Tornado Season Isn't What It Used To Be: December is hot, June is not, there is no fall season, etc...
- Andrew Pritchard
- Oct 2
- 7 min read
Updated: Oct 3
I’m an operational meteorologist who spends most of my time forecasting high-impact weather and helping people make safety and logistics decisions. I'm not of the academia world, I don’t claim to be a pure atmospheric researcher or data scientist—but after years of forecasting (and chasing) big storms, I’ve noticed some trends I wanted to test with a simple data exploration.
Those trends have to do with severe weather season, or more specifically tornado season in the United States. As a meteorologist and storm chaser, lately, it feels like tornado season itself is shifting. Specifically:
The most active and violent portion of the annual tornado season is shifting earlier in the spring (March & April)
The primary tornado season is ending earlier and more abruptly (late May and June)
Winter is becoming more active (more outbreaks in December and January)
There is not really a fall season
The methodology here was pretty simple - I just wrote a basic python script to scrape the data from https://www.ncei.noaa.gov/access/monitoring/tornadoes/ and spit out some trends based on monthly tornado numbers in the United States from 1990 to 2025 along with a trend line showing the increase/decrease in tornadoes each month during that 35 year period. My goal was to take the data and lay it out in an easy to digest visual.
Of course, there are limitations. Monthly tornado counts can be heavily skewed by single large outbreaks. A quieter month with one blockbuster day might look artificially “active.” A more robust research approach—like the “Shifting Tornado Alley” studies that use tornado-favorable environments instead of raw counts—would smooth out these extremes. Read some of this research (much more thorough than mine!) here: https://www.nature.com/articles/s41612-018-0048-2 and here https://journals.ametsoc.org/view/journals/apme/63/6/JAMC-D-23-0143.1.xml#bib26
All that being said, the trends that I uncovered were actually quite loud and back up a lot of my recent suspicions without a whole lot of "eh that was just one big day" hemming and hawing. For example, I think the back to back tornado outbreaks of December 10th and December 15th are in fact, part of a bigger trend instead of a couple outliers skewing the data.
So, check it out:
Looking at monthly trends verbatim, some of those feelings start being validated. The biggest changes?
June has become a LOT quieter.
This is one of the more undeniable monthly trends. From 1990 to 2010 the month of June produced 300+ tornado six times, but that has not happened once since 2010. Always a rather volatile month with big year to year swings, big synoptic tornado outbreaks are becoming a rarity while a quiet June with a tornado count between 100 to 150 has become much more common. This makes logical sense when you consider an analysis of mid-latitude jet stream trends from 1990 to 2025 shows an increasingly weak, and northward displaced jet stream overall by June. A hotter air mass with a stronger cap (EML) and weaker flow is going to result in fewer tornadoes.
2025 produced some notably photogenic tornadoes on the Northern Plains this year, but these were primarily localized mesoscale events not featuring large upper-level disturbances. Point being - June seems to more often feature a weakened jet stream that is retreating north into Canada earlier in the season which means less in the way of geographically widespread, high-end severe weather events and a higher reliance on locally favorable, mesoscale days producing a photogenic tornado or two.
The other big feeling that had a pretty loud confirmation was the idea that December tornadoes / December tornado outbreaks are becoming more common. Over the last five years, my body is just starting to get this feeling that with the seasonal return of fast jet stream flow over the CONUS in December comes a renewed potential for organized severe weather events and high volume tornado days.
From 1990 to 2011 the month of December only produced 50+ tornadoes one time in an anomalous 2002. Since 2013 the month of December has surpassed 50 tornadoes 7 times. That's a jump from 1-in-22 years to 7-in-13 years. Put another way: for decades, December almost never produced 50+ tornadoes in the U.S. Now, it happens more than half the time.
This makes logical sense, too. In a warming climate with more moisture, you're seeing seasonally intense jet stream disturbances that once passed over the CONUS quietly, are now tapping in to a little instability in the warm sector and with a highly shear storm system, that's all you need!
The December 10-11, 2021 tornado outbreak featured a highly unusual, deep warm sector that advected north into the Midwest on several days of return flow from a strong low-level jet. However, a few days later a classic example of this "big flow + just enough juice" wintertime setup occurred on December 15th as a record number of tornadoes raced across the Upper Midwest. This was a highly unusual tornado outbreak, but has the framework of something I think we'll see more often in coming years/decades as higher end severe weather days become more common during the holiday season in the U.S.
2024 featured big December tornado outbreaks with 100+ tornadoes impacting the Southern U.S. from December 28-29. In the midst of another fall drought in the Central U.S. I'll be very curious to see if early winter brings a return of jet stream flow and organized severe storms.
January largely mirrors December, a month that prior to the 2010s almost never produced 50 tornadoes (except for an exceptional 1999) is now quite variable - some years it's quiet, other years we're bringing just enough instability north into the CONUS to fuel organized severe weather events. January has seen the fifth biggest increase in tornado activity since 1990 in the U.S.
Along with the surge in wintertime tornadoes, we're seeing high volume tornado outbreaks and tornado sequences earlier in the season with big moderate/high risk days featuring significant tornado outbreaks becoming more common in March and April.
May has seen an annual increase, but there's not much to see as far as trends since 1990. April and May will probably always be the two most consistently active months for tornadoes in the U.S. as it provides the best combination of strong jet stream flow and availability of warm, moist air to fuel thunderstorms.
We've already looked at the rapid decline in June tornadoes, and while the trend is negative in July as well I think there's some potential for July to get weird.
A weakened, northward displaced jet stream means less in the way of ejecting storm systems and organized severe weather events over the CONUS, sure. But, as we examine 500 mb height increases over time we've seen a bigger increase over the western U.S. compared to the central and eastern U.S.
In an oversimplified way, this means the month of July is more frequently featuring a persistent upper-level ridge over the western U.S. compared to the central and eastern U.S. allowing portions of the Plains, Great Lakes, and Midwest to benefit from "northwest flow". Less reliable than the prime March, April, and May season driven by organized jet stream flow, ridge-riding storm clusters and remnant MCV can at times provide periods of organized severe weather over this corridor of the Northern Plains to the Midwest. This corridor has been dubbed "derecho alley" by prior research, read some of it here: https://journals.ametsoc.org/view/journals/bams/104/10/BAMS-D-22-0217.1.xml
As our ability to detect and confirm tornadoes has increased, and we theoretically see more of these derecho-type events in a warming climate with more available fuel, another dumb hypothesis of mine is that we'll see an increase in high volume tornado outbreaks from derechos, similar to August 10, 2020 and July 15, 2024. July probably stays a big wildcard month, often quiet, but occasionally allowing bursts of tornado activity across the Northern Plains and Midwest during periods of favorable northwest flow.
Finally, there is no fall season. I've fallen (pun intended!) for this misconception since my earlier years in the 2000s. As the jet stream migrates south and we see flow return, we should see severe storms and tornadoes return, right?? In theory, yes! More often, the jet stream has taken the scenic route in October and November with the CONUS seeing continued weakened, displaced flow with drought becoming a more common autumn hazard.
I combined October and November into one chart and the line is nearly flat, a slight decline. Tropical cyclones produced high volume tornado outbreaks over the southeast U.S. in 2002 and 2004, and there have certainly been notable synoptic tornado outbreaks from the Plains into the Midwest over the last 35 years. November 2005 was very active in the Midwest, November 2011 produced in the Southern Plains, the November 17th, 2013 tornado outbreak in the Midwest was extremely violent. These events are anomalies though, and there's no data to suggest the fall season is getting more or less active overtime.
In a lot of ways tornado season may no longer be a season at all, but a moving target that increasingly stretches across the calendar. My general vibe as a meteorologist and storm chaser living in the Midwest is that fall is rest time. A time to enjoy the mild autumn air and head out to the woods for fall colors... because with wintertime comes the beginning of tornado season in the U.S. as December and January tornado days become more common, ushering us into that March - April run of atmospheric violence, followed by consistently consistent May, and the increasingly inconsistent June and July.
