Thoughts on the Midwest Tornado Outbreak

November isn’t generally known for tornado outbreaks in the U.S., but mid-way through the month a vicious and deadly swathe of tornados swarmed through the Midwest States of Illinois, Indiana, Kentucky, Missouri, Ohio and Michigan leaving widespread damage and destruction in its wake.

The Storm Prediction Center recorded 85 tornado reports with 455 reports or high winds and 32 reports of hail, making a total of 572 reports of severe weather.

Within this log, two preliminary EF-4 tornados (in Washington, IL, and New Minden, IL), and an EF-3 were included. This outbreak of severe weather also took the lives of at least eight people.

General Meteorological Situation

November 16th saw a weak low pressure system over the central U.S. It deepened over the northern states around the Great Lakes, and engaged with a strong jet stream, which encouraged cyclonic development in the presence of an unseasonably warm and humid air mass.

Well-defined mid and low level jets were also present which aligned to produce favourable conditions for Tornado development.

The Storm prediction Center issued a moderate warning of severe weather on the 17th, and as the low deepened and confidence in a severe convective event increased, the threat was upgraded to high risk for the Ohio Valley and the southern portion of the Great Lakes.

A squall line developed during the afternoon producing widespread wind damage which tracked eastwards before being pushed along by a cold front overnight and into Monday the 18th.

Numerous factors came together to require the severe warning together to encourage tornadic development, namely:

  • a well-organised and mature negatively-tilted system
  • the coincidence of favourable low and mid level winds with an outbreak of convective storms in the warmer and more humid air
  • the fact that supercells were expected to remain discreet well into the afternoon
  • very favourable low level wind shear profiles, i.e. significant speed and directional shear was present basically describing a twisting of the atmosphere


Tornados can happen at any time of year at any time of day or night, however the peak period for historical tornado reports is normally referred to as ‘Tornado Season’.

The peak period for the southern plains is from May to early June, but around the Gulf Coast is it nearer spring, and for the northern plains and upper Midwest the peak is June or July. It should be noted that there is a northward shift through the year from late Winter to mid-Summer before shifting southward again.



Tornado Alley

Tornado alley is a term used to describe a fairly broad swathe of the central U.S. where there is a relatively high occurrence of tornados. There are many versions of tornado alley based on different metrics for tornado frequency, but they roughly show similar regions.

These maps only show the most frequent areas, but tornadoes can happen anywhere that conditions allow, in fact violent tornadoes happen outside of tornado alley every year, and occur outside of the peak season.

This tornadic event therefore was not so unusual spatially, since tornados frequently occur in the affected areas of the Midwest. However, for such a widespread outbreak so late in the year, it is a fairly rare event from a meteorological perspective.

Tornado History Project

Using the Tornado History Project website which shows tornado data back to 1950, it appears that only 20 EF-4 tornados have formed in November according to the historical record dating back to 1950, and 15 of them have been reported since 1990.

It is also preliminarily the second most northerly EF-4 in the record. This kind of information, although intriguing, should be used with caution due to the poor quality of tornado reporting and will incorporate considerable uncertainty into actual meteorological factors rendering any direct comparison flawed.

Measuring return periods of tornado outbreaks reliably is notoriously difficult due to the sporadic nature of the observations and the relatively hard-to-measure small-scale processes and that form tornados.

When simulating tornados using computer models, these factors make them difficult for all but the highest resolution weather models to represent.

The most expensive and deadliest outbreak of severe weather this year was in the height of Tornado Season, in Moore, Oklahoma where an EF-5 touched down killing 23 people and causing around $2 billion in damage.

Tornados and Climate Change

The destructive force of a tornado, although devastating and uncompromising, is a secondary effect when looking at climate change. To shed light on the question of how tornado frequency will change we should first look at how a changing climate will influence thunderstorm activity.

Despite research into trying to elucidate a signal from climate models to represent the convective systems that spawn tornados (and other small scale processes like hail and lightning), to this date there has been limited success.

The main difficulty comes from the fact that tornados are one of the most extreme examples of weather – they manifest in a matter of minutes, ever-changing and tempestuous, as they leave footprints often only a few miles wide.

This small spatial and temporal scale is achievable with a degree of accuracy when looking at weather models which forecast just a few days ahead, but this is the opposite end of the spatial and temporal spectrum examined by climate models.

Climate models look further into the future and generally cover the whole globe, and therefore are good at indicating broad characteristics.

They could perhaps find favourable conditions for convective systems to develop, and so give us clues from which to derive a link to potential for deep convection and thunderstorms, and therefore finally deduce an influence on tornado frequency – but suffice it to say, the link is tenuous.

Model Resolution Improving

However, model resolution is improving and cutting edge models can now start to finely resolve the structure of hurricanes and winter storms realistically, and with research into high resolution climate modelling, we will begin to see a clearer picture of how potential future climates will affect our weather and perhaps start to understand the relationships to extremes of weather, like tornados.

On the broad scale in time and space that is required for looking at climate change, our physical understanding of the atmosphere gives mixed messages. Some indicators may increase instability (good for tornados) while others may decrease low level shear (bad for tornados).

If we suggest that severe tornados seem to be becoming more frequent, or at least there is a shift further north in the Midwest of the U.S., it would be very difficult to find the evidence to back our suggestions up, and therefore there is limited impact that such an assumption could have on policy and decision makers. Much more research needs to be conducted into climate change impacts with respect to severe weather events.

As mentioned before, historical tornado records are also very poor and don’t even exist for most parts of the world, and so it is also very difficult to look back at past data to get an idea about how tornado trends have changed already.

Therefore, it is very difficult to answer the question of how climate change will influence tornado frequency and severity. It is for this reason that we are reliant on models.

It will likely be a few years yet before the factors that influence tornados can be reliably represented in climate models, but until that time, the best we can do is keep an eye on the forecast, listen to weather warnings and make sure we have a plan of action to make our homes, businesses and communities as resilient as they can be, if we live in an area at risk from tornados.

About Geoffrey Saville

Geoffrey Saville is a member of Willis Towers Watson's Analytics Technology Team, having joined the company in 2013…
Categories: Analytics, Reinsurance | Tags: , , ,

Leave a Reply

Your email address will not be published. Required fields are marked *