StormNet Event Summary
- Contrasting with some of the other case studies, this event was NOT synoptically evident several days out.
- Beyond +9d, the StormNet forecasts were generally inconclusive with probabilities below 1% across CONUS.
- At +7d, StormNet really started honing in on the mid-south as a potential hotspot for tornadoes.
- By +5d, the forecast is essentially 'solved', as StormNet has honed in on the affected corridor with few spatial changes thereafter.
- The Storm Prediction Center missed several tornadoes in the mid-Atlantic which StormNet caught at low probability levels.
On Friday May 16, 2025, a tornado outbreak impacted much of the Ohio River Valley region. This tornado outbreak led to 30 fatalities and at least 165 injuries, along with $5.9 billion in damages.
This event, alone, stands as one of the most costly severe weather events of this decade. This event is one of the events that we closely evaluate new versions of StormNet against. Not only due to the number of tornadoes, but primarily due to the heavy impacts of this event on so many lives.
In this first case study, we'll go through the forecasts leading up to this event from 9 days out to 3 days out. The aim of these case studies is to help meteorologists, storm chasers, emergency managers, and the general public understand StormNet forecasts, what they mean, and how they evolve with time leading up to an outbreak of this magnitude.
Tornado tracks from the NWS on May 16, 2025. Color coded by intensity. Damage Assessment Toolkit.
This tornado outbreak consisted of several EF2+ tornadoes across SE Missouri into the Ohio Valley region, with additional weak tornadoes in the Mid-Atlantic.
This particular case wasn't that obvious from several days out -- the upper air pattern was mostly zonal and there was considerable model uncertainty.
9 Days Before the Outbreak
We'll start on May 8th, 2025. At this point, there was really no discussion of a potential impending tornado outbreak. Models were inconsistent from run-to-run, and from model-to-model. This is to be expected at this range. StormNet was already honing in on a tornado potential, however.
These probabilities may seem quite low at first glance, only around 1% to 10%. From our climatological analysis, we can see that probabilities at these levels within 216 hours are somewhat rare, occurring around 75-100 days per year.
The Storm Prediction Center, who are the best in the business when it comes to severe weather prediction, do not produce forecasts beyond 8 days, so we won't compare them at this range.
7 Days Before the Outbreak
Next, we'll move on to May 10th - 7 days before the outbreak. The Storm Prediction Center's day 7 outlook had no mention of a tornado outbreak, stating that 'predictability is too low'.
And yet, from the same morning, the 7-day StormNet tornado forecast is starting to loudly broadcast that a tornado outbreak may be possible. Probabilities have been steadily rising and now exceed 10-15%, a feat that only occurs a few times per year at this lead-time.
These probability levels at day 7 raise serious alarm that a rare tornadic event may be on the horizon.
5 Days Before the Outbreak
By May 12th, 5 days before the outbreak, the SPC has highlighted a small portion of the Mid-South in a 15% ('Slight' risk) for possible severe storms. In their outlook, they state that "This synoptic regime suggests that one or more organized clusters/linear segments may propagate along the boundary and into parts of the mid-MS Valley, lower OH Valley, and potentially into the TN Valley."
The Day 5 StormNet tornado forecast was quite a bit more robust -- an incredible rare >50% tornado threat 5 days out. StormNet highlights a large expanse from SE Missouri through Kentucky.
These are very rare probability levels for 5 days out, only occurring a few times per year in our validation study. This is essentially StormNet screaming from the rooftops that a tornado outbreak is highly likely 5 days out for a specific region of the country.
3 Days Before the Outbreak
May 14th, 2025: only 3 days before a major tornado outbreak. The SPC has issued a day 3 'slight' risk (level 2/5) for severe storms, stating: "Some combination of scattered supercells and organized clusters, including the potential for a long-lived MCS as hinted by some deterministic solutions, appears likely somewhere across the Ozarks/OH Valley region Friday afternoon and evening."
StormNet at this point has consistently increased tornado probabilities over the last several days, now with probabilities exceeding 60%! These probabilities are very rare for 3 days before an event.
The SPC is calling for a run-of-the-mill severe weather event, with considerable uncertainty. StormNet, on the other hand, is calling for a rare tornado outbreak with considerable certainty.2 Days Before the Outbreak
Now, 48-hours out from the outbreak, confidence is increasing across the board that an impactful tornado event may occur.
The 48hr HRRR forecast indicates a line of cells extending from the Ohio River area through central Arkansas and Tennessee.
At this point, it is becoming more probable on conventional models that there may be supercells, and an environment potentially conducive for tornadoes.
The Day 2 SPC forecast had an enhanced risk (level 3/5) of severe storms from the Mid South through the Ohio Valley region, including a 10% tornado risk!
The SPC notes in their forecast that intense supercells followed by upscale growth and tornadoes will be possible:
During the late afternoon into evening, this may include a few intense supercells posing a risk for large hail and tornadoes. During the evening into the overnight hours, this probably will include an upscale growing and organizing cluster posing a risk for strong, damaging winds and perhaps a continuing risk for tornadoes.
The StormNet tornado forecast at 48hrs out remains extraordinarily high; a steady increase from day 9 all the way through day 2, with tornado probabilities now exceeding a rare 75%!
Our analysis shows that probabilities this high only occur around 5-10 times per year, making this roughly equivalent to an SPC day 2 high-end moderate risk (level 4/5).
Inside of 48 hours
Inside 48 hours, the forecast breakdown becomes a bit more complex. We start focusing on specific corridors, specific towns, and more narrow timeframes. StormNet forecasts in the short-range output at an hourly interval and update every 2 minutes. As a result, we'll cover short-to-intermediate range forecasts in another case study article!
Visit StormNet.ai and download the StormNet app (Android + iOS) to get started today.
Interested in the StormNet API? Tap here.
Andrew Brady