One night in April: Champagne's 2026 frost at our grid point and at thirteen stations
Our Champagne grid point recorded no sub-zero night in April or May 2026, but ten of the thirteen Météo-France stations reporting in the Marne did, nine of them on 2 April; set against six long station records, the grid point resembles four and misses much of the frost at the other two.
Published 21 November 2026 · Figures to 31 May 2026
Spring frost is being talked about again in Champagne, so we counted it two ways. The first is the daily weather record we keep for the region: ERA5-Land, the gridded reanalysis published by the Copernicus Climate Change Service, read at a single point at 49.04° N, 4.00° E, from 1950. The second is Météo-France's open daily record for weather stations in the Marne. In both, a frost day is a day on which the minimum air temperature fell below 0 °C, and the window is 1 April to 31 May.
At the grid point the count for 2026 is nought. The coldest night of the two months was 2 April, at 1.1 °C, and only one other, 3 April, went below 2 °C. The grid's last sub-zero night was 27 March, at −2.0 °C, twelve days after another on 15 March at −1.2 °C. March as a whole averaged 8.3 °C against 6.8 °C for the Marches of 1991–2020, taking each day as the midpoint of its maximum and minimum, the eighth warmest of the seventy-seven in the record.
The stations agree about March and disagree about April. All thirteen Marne stations with a complete April and May were below zero on 27 March. Nine of them were below zero again on 2 April, the night of the grid's own low, and ten recorded at least one sub-zero night in April or May. The two nearest the grid point, both within five kilometres, sit either side of it: Chouilly read 0.7 °C on 2 April and had no frost day; Dizy read −0.7 °C and had one. Further away the counts climb. Reims-Prunay had ten frost days, the last on 26 April. Mourmelon-Grand had seventeen, with −6.8 °C on 2 April and a last sub-zero night on 12 May. The median across the thirteen was two.
The spring of 2025 shows the same split. The grid point had no frost day and a lowest minimum of 2.8 °C; Chouilly and Dizy had none either; nine of the thirteen stations recorded at least one.
In its own long record the grid point has seen spring frost thin out. Twenty-two of its seventy-seven springs passed without a frost day. It averaged 3.2 a spring in 1961–1990 and 1.6 in 1991–2020; the 1970s averaged 4.8, the highest of any decade, and the 2010s 1.0, the lowest. Four springs had nine, 1956, 1970, 1975 and 1984, and none since 1984 has had more than six. The coldest April or May night it holds is 12 April 1986, at −5.0 °C, and it has recorded a sub-zero night in May three times, in 1957, 1960 and 1997.
Six Marne stations with long records, none of them reporting after 2022, show how far that can be trusted. Over the same years the grid point resembles four of them. Épernay, Mailly, Avize and Bouzy averaged between 1.04 and 1.95 frost days a spring, against 2.18 to 2.78 at the grid point, and at Épernay, where the station stood within a kilometre of the grid point, there is no spring from 1959 to 2004 in which the station recorded frost and the grid recorded none. The other two are a different matter. Reims-Courcy averaged 4.9 frost days a spring to the grid's 2.8 and recorded frost in fourteen springs the grid shows as clear. Chambrecy averaged 8.6 to the grid's 2.2, had twenty frost days in 1997, and recorded a sub-zero night in May in twenty-nine of its forty-six springs.
On direction the seven records agree. All six stations had fewer frost days in their later years than in their earlier ones, by between 16 per cent, at Reims-Courcy, and 55 per cent, at Bouzy; the grid's own fall of a half sits near the top of that range. The last sub-zero night of the first half of the year also came earlier at all six, by about three to nine days. At the grid point it moved from 10 April in 1961–1990 to 7 April in 1991–2020, while the April-to-May average temperature there rose from 10.5 to 11.8 °C.
Three cautions belong with the figures. A grid cell gives one number for an area several kilometres across, and these comparisons show that it stands for some sites and not for others; nothing in our tables says which vineyards resemble which station. Both series are air temperatures, not readings at the vine, and neither says anything about damage, which depends on how far growth has gone when the cold arrives; we hold no record of when Champagne's vines break bud. And the station periods differ from one another, so their percentages are not like for like. What the two records show together is narrower: 27 March 2026 was a frost night across the Marne, 2 April was one at nine stations in thirteen and not at our grid point, and the grid point alone would have missed it.
Contains modified Copernicus Climate Change Service information 2026. Station data: Météo-France, Données climatologiques de base – quotidiennes, via data.gouv.fr; the Marne file was updated 24 June 2026; files downloaded 1 October and 9 November 2026. Licence Ouverte / Open Licence 2.0.
A frost day is a day whose minimum air temperature fell below 0 °C between 1 April and 31 May; the grid series is ERA5-Land at 49.04° N, 4.00° E; station counts use the same definition on Météo-France daily minima; period averages are of daily (tmax + tmin)/2; station periods differ and are stated where used.
Sources: ERA5-Land (region_weather_daily, slug champagne, daily from 1950); Météo-France daily station data for the Marne (six stations 1950–2022 from the 1 October extract; thirteen stations 2025–26 from the file downloaded 9 November 2026)
Contains modified Copernicus Climate Change Service information 2026. Station data: Météo-France, Données climatologiques de base – quotidiennes, via data.gouv.fr; the Marne file was updated 24 June 2026; files downloaded 1 October and 9 November 2026. Licence Ouverte / Open Licence 2.0.
Temperature measures warmth, not wine. This issue records what the weather record shows and makes no claim about any wine or vintage.