FORECASTSEASON.COMRoute 19 · Degree DaysSEAS5 ExplorerProbabilityAnalogsSecond OpinionCitiesModelsDriversAbout
SEASON90Forecast SeasonSeasonal Outlooks · U.S. · Europe · Global
Forecast Season/Degree Days
Route 19 · Energy demand

Degree days

How much heating and cooling the season will ask for, from every one of SEAS5's 51 members — each member's daily high and low turned into a daily degree day, summed by month, weighted by where people live. Corrected against the observed record, and checked against last winter.

ENERGY19

US heating, Nov–Mar

96% of normal
3,114 °F-days population-weighted, against a 1991–2020 normal of 3,260

Europe heating, Nov–Mar

93% of normal
1,881 °C-days population-weighted, against a 1991–2020 normal of 2,031

US winter in the cold third

8%
of 51 members; 61% in the mild third, against 33% each by chance

Last winter's error, US

6%
typical monthly miss of the corrected median, US HDD, 2025 runs (11% uncorrected)

What the forecast says

SEAS5 started October 2026 · 51 members

The lower 48. The corrected median puts November–March heating demand at 3,114 °F-days, 96% of the 1991–2020 normal, with 80% of members between 2,784 and 3,319. 23 of the 35 winters since 1991/92 needed at least that much; last winter took 3,008.

Europe. The corrected median puts November–March heating demand at 1,881 °C-days, 93% of the 1991–2020 normal, with 80% of members between 1,737 and 2,097. 29 of the 35 winters since 1991/92 needed at least that much; last winter took 1,898.

Month by month. For the US the mildest month relative to normal is April 2027 (81% of normal heating demand) and the closest to a cold month is February 2027 (96%). Every month's spread across members is wider than its shift from normal — this is a tilt, not a certainty.

Last winter as a check. Across the 18 months the three 2025 runs covered, the corrected US median missed observed heating demand by 6% on average, against 11% before correction; the observed value fell inside the members' 10–90% range in 94% of months. The correction is doing its job.

Month by month

every member, against the record
Region
N. AmericaEurope
Degree days
Weighting
Values
Run
each member10–90% of membersmedian1991–2020 normal1991–2020 rangeobserved (AgERA5)

Every region

corrected median as a share of normal
Heating demand vs normal≥ 125%112–125%104–112%±4%88–96%75–88%≤ 75%blue = colder
Oct 2026Nov 2026Dec 2026Jan 2027Feb 2027Mar 2027Apr 2027Nov–Mar
North America · °F-days
United States (lower 48)341.7 M people · 934 cells90%19594%44294%68194%76796%65892%51581%26096%3,114
New England15.3 M people · 36 cells110%40795%62993%91792%1,07891%93986%80377%45890%4,346
Middle Atlantic42.6 M people · 46 cells108%33296%57793%83893%1,00791%87188%73877%38793%4,072
East North Central49.7 M people · 111 cells97%34399%68296%98592%1,10893%96188%75278%38993%4,504
West North Central20.9 M people · 168 cells82%30496%70796%1,07092%1,17196%1,00391%73775%33695%4,758
South Atlantic68.0 M people · 104 cells93%10295%28298%467101%570101%45894%32084%12498%2,098
East South Central19.7 M people · 60 cells67%9794%377101%642102%749105%607101%42986%160102%2,830
West South Central43.7 M people · 142 cells27101%226106%452103%500110%38498%20769%48106%1,805
Mountain26.3 M people · 252 cells71%23796%59394%84894%853103%770104%64284%36798%3,729
Pacific55.5 M people · 116 cells66%10886%32184%48285%48796%46292%39481%24888%2,149
Canada37.5 M people · 1818 cells95%53497%88190%1,14290%1,30589%1,12787%97980%58790%5,452
Europe · °C-days
Europe (EU, UK, Norway, Switzerland)497.2 M people · 846 cells78%15194%30591%40393%44690%37498%36192%22693%1,881
Germany82.9 M people · 64 cells78%18694%35689%43993%48789%40999%39895%25294%2,114
France62.1 M people · 87 cells77%11993%27589%35490%38289%33299%31893%21192%1,665
United Kingdom62.3 M people · 58 cells80%17395%29992%36794%38994%35195%34891%26494%1,753
Italy52.6 M people · 62 cells66%6894%22488%33091%37389%31693%27784%15590%1,512
Spain39.9 M people · 73 cells82%4989%17191%26684%27384%23196%21987%13287%1,143
Poland39.4 M people · 55 cells82%21897%40091%50194%56394%48895%43792%25494%2,394
Netherlands16.4 M people · 11 cells79%14892%28591%37592%40691%36298%35798%25194%1,808
Nordics24.9 M people · 269 cells85%28398%44493%52899%613101%57497%52795%37598%2,709

Each cell: the corrected median as a share of the 1991–2020 normal, with the value itself underneath. Hover for the 10–90% range and the chance of landing below, near or above normal (the terciles of the 30 observed years). Cooling cells are coloured the other way round: rust is more cooling demand. A month with almost no normal demand (under 60 °F-days or 20 °C-days) shows the value alone. Controls above choose HDD/CDD and the weighting.

Maps

departure from normal, land only
Month
Heating degree days, October 2026, North America, departure from normal
Heating degree days, October 2026, Europe, departure from normal

Corrected ensemble-mean degree days minus the 1991–2020 normal. The HDD/CDD control above switches these too; cooling maps are drawn only for months where the normal has cooling demand to speak of.

Cities

the 1° cell each city sits in
Oct 2026Nov 2026Dec 2026Jan 2027Feb 2027Mar 2027Apr 2027
North America · °F-days
New Yorkcell 40.5°, -74.5°291108%53094%79392%94192%81391%64384%29670%
Bostoncell 42.5°, -71.5°411110%64395%94193%1,09191%95292%77584%40773%
Philadelphiacell 39.5°, -75.5°240112%47795%72692%86592%75993%58686%25372%
Washingtoncell 38.5°, -77.5°211105%45795%70893%82093%71396%51087%19773%
Atlantacell 33.5°, -84.5°7570%33596%575103%673106%515106%373109%11787%
Chicagocell 41.5°, -87.5°334100%67599%98795%1,11291%95893%73787%37176%
Detroitcell 42.5°, -83.5°399100%729100%1,00394%1,15592%99192%78085%36568%
Minneapoliscell 44.5°, -93.5°41385%85695%1,22592%1,36690%1,14290%84083%43475%
Dallascell 32.5°, -96.5°19231101%476104%527102%423113%21396%4366%
Houstoncell 29.5°, -95.5°3116100%294111%341108%254120%124110%13
Denvercell 39.5°, -104.5°28964%73295%1,02696%1,05798%968103%857107%50285%
Phoenixcell 33.5°, -112.5°09794%30595%26390%237123%98114%9
Los Angelescell 34.5°, -118.5°2937%23082%43788%42488%442106%374106%20285%
San Franciscocell 37.5°, -122.5°5636%12843%24858%28764%28274%30480%25877%
Seattlecell 47.5°, -122.5°33787%56594%67387%66487%59589%51881%36173%
Torontocell 43.5°, -79.5°404103%71099%95493%1,13892%1,00490%88388%48875%
Montréalcell 45.5°, -73.5°517104%81395%1,14690%1,38890%1,19990%1,01289%54179%
Calgarycell 51.5°, -114.5°62180%1,07695%1,24887%1,33691%1,12389%1,00987%66784%
Vancouvercell 49.5°, -123.5°50592%73896%81488%77584%70687%65982%49376%
Europe · °C-days
Londoncell 51.5°, -0.5°15076%28493%36693%38093%34292%33696%24892%
Pariscell 48.5°, 2.5°13182%27491%35387%39292%32988%31498%21396%
Berlincell 52.5°, 13.5°17675%36496%45391%50094%43194%38995%23193%
Amsterdamcell 52.5°, 4.5°12674%26091%34991%38793%35292%35798%26398%
Madridcell 40.5°, -3.5°7887%25095%34593%34486%28786%272100%16389%
Romecell 41.5°, 12.5°06772%17278%22186%19081%18089%8982%
Milancell 45.5°, 9.5°7963%27192%39489%42090%34087%27792%13480%
Warsawcell 52.5°, 21.5°23785%42199%52092%57193%50495%43994%25693%
Viennacell 48.5°, 16.5°14568%34493%47290%50691%41590%35793%19290%
Stockholmcell 59.5°, 18.5°26984%41696%50493%590101%551101%52799%38094%
Oslocell 59.5°, 10.5°27483%44298%52690%634101%586104%49494%31484%

Corrected median for the month, with its share of normal. A 1° cell is about 100 km across, so these are the city's surroundings, not its airport; the normal is taken on the same cell, so the comparison is fair.

Last winter, checked

three 2025 runs against AgERA5

Pick a 2025 run in the Run control above and the chart shows it against what happened. The summary, population-weighted heating degree days, every month each run covered:

Typical miss,
raw		Bias,
raw		Typical miss,
corrected		Bias,
corrected		Observed inside
10–90% of members		Correction
United States (lower 48)18 run-months11%+2%6%0%94%✓ better
Canada23 run-months9%−4%9%+1%83%✓ better
Europe (EU, UK, Norway, Switzerland)22 run-months13%+10%10%+5%100%✓ better
Germany22 run-months12%+7%10%+4%100%✓ better
France20 run-months18%+10%13%+4%85%✓ better
United Kingdom22 run-months18%+12%14%+6%91%✓ better
Italy20 run-months19%+19%12%+3%80%✓ better
Poland22 run-months10%0%12%−2%91%✗ worse

Typical miss: the mean absolute difference between the members' median and AgERA5, as a share of AgERA5, over every month with real heating demand. Bias: the same with the sign kept, so positive is a forecast that was too cold. April is past the monthly reforecast's six months and borrows February's correction; February is past the monthly reforecast's six months and borrows January's correction; March is past the monthly reforecast's six months and borrows February's correction; the October 2025 run has no reforecast of its own and uses the September start's; the October 2026 run has no reforecast of its own and uses the September start's.

How it is made

six things to know

Conventions

Each continent's own

North America in °F-days against 65 °F, as NOAA and the US gas and power markets count them. Europe in °C-days by Eurostat's rule: heating is 18 °C minus the daily mean when the mean is 15 °C or below; cooling is the mean minus 21 °C when it is 24 °C or above. The daily mean is (Tmax + Tmin) / 2 on both sides.

Order of operations

Days first, members kept apart

Each member's every day becomes a degree day before anything is averaged or summed. A degree day is not linear in temperature, so the degree days of an average temperature are not the average of the degree days — a mild ensemble mean hides the cold members that make the bill.

Correction

Anchored to the observed record

SEAS5 has no daily Tmax/Tmin reforecast on file, so each calendar month and 1° cell is shifted by the model's 1993–2016 mean temperature (from its monthly reforecast, plus its own gap between the max/min midpoint and the daily mean) minus AgERA5's over the same years. The forecast's anomaly — including any warming — is kept.

Observed

AgERA5, on the same grid

Copernicus' AgERA5: daily day-time maximum and night-time minimum, local days, 0.1° land averaged into each SEAS5 1° cell. SEAS5's days are UTC days; summing by month absorbs most of that difference, which is why nothing here is published at a daily step.

Weighting

People or area

Population weights are GHS-POP 2025 counts; regions are Natural Earth countries and, for the US census divisions, states. Area weighting treats every square kilometre alike — what a climatologist would average, where population weighting is what a utility pays.

Where next

The temperature behind it

The monthly temperature outlook these numbers rest on is on the Probability Plate; individual members day by day on the Daily Forecast; and the other systems' view on Multi-model.