September Didn't Stand Down: 25 Tornado Warnings, 29 Reports, and 10 Surveyed Paths in Eleven Days

September 16, 2026
September Didn't Stand Down: 25 Tornado Warnings, 29 Reports, and 10 Surveyed Paths in Eleven Days

The 2026 shoulder season, told the way tornado history actually arrives — warnings first, surveys later

September is supposed to be the exhale. The Plains quiet down, the jet stream hasn't yet returned for fall's second season, and tornado climatology's long tail begins. September 2026 didn't cooperate. Between September 4 and 14, the National Weather Service issued 25 tornado warnings, observers filed 29 tornado local storm reports — on seven of the window's nine storm-active days — and in the days that followed, survey teams logged ten tornado damage paths across nine forecast-office areas, from the Carolinas and the Gulf Coast through Illinois, Pennsylvania, and Minnesota to New Mexico. A parallel cluster put tornadoes on the ground in southern Manitoba and southwestern Ontario. None of it made national news. All of it is in the event record.

This is the anatomy of a shoulder-season tornado run — dispersed, small, easy to miss, and exactly the pattern that accumulates into portfolio risk without a headline.

The month opened loud

The run didn't start from silence. September's first week brought a multiday severe-weather event across the Great Lakes, Ohio Valley, and Mid-Atlantic — more than 500,000 customers without power at its peak, roughly 280,000 in Michigan alone, with damaging winds taking down trees and lines from Michigan through Pennsylvania before the pattern slid into the Carolinas over Labor Day weekend. Reported tornado activity in that opening act included a confirmed tornado near Somerset, New York — a waterspout that came ashore northeast of Buffalo — and, per widespread reporting, Detroit's first tornado since 1997 on September 3.

By September 4, as that system faded east, the warning run began.

Eleven days, 25 warnings, tornado reports on seven of them

The complete National Weather Service record for the window — warning polygons from the IEM archive, local storm reports, and SPC's daily tallies — shows 25 tornado warnings and 29 tornado LSRs between September 4 and 14, with tornado reports filed on seven of the nine storm-active days and the busiest single day at the very end: eight tornado reports on September 12. Around the tornado activity, the same window logged 989 severe-thunderstorm warnings, 252 hail reports, and 1,783 damaging-wind reports — a full severe-weather machine running at shoulder-season volume.

A tornado warning is a short-fuse product — typically 25 to 40 minutes, drawn around detected rotation or a report. Among the 25, the warnings our hazard feed elevated to event records show how small these polygons ran, with footprint populations attached:

  • September 4 — Clark County, Wisconsin (~1,500 people in the polygon) and southeastern Virginia (~7,500), warned within a minute of each other in separate systems
  • Night of September 7 (local) — a three-warning cluster across the Dakotas: Sheridan County, ND, then Charles Mix and Bon Homme counties, SD (~250 to ~2,900 people each)
  • Night of September 9 (local) — the Ohio Valley round: Licking County, Ohio (~5,300), Potter County, Pennsylvania (~900), and north-central Pennsylvania (~1,300)
  • Night of September 11 (local) — west-central Minnesota (~1,100)

Small polygons, small populations, no single day worth a national story. That's the shoulder-season signature: the atmosphere doesn't produce an outbreak, it produces retail-scale rotation, one county at a time — 25 times in eleven days.

What the surveys found

Tornado history is written twice — once by radar in real time, once by survey teams in the days after. For this window, the NWS Damage Assessment Toolkit now holds ten surveyed tornado paths and 69 damage-survey points, spread across nine forecast-office areas: the North Carolina coast (two paths), the New Orleans area, Chicago's and central Illinois's areas, western and central Pennsylvania, New Mexico, the Omaha office's area, and Minnesota. Among the touchdowns, the ones local reporting attached names and dates to:

  • McKean County, Pennsylvania — a tornado touched down near Bradford Airport on September 9, from the same night's storms that drew the Potter County and north-central Pennsylvania warnings
  • Guernsey County, Ohio — a tornado confirmed from the September 9 storms that also warned Licking County
  • Stevens County (Artichoke area), Minnesota — a confirmed tornado, with storm video widely shared, from the west-central Minnesota activity
  • Surry County, Virginia — an EF-0 confirmed along Route 10 on Saturday, September 12

North of the border, the same continental pattern was busier still: two tornadoes touched down in southern Manitoba — including one in a Winnipeg neighborhood — with one injury reported, and southwestern Ontario's survey teams confirmed a cluster including an EF-1 near Monkton, with researchers ultimately counting a sixth tornado from the early-September Ontario storms.

Every US touchdown in the run closed with zero reported casualties. The warnings, small as their polygons were, did their job.

Open map full screen ↗

The map opens on the tornado story — warning polygons, report dots, and the ten surveyed damage paths, each marked with a ringed dot (the paths run 0.3 to 2.8 miles, so use the built-in "zoom to a surveyed path" links). Then press Everything on: 989 severe-thunderstorm polygons, 1,244 radar-measured hail swaths, 243 hail reports, and 1,783 damaging-wind reports fill in the machine the tornadoes were embedded in — that view is the eleven days.

Was this much severe weather normal for September? No.

That 989 severe-thunderstorm warning count deserves its own context, because it's the part of this window that genuinely departed from climatology. Running the identical query — NWS warning polygons for September 4–14, from the Iowa Environmental Mesonet archive — across the last seven years:

  • 2020 — 9 tornado warnings · 121 severe-thunderstorm warnings
  • 2021 — 15 · 372
  • 2022 — 18 · 227
  • 2023 — 8 · 1,054
  • 2024 — 22 · 116
  • 2025 — 55 · 630
  • 202625 · 989

The 2026 window ran at roughly three times the seven-year median severe-thunderstorm warning volume — second only to 2023's remarkable 1,054. The tornado-warning count sits above the period's typical teens but below 2025's outlier 55. And the pattern in that list is worth sitting with: three of the last four Septembers (2023, 2025, 2026) ran hot in the same eleven-day window. Whether that's variance or a shifting shoulder season is a question this seven-year sample can't settle — but it's exactly the kind of question that becomes answerable when every warning, report, and survey lives in one queryable archive.

The Pennsylvania backdrop: a record already broken

Here's the context that makes September's Pennsylvania touchdowns more than a footnote: western Pennsylvania entered September having already broken its all-time annual tornado record. The National Weather Service office in Pittsburgh confirmed its 30th tornado of 2026 in mid-August — an EF-0 at Orchard Hills in Armstrong County — passing the previous record of 29 set just two years earlier, in 2024. Tornado activity in the region normally peaks in June and July and fades through August; 2026 kept producing anyway, and September kept the streak alive: the survey record from this run includes a tornado path inside the Pittsburgh office's own warning area, on top of the September 9 McKean County touchdown next door in the State College office's area — a year that has refused to follow its own climatology.

Two record years in three seasons is the kind of signal a single event never shows — it only appears when events are kept as structured history rather than headlines.

Why a quiet-month run matters

A September like this is easy to shrug off — no outbreak day, no EF-3s, no fatalities. But it's precisely the pattern that stress-tests how organizations consume severe-weather intelligence:

  • The alert tier and the observed tier diverge by days. On September 10, the truthful answer to "did a tornado hit Ohio last night?" was "a warning was issued; surveys pending." The confirmation arrived later — the same watch–warning–confirmed lag we walked through for Omaha, playing out in four states at once.
  • Dispersion defeats attention. Seven states in eleven days means no local story becomes a national one — but a portfolio with exposure across the Midwest and Mid-Atlantic experienced all of it, cumulatively.
  • Small polygons still contain assets. A ~1,000-person warning footprint is invisible at the national scale and decisive if your facility is inside it. Footprint-level intersection — which locations sat inside which polygon, when — is the only view at which this month was legible.

The season's long tail is still ahead

Climatologically, the year's second tornado season — the fall transition, October into November — hasn't arrived yet. 2026 has already delivered a record year in western Pennsylvania, a first-since-1997 tornado in Detroit, and a September that produced warnings at a summer cadence. The honest read isn't a prediction; it's a posture: the alert tier will keep firing through fall, the surveys will keep arriving days late, and the organizations that fare best are the ones whose intelligence connects both tiers to their own map.

Track any US city's live tornado status and history at explore.disasteraware.com — or start from the full hazard picture for your locations.

Every warning, every survey, every footprint — one record

DisasterAWARE Historical Event Intelligence connects alerts, observations, and exposure into event-level records — the dataset behind this analysis.

START FREE TRIAL →TALK TO AN EXPERT →

Explore the historical record

Decades of reconstructed, verified catastrophes — browse the Historical Event Intelligence overview, or request sample data for your own portfolio.

HISTORICAL EVENT INTELLIGENCE →REQUEST DATA →

Frequently Asked Questions

How many tornadoes hit the United States in early September 2026? Between September 4 and 14, the National Weather Service issued 25 tornado warnings, observers filed 29 tornado local storm reports, and NWS surveys logged ten tornado damage paths across nine forecast-office areas — from the Carolinas and Gulf Coast to New Mexico. Named confirmations in local reporting included McKean County, Pennsylvania and Guernsey County, Ohio (both September 9), Stevens County, Minnesota, and an EF-0 in Surry County, Virginia (September 12). The month's first week separately produced a confirmed tornado near Somerset, New York and reported Detroit's first tornado since 1997. Zero casualties were reported from the US touchdowns.

Is September usually an active tornado month? No — September sits in the shoulder between the summer peak and the fall secondary season. Tornado activity in most of the Midwest and Northeast normally peaks in June and July. That's what made 2026's September run notable: warnings at a near-summer cadence, spread across seven states, in a month that usually exhales.

Why do tornado confirmations come days after the storms? Warnings are issued in real time from radar rotation and spotter reports. Whether a tornado actually touched down — and its EF rating — is determined by National Weather Service ground surveys of the damage path, which take days and sometimes find additional tornadoes on a second pass. The alert tier and the observed tier of tornado history are written on different clocks.

What tornado record did Pennsylvania break in 2026? The National Weather Service Pittsburgh office, covering western Pennsylvania and adjacent parts of Ohio and West Virginia, confirmed its 30th tornado of 2026 in mid-August — an EF-0 in Armstrong County — surpassing the office's previous annual record of 29, set in 2024. Two record years within three seasons.

Did the September 2026 storms hit Canada too? Yes. The same early-September pattern produced a confirmed cluster in southern Manitoba — two touchdowns including one in a Winnipeg neighborhood, with one reported injury — and southwestern Ontario, where surveys confirmed multiple tornadoes including an EF-1 near Monkton, with researchers ultimately counting six from the early-September storms.

Related Reading

Get the Weekly Hazard Brief
Curated global hazard intelligence — earthquakes, severe weather, conflict, and cascading risk — delivered to your inbox each week. Free.

More Reading Material