Winter Lows at Six U.S. Airports: 66 Winters in Context

The later 30-year winter average is higher at each of these six U.S. airport stations. In NOAA’s daily records, the mean winter low rises by about 0.7°F to 4.1°F when we compare winters ending in 1961–1990 with winters ending in 1991–2020. The size of the difference varies by airport, and a recent cold winter can still sit below the later reference.

That answers a specific question about the records at Seattle–Tacoma, Boston Logan, Chicago O’Hare, New York JFK, Phoenix Sky Harbor and San Francisco International. It does not measure the change in every neighbourhood, establish a national warming rate or isolate the cause of a station’s change. Several observing sites moved, and their instruments and reporting practices changed too. Those details belong beside the temperature history.

Compare the same winter measure

Here, “winter low” means the average of the daily minimum air temperatures during December, January and February. It is not the winter’s single coldest reading, an average of every nighttime hour or a prediction of frost in a garden.

Each winter is labelled by the year in which it ends. Winter 2026 covers December 2025 through February 2026. The first comparison period contains 30 winters ending in 1961–1990; the second contains 30 ending in 1991–2020. We give each winter equal weight within its period.

Differences are calculated before rounding, so subtracting the displayed one-decimal averages can give a slightly different result. For a temperature difference, 1°F is about 0.56°C: the changes here range from roughly 0.4°C to 2.3°C. These are descriptions of the selected station records, with no adjustment applied here for changes in site, equipment or surrounding development.

One winter can run against the longer comparison

The chart shows all 66 winters ending in 1961–2026. Each airport has its own zero line: its calculated 1991–2020 winter reference. A point at +5°F means that winter’s average daily low was 5°F above that airport’s reference. The panels share the same vertical scale, so a larger departure remains larger across airports.

Winter 2026 provides a useful counterexample to a simple “every winter gets warmer” interpretation. Boston’s mean daily low was 22.7°F, about 3.2°F below its reference. JFK was 24.6°F, about 4.6°F below its reference, and O’Hare was 19.4°F, about 0.7°F below. Their later 30-year averages can be higher while an individual winter is colder than that average.

The western examples went the other way that winter: Phoenix averaged 52.3°F, San Francisco 47.6°F and Seattle–Tacoma 39.2°F, all above their respective references. A national headline can hide those different local experiences. Even an airport-specific winter average smooths out the cold spells within its three months.

The airport name is not a guarantee of an unchanged site

NOAA’s Historical Observing Metadata Repository, or HOMR, records physical relocations for five of the six shortlisted airports. Examples include Seattle–Tacoma’s October 1996 move, O’Hare moves in March 1985 and January 1989, and Phoenix moves in 1994, 1997 and 2000. San Francisco’s record includes a January 1979 move; JFK’s includes a December 1957 move, before the chart’s period.

A move marker is a reason to inspect the history. It does not prove a temperature change was caused by the move, and it does not tell us the size of any effect. Equipment and observing practices can change without a physical relocation. Boston’s empty relocation field also does not establish that every part of its record is unchanged. Download the recorded moves and their source links.

This is why we describe airport station records rather than a single fixed measurement outside every home in a metro area. The six airports were selected as contrasting examples from our city-comparison coverage, before inspecting the temperature differences. They are not a geographically weighted national sample. Estimating a regional or national climate trend requires an appropriate dataset and treatment of non-climatic changes.

Use the history for a practical decision

For a winter trip or a move, start with the absolute temperatures in the table. O’Hare’s later reference is about 20°F, while Phoenix’s is about 47°F. A chart of departures alone would hide that large difference in the underlying climates. Then look at the variation between winters and the month that matters to your plans.

If your concern is how often temperatures fall below freezing, use a separate count of days with a daily minimum below 32°F. A mean of 35°F does not mean every day stays above freezing. We retain those counts only for winters with every accepted daily low; the two winters affected by excluded JFK readings have no complete count. A station’s air-temperature threshold still does not predict frost at a particular garden or road surface.

Eranorms helps put a selected station’s weather into historical context. Its useful role here is to make the location and period explicit when exploring a warm day, cold spell or unusual month. Use the history to understand the range of past conditions, then check a current forecast for the actual travel dates. The city weather comparison guide helps with broader seasonal planning.

Sources, calculation and limits

The published calculation uses NOAA GHCN-Daily TMIN values from six named station files: USW00024233, USW00014739, USW00094846, USW00094789, USW00023183 and USW00023234. Sources were retrieved on 7 October 2026. We checked station identities against the local app metadata and retained the NOAA histories. The figures use NOAA’s daily files rather than combining values from different daily-summary sources.

NOAA stores these temperatures in tenths of a degree Celsius. We convert them to Fahrenheit for display, omit missing values and exclude every nonblank quality flag. Two JFK readings, 22 December 1962 and 3 December 1964, carry an internal-consistency flag and are excluded. A blank flag means the value did not fail the documented checks; it does not establish that every possible error has been removed.

A monthly mean uses the accepted daily lows. We weight the three monthly means by their calendar lengths, including leap days. A winter is eligible with no more than two missing or excluded days in any month and four in the entire winter. All 66 winters pass this temperature-coverage gate at every station, including all 30 reference winters. The two affected JFK winters use the remaining accepted days; we do not insert replacement observations.

The reference spans winters ending in 1991–2020: December 1990 through February 2020. It is our calculated mean of those 30 winter means. It is not NOAA’s official 1991–2020 Climate Normal, whose production includes additional methods for station changes and missing observations. Our figures apply no station-move, equipment, urbanization or attribution adjustment. The observations support the period comparisons shown here; they do not isolate a cause or establish a nationwide rate.

• NOAA NCEI: GHCN-Daily data and methodology, with the daily-file format and quality-flag definitions.
• NOAA HOMR: station locations, observing practices and equipment history.
• NOAA U.S. Climate Normals: official reference periods and methods.
• Reproduction files: daily inputs and flags, monthly means and coverage, period comparisons, excluded readings and source URLs, hashes and method.

For the next step, read what climate normals mean and how to read a climate stripe chart. Both explain why a baseline, a location and a stated measurement belong with every weather comparison.