Winters are dropping less snow across most of the country, and school closures haven't followed that decline. That's not a hunch or a nostalgia trip — long-term weather station records back up the first half of the claim, and district calendar histories back up the second. The gap between the two is the real story, and it comes down to what actually triggers a closure decision, which was never snowfall totals alone.
The Claim Worth Testing
A recurring observation from one North Carolina county is that its wintry precipitation days — days with any measurable snow, sleet, or freezing rain — fell by roughly a third over six decades, while the number of school closures charged to weather stayed essentially flat. That's a local finding, not a national one, and it hasn't been checked against other regions using the same method. Treated as a hypothesis rather than a settled fact, it's worth testing against what long-run station data and district records actually show elsewhere.
The honest answer sits in two parts: the snowfall decline is real and well documented at a national scale, but the closure side of the equation depends on local calendar archives that aren't centralized anywhere. No single public database lets you pull "closures by district by year" the way you can pull daily precipitation from a weather station. That mismatch in data availability is itself part of why this kind of local finding is hard to independently confirm — and why it's worth being precise about what's verified and what isn't.
What Long-Term Station Records Actually Show
The clearest national-scale look at this comes from an analysis of long-term snowfall observations at 2,041 U.S. monitoring stations, comparing 1970 to 2023. Roughly two-thirds of those stations — 1,301 of them — recorded less snowfall by the end of that period than at the start. The remaining third, concentrated in the lee of the Great Lakes and areas that get lake-effect snow, actually saw increases.
That regional split matters. A county in the Northeast or Mid-Atlantic, where winter precipitation has been shifting from snow to rain as the snow-rain transition line moves north, is a very different case than a county downwind of Lake Erie, where warmer lake water can mean more lake-effect snow, not less. Testing the "closures didn't shrink" hypothesis somewhere like western New York would need to account for the fact that snowfall there may not have declined at all.
| Region | Long-term snowfall trend (1970–2023) | Primary driver |
|---|---|---|
| Northeast / Mid-Atlantic | Declining; more precipitation falling as rain | Rising winter temperatures narrowing the snow window |
| Southwest | Declining; earlier snowmelt | Warming, shrinking mountain snowpack |
| Lee of the Great Lakes | Increasing in some locations | Lake-effect snow intensifying with warmer lake water |
| Interior South / mid-Atlantic coast | Declining since the 1920s in federal station analyses | Long-run warming trend |
Winter is the fastest-warming season in most of the U.S., which is also why the total count of nights below freezing has been dropping since 1970 across the majority of stations tracked. Less consistent cold means more winter precipitation arrives as rain, sleet, or a rain-snow mix rather than accumulating snow — which changes what a "snow day" storm even looks like on paper, separate from whether roads are still dangerous.
Why Closures Didn't Shrink With the Snow
If snowfall totals were the only input into a closure decision, you'd expect closures to track the decline. They mostly haven't, for reasons that have nothing to do with how much snow actually fell.
Road conditions, not snow depth, drive the call. Superintendents and transportation directors are making a bus-safety decision, not a snow-measurement decision. A quarter inch of freezing rain on untreated roads is more dangerous to a bus fleet than four inches of dry powder on plowed streets. As winter precipitation shifts from snow toward rain, sleet, and freezing rain in warming regions, the roads can get more hazardous even while the snowfall total on the gauge goes down.
Liability posture has hardened, not loosened. Districts that once tolerated marginal conditions face more scrutiny after any weather-related incident, and risk-averse administrators tend to keep the same closure threshold even as the underlying storms get milder on average, because the downside of guessing wrong is asymmetric — a canceled day costs make-up time, but a bus crash costs far more.
Remote learning changed the accounting, not the underlying trigger. Since the pandemic, a large share of districts converted some snow closures into remote-instruction days rather than eliminating closures. A 2020 survey of principals and district officials found that around 70% had converted or were considering converting snow days to remote learning days. That's a change in what the day counts as on the school calendar — it is not necessarily a change in the threshold that gets a school building closed to buses and in-person attendance in the first place.
Parents don't want the trade fully eliminated. A 2026 survey of over a thousand parents found that 74% consider snow days an important part of childhood, and majorities said they wouldn't trade snow days for a longer summer or for using them to cover other weather closures. That kind of public sentiment pushes back against districts that try to close the loophole entirely, which keeps the "closure" option alive even where remote instruction is technically possible.
The Remote-Learning Wrinkle, State by State
Policy on this varies enormously and is still shifting year to year, which is part of why a single county's closure count is hard to generalize from.
| State / district | Approach | Notes |
|---|---|---|
| Kentucky | Non-Traditional Instruction (NTI) days, capped at 10 per year, count as full school days | Districts that burn through the cap by February are forced back to traditional closures and make-up days |
| Virginia | Proposed legislation would require virtual learning instead of closures | Not yet a mandate statewide as of the most recent session |
| Washington, D.C. | Kept traditional closures | Cited persistent gaps in home internet and device access as the reason to avoid mandatory remote days |
| Seattle | Ended built-in snow days after the pandemic | Now defaults to either a delay or a remote-instruction day rather than a full closure |
| New York City | Shifted large storms to remote learning | Still closes for the most severe events rather than running remote instruction through everything |
Roughly 20 states leave the remote-versus-closure decision to district discretion, mostly in places that don't see much snow to begin with. Another 27 cap how many remote days can substitute for a closure before make-up days are required, and three states plus D.C. don't let remote instruction count toward the requirement at all. None of that variation is captured by looking at snowfall totals — it's set by state education law, and it changes independently of the weather.
What the Instructional-Loss Research Actually Says
Part of why districts haven't felt urgency to tighten closure thresholds as snowfall declined is that the instructional cost of routine one- or two-day closures looks smaller than assumed. Research from NWEA, an education research nonprofit, found that measurable test-score disruption is associated with closures lasting five or more consecutive days — the kind typically caused by hurricanes or wildfire smoke, not an ordinary winter storm. A single missed day here and there doesn't show up as a meaningful academic cost in that research.
That finding cuts against the idea that districts are under pressure to become more permissive about staying open. If the data superintendents are seeing says short closures don't hurt outcomes, there's no measurable incentive on the instructional side to raise the bar for calling a snow day, even as the storms behind those calls get milder on average.
Reading the North Carolina Finding Correctly
A third fewer wintry precipitation days over six decades, with a flat closure count, is a real and checkable local result if it comes from an actual paired analysis of a GHCN station near the county and that county's calendar archive. It's also, on its own, one data point. The mechanisms above — road-condition thresholds, liability posture, remote-learning accounting, parent sentiment, and the weak measured cost of short closures — are general enough that a similar decoupling is plausible in a lot of places. But "plausible" isn't "confirmed everywhere," and the honest position is that nobody has run this same paired test across a representative sample of counties and published the result.
What a real test of the hypothesis requires:
- A long-run daily record from a GHCN station reasonably close to the district, spanning at least three or four decades
- A count of days meeting a fixed threshold for wintry precipitation (not just heavy snowfall — sleet and freezing rain matter for road decisions)
- A district's actual closure calendar, not a summary — board minutes, superintendent memos, or archived calendars, since public reporting on closures is inconsistent
- A normalized closure rate rather than a raw count, because the number of school days and district size can shift over decades independent of weather
That's a reproducible method. It's also slow, county by county, because there's no national dataset that already links weather stations to school-closure calendars. Anyone testing this claim in their own region would need to build that pairing from scratch, which is exactly why one county's result hasn't yet been checked against others.
Common Questions
Does less snowfall automatically mean fewer snow days? No. Closure decisions weigh road conditions, ice, wind chill, and bus-route safety more heavily than total accumulation. A storm that drops less snow but more freezing rain can be more dangerous, not less.
Are remote-learning days the same thing as a closure? Not for this kind of analysis. Some states count a remote-instruction day as a full school day (Kentucky's NTI program, for example), which means it wouldn't show up as a "closure" in calendar records even though the building was physically shut. Any comparison across districts needs to define closure consistently — building-closed versus instructional-day-lost are different measures.
Why do some regions see more snow, not less? Lake-effect zones near the Great Lakes are the main exception. Warmer lake water can produce more intense lake-effect snowfall even as the broader region trends warmer overall, which is why a national average obscures a lot of local variation.
Is this a case where districts are just being overly cautious? That's a judgment call outside what the data settles. What the data does show is that the inputs to a closure decision — road conditions, liability exposure, remote-learning infrastructure, parent expectations — have all shifted independently of snowfall totals over the same decades. Any of those shifts alone could hold closure rates flat even with less snow falling.
The Blunt Summary
Snowfall has genuinely declined at roughly two-thirds of long-term U.S. weather stations since 1970, most sharply in the Northeast, mid-Atlantic, and Southwest. School closure rates haven't tracked that decline in a way anyone has confirmed at scale, because closures were never a direct function of snow totals — they're a function of road safety, liability exposure, remote-learning policy, and public expectation, all of which have moved on their own timelines. The one North Carolina county with a documented gap between falling snow days and flat closures is a real, checkable local result. Whether it holds up elsewhere is an open question, and answering it means pairing station records with actual district calendars, county by county, not assuming the pattern generalizes from a single case.
Has your own district's closure pattern matched what the snowfall data says it should, or diverged from it? If you've got local calendar records or know where to find them, that's exactly the kind of county-level comparison this question needs more of.
0 Comments