Frost dates are shifting
Comparing the two most recent U.S. Climate Normals across the 3,263 cities where the same weather station reports both. We measure the 50%-probability 32°F last-spring and first-fall frost dates and how far each moved. 63% of cities now see an earlier last spring frost, and 66% a later first fall frost.
How much the frost-free season changed
Each bar counts cities whose frost-free season changed by that many days (1991–2020 minus 1981–2010). Bars right of zero gained growing season.
Median change by state
Frost-free days gained across each state’s published cities, longest gain first.
The 50 biggest movers
Cities whose frost-free season changed the most between the two normals.
By state
50 states| State | Cities | Spring | Fall | Season |
|---|---|---|---|---|
| Delaware | 11 | −7 d | +6 d | +13 d |
| Nevada | 17 | −9 d | +4 d | +12 d |
| Vermont | 4 | −5 d | +7 d | +12 d |
| Maine | 11 | −4 d | +5 d | +10 d |
| District of Columbia | 1 | −4 d | +5 d | +9 d |
| Montana | 5 | −4 d | +4 d | +9 d |
| Idaho | 22 | −3 d | +5 d | +8 d |
| New Hampshire | 12 | −4 d | +4 d | +8 d |
| Rhode Island | 7 | −2 d | +5 d | +7 d |
| Wyoming | 5 | −3 d | +2 d | +7 d |
| Alabama | 52 | −5 d | 0 d | +5 d |
| Alaska | 3 | −2 d | +3 d | +5 d |
| Arkansas | 35 | −3 d | +1 d | +5 d |
| Connecticut | 21 | −2 d | +3 d | +5 d |
| Illinois | 182 | −1 d | +4 d | +5 d |
| Massachusetts | 76 | −2 d | +4 d | +5 d |
| Minnesota | 70 | 0 d | +4 d | +5 d |
| Mississippi | 38 | −4 d | 0 d | +5 d |
| Nebraska | 11 | −1 d | +4 d | +5 d |
| New Jersey | 124 | −2 d | +4 d | +5 d |
| New Mexico | 15 | −3 d | +2 d | +5 d |
| Oregon | 50 | −4 d | +1 d | +5 d |
| Pennsylvania | 63 | −1 d | +3 d | +5 d |
| Virginia | 97 | −3 d | +1 d | +5 d |
| West Virginia | 13 | −1 d | +3 d | +5 d |
| Colorado | 64 | 0 d | +3 d | +4 d |
| Florida | 241 | +3 d | +6 d | +4 d |
| Kentucky | 38 | −1 d | +3 d | +4 d |
| Louisiana | 52 | −3 d | −1 d | +4 d |
| Maryland | 99 | −2 d | +3 d | +4 d |
| Missouri | 72 | −2 d | +2 d | +4 d |
| New York | 145 | −2 d | +3 d | +4 d |
| North Carolina | 74 | −3 d | +1 d | +4 d |
| Ohio | 161 | 0 d | +3 d | +4 d |
| California | 399 | −3 d | 0 d | +3 d |
| Georgia | 95 | −2 d | −1 d | +3 d |
| Iowa | 39 | +1 d | +4 d | +3 d |
| Michigan | 68 | 0 d | +4 d | +3 d |
| Tennessee | 58 | −2 d | +2 d | +3 d |
| Wisconsin | 67 | 0 d | +3 d | +3 d |
| Indiana | 70 | +1 d | +4 d | +2 d |
| North Dakota | 5 | 0 d | +2 d | +2 d |
| South Carolina | 50 | −2 d | 0 d | +2 d |
| Texas | 213 | −2 d | −1 d | +2 d |
| Utah | 53 | −2 d | 0 d | +2 d |
| Arizona | 53 | −1 d | +1 d | +1 d |
| Kansas | 29 | 0 d | +1 d | +1 d |
| South Dakota | 11 | +1 d | +3 d | +1 d |
| Oklahoma | 34 | 0 d | 0 d | 0 d |
| Washington | 128 | −1 d | 0 d | 0 d |
Methodology & caveats
Both figures come from NOAA NCEI U.S. Climate Normals: the last-spring (PRBLST) and first-fall (PRBFST) freeze-probability products at 32°F, 50% probability. For each published city we take its nearest complete-normals station and compare that station’s 1991–2020 anchor against its 1981–2010 anchor, matched by GHCND station id. Shift = 1991–2020 date − 1981–2010 date, as a signed circular day distance.
This compares two published normals, not raw temperature trends. Part of a station’s change can come from instrumentation or siting changes, a shifted station location, or the sparse, noisy frost record at near-frost-free sites. The national and per-state medians are far more robust than any single city. Coverage: 3,263 of the published cities (78.6%) have both normals vintages at their matched station. How we compute this.