can also provide additional low-level shear as winds shift direction with height near the surface. This is in addition to the large-scale ascent already expected in front of the MCV. Too much ascent, however, can suppress discrete supercell formation in favor of clusters and/or lines of storms that are often less significantly tornadic.
While the Zombie MCVs in both cases imparted favorable conditions for tornadic supercells, there was one ingredient that was relatively lacking in both cases and may have precluded any tornado forecasts in the days before each event. That ingredient was deep-layer wind shear (the change in winds from the surface to about 6km altitude). This ingredient is important for forecasting supercell development and since supercells are the most prolific producers significant tornadoes, one should be cautious issuing a tornado forecast without accounting for it. However, despite the modest shear, supercells were able to develop in both cases. This added another level of uncertainty to the forecasts.
Another subtle consistency between the two events was drone footage of damaged swaths of corn depicting where the tornadoes had tracked. Despite being a reflection of the latest in unmanned aeronautic imagery the pictures also highlight the coincidence of maturing corn and climatological peak MCS season in the Midwest. MCSs are an important contributor to the annual rainfall in the Corn Belt; without them, it’s likely wouldn’t associate the Midwest with corn and agriculture.
A recent field campaign was conducted in 2015 on the Great Plains called PECAN (Plains Elevated Convection at Night), the aim of which was to study the persistence of these nocturnal thunderstorm clusters their subsequent impacts. Below is a map displaying the results of one of the climatological studies done in preparation for PECAN. One focus of the study was to determine where summertime nocturnal thunderstorm activity was most common (areas of black, blue, and purple). The results show that nocturnal thunderstorms and MCSs are most common over the central Great Plains. One could extrapolate that if an MCS developed in this region and formed an MCV, the remnants of the latter would emerge somewhere over the eastern Corn Belt the next day.
A different visualization of this thunderstorm activity is depicted in the tweet below. The animation is from the National Center of Atmospheric Research (NCAR) and displays output from their WRF (Weather Research and Forecasting) Hydrological model. This model tracks rainfall as it flows into streams, creeks, and eventually into major river systems. This animation is from May 2015, and the development of MCSs over the High Plains and subsequent eastward tracks are apparent from the water entering the river systems. This rain is vital to much of the Corn Belt also provides both for shipping and irrigation downstream. In some years when thunderstorm activity is very frequent the Mississippi River will flood causing a threat to life and property as was the case in the Flood of 1993.
— NCAR & UCAR Science (@AtmosNews) October 17, 2016
This year’s events in Illinois and Indiana lends support to a broader connection between late-season Midwestern tornadoes and MCS activity. One would also expect to find similar events in recent years. A simple search of SPC products, yielded several similar tornadic events over the past few years. In reviewing these cases, I focused on set-ups with consistencies to those discussed above.
- The event had to produce at least two supercell tornado reports (per SPC).
- The MCV had to be remnant, or undead if you will. That means that its parent MCS had to originate the evening before the event. I also looked for MCVs that didn’t have a lot of convection or ascent surrounding them as we discussed how we wanted to differentiate between tornadoes that form within MCSs and those spawned by discrete supercells. All the events had explicit references to “remnant MCVs” in the SPC Day 1 products.
- I focused on events in the Midwest to highlight the connection between Plains MCSs and their remnant MCVs that bequeath tornado threats farther east.
June 7th, 2014 (Multiple reports of weak tornadoes near St. Louis, MO).
A large MCS developed over western Kansas in the early evening on June 6th. Archived radar imagery depicts the complex stretching from central Nebraska to northern Texas. The storms tracked east overnight before weakening markedly as they approached Missouri early the next morning. The remnant MCV emerged over northern Missouri by mid-morning on June 7th. Supercell storms formed on its southeast side by mid-day producing tornadoes in SE Missouri and SW Illinois by mid-afternoon. Event archive
May 1st, 2012 (Multiple reports of weak tornadoes in Illinois and Indiana)
An MCV developed over Missouri on the morning May 1st. It then tracked northeast into northern Illinois that afternoon. Supercell thunderstorms formed within its southeast quadrant by mid-afternoon and produced multiple tornado reports in eastern Illinois and western Indiana. Event Archive
June 27th, 2010 (Multiple tornado reports in Michigan and one in Wisconsin)
An MCS developed over Minnesota on the evening of June 26th and left wind and tornado reports over parts Minnesota and Iowa. The next morning the remnant MCV drifted eastward across Lake Michigan initiating storms which would produce tornadoes by mid-day in Wisconsin and Michigan, again on the southeast side of the feature. Event Archive
Below are the radar images for each day (6/7/2014, 5/1/2012, and 6/27/2010) as the tornadic storms began to initiate. I have overlaid the locations of the relevant remnant MCVs (red) and circled the areas (light blue) where the storms would track and subsequently produce tornado reports. Both events show that the southeast quadrants of the remnant vortex were largely clear of storms and precipitation just as the new supercells began to form.
Each of the events mentioned had a number of consistencies beyond the MCVs. The local environments were moist both in an absolute sense (dewpoint) and also in a relative sense (relative humidity). The influence of the latter is to lower the cloud base of the thunderstorms. The level that marks the base of a thunderstorm is called the lifted condensation level or LCL. Along with low-level wind shear, lower LCL heights have also been found to be contributing factor in the formation of significant tornadoes.
The reason for the higher moisture in these late-season scenarios is likely due to the fact that dewpoint temperatures (a measure of moisture in the sair) are often maximized during mid-summer in the Midwest. This maximum is aided by the maturing corn in the area that can locally increase dewpoint measures by 3-5 degrees Fahrenheit through evapotranspiration. Whether or not this effect was a significant contributor in any one event is left to further research.
Another consistency between these events was the aforementioned paucity of deep-layer wind shear. The MCVs often formed on the northwest side of an upper-level high and downstream of a low-amplitude upper-level trough over the northern Rockies. They also formed in areas of relatively weak upper-level flow, as is typical of mid-summer.
In examining recent cases, another feature was found contributing to localized tornado reports in Illinois and Indiana that shares a defining characteristic with zombie MCVs. Atmospheric features characterized by relative warmth within them, such as MCVs, are called warm-core systems. Tropical cyclones are also warm core systems but originate over the warm tropical oceans as opposed to the Great Plains. Tropical cyclones that form in the Atlantic Ocean in late summer and early fall often track northwestward toward the US before either impacting the coastline or, more commonly, turning east and missing it altogether. When these storms impact the coast, they ubiquitously weaken but can to track inland for hundreds of miles as they slowly spin down.
Once inland, these remnant lows behave similarly to remnant MCVs, and can initiate new storms. In some cases, they can even produce tornadoes. In two recent cases, a remnant tropical low tracked inland and produced tornadoes in the Midwest.
August 31st – September 1st, 2012 (Multiple tornado reports in Illinois)
Hurricane Isaac made landfall as a Category 1 Hurricane on August 28th in Louisiana before moving northward into the Midwest on August 31st. As it tracked northward, it produced several tornadoes, including a significant EF2 tornado in Arkansas. It then produced multiple EF0 (weak) tornadoes in central Illinois on both August 31st and September 1st.
June 19th, 2015 (Multiple tornado reports in Illinois)
Tropical storm Bill made landfall on the Texas coast on June 16th and continued to track north and then east until it reached southern Illinois on June 19th. What makes this system particularly notable was how it retained warm-core characteristics for nearly three days while over land. This has been theorized to be due in part to the Brown Ocean Effect, in which moist surfaces are able to continually resupply the system with moisture as a substitute to the warm ocean water. The effects of Bill’s remnants in southern Illinois, including tornadoes, was well cataloged by storm chaser Dan Robinson of stormhighway.com
The final event I will discuss is also of tropical origin and also produced tornadoes in Indiana in August 2016. On August 15th, just nine days before the August 24th outbreak, seven tornadoes struck Indiana in what was seen as a notable event itself if only for its late season occurrence. The system that produced these storms had origins farther south and while it wasn’t a remnant tropical cyclone, it did possess warm-core characteristics.
Starting August 9th, flooding rains began to impact Louisiana as the result of an upper-level tropical low. This disturbance then parked itself over the state for the next 72 hours produced over 20 inches of rain in parts of central Louisiana. The excess rains devastated the region in what might be the worst US national disaster since Hurricane Sandy.
The continual rainfall and release of latent heat formed a warm-core disturbance that resembled an MCV or remnant tropical cyclone. The feature then moved northward as upper-level winds finally steered the system out of Louisiana on August 14th. By the next morning, the feature was located over central Illinois, where augmented the tornado threat to its east in Indiana. Again, the SPC did not note a tornado threat in its forecast the day before but included it the morning of the event as it was clear that the disturbance was going to be in a favorable position to enhance low-level shear in Indiana. The set-up was even complete with a warm front as was present in nearly each case analyzed above.
This event allows us to come full circle with drone imagery as a drone video captured one of the day’s tornadoes as it was in progress over the cornfields of Indiana.
The final similarities between the events, including tropical cases, pertain to the structure of the storms and tornadoes themselves. While a few of the scenarios were able to produce tornadoes categorized as significant, the majority of the tornadoes were weak and resulted in little damage. Even when stronger tornadoes occurred, they never exceeded EF3 strength.
Another common characteristic of these events was a lack of reports of other severe weather types — hail and wind. One reason for this appears to be the tendency for mini, or low-topped, supercells to form during these events. Mini supercells are characterized by both relatively low (30,000 feet) cloud tops, a smaller spatial extent, and a lack of hail. The combination of ingredients that typically support these storms are lower instability and lower equilibrium levels (the height of the cloud tops). Each of the cases investigated had a combination of only modest instability (CAPE around 1000 J/kg) and lower equilibrium level heights. These factors are related both to one another and to their associated warm-core environments. Warmer air aloft suppresses hail formation and reduces instability. Fittingly, tropical storms are also noted by their lack of hail and lightning. This provides yet another challenge to forecasting warm-core set ups as we have established yet another precluding factor for widespread severe storms in the proximity of such vortices: modest instability.
The challenge then becomes issuing a tornado forecast when the major ingredients one looks to when considering severe weather, atmospheric instability and deep-layer wind shear, are only modest around persisting MCVs and other warm-core vortices. It appears then that the ingredients that are enhanced around the feature can compensate for those reduced. It is also interesting that the enhanced ingredients, low-level wind shear and lower LCL heights, are the two ingredients found to be the most predictive for significant tornadoes and are also theorized to possibly be the most important to tornadogenesis.
The lack of severe weather reports, besides tornadoes, is consistent with each event mentioned above. A visualization of this idea is the distribution of storm reports from the August 24th Indiana outbreak shown below. Portions of Indiana and Ohio are littered with red dots (marking tornado reports) but very few other severe reports (green: hail and blue: wind) are nearby. In the Corn Belt, these features can extend the tornado season longer into summer than would otherwise be expected given the poleward retreat of the primary midlatitude jet stream. It may be that a significant portion (possibly greater than 15%) of late-season tornadoes in this area are associated with remnant warm-core features.
Tornadoes remain a sometimes elusive phenomenon, for forecasters, storm chasers, and residents alike. To the residents of the Corn Belt, tornadoes are not seen as uncommon. Tornado sirens are even treated as routine in many areas. However, the likelihood of tornadic storms is generally known to decrease as the summer days pass. The windy spring afternoons are replaced by a humid lull that often overspreads the Corn Belt by mid July. However, this summer swelter from the increased moisture can provide the hint of storms to come. In rare occurrences, complexes of nighttime storms that originate several states away can manifest a spin that imparts the following day with a tornado threat otherwise overlooked. In my own failed efforts to merge poetry and weather, I leave you with a quote by Illinoisan author David Foster Wallace.
But in the odd central pocket that is Champaign-Urbana, Rantoul, Philo, Mahomet-Seymour, Mattoon, and Tolono, Midwestern life is informed and deformed by wind. To the west, between us and the Rockies, there is basically nothing tall, and weird zephyrs and stirs join breezes and gusts and thermals and downdrafts and whatever out over Nebraska and Kansas, and move east like streams into rivers and jets and military fronts that gather like avalanches and roar in reverse down pioneer ox trails toward our own unsheltered asses. — David Foster Wallace (1991)
We should then be fortunate this year that as Halloween comes around yet again, we are haunted only by ghosts, ghouls, and goblins and not the forecaster’s nightmare that is the Zombie MCV.