Degree-days sum each day’s departure from a base temperature — CDD = Σ max(0, T̄ − base), HDD = Σ max(0, base − T̄) — added up over a season. Degree-days are an approximate indicator of cooling (CDD) and heating (HDD) demand. This tool computes one figure per year from daily records rather than monthly averages, with T̄ the mean of the day’s maximum and minimum. The base is adjustable per city at 55, 60 or 65 °F. Figures are held as Fahrenheit degree-days; the Celsius values are those numbers divided by 1.8, which is exact because the two sets of bases correspond.
Calculations use these seasonal bins: winter is December to February, spring March to May, summer June to August, fall September to November. A season counts only when it holds at least 78 days of observations, and an annual total is formed only when all four seasons qualify, so the current year contributes the seasons it has finished and no annual figure.
Three chart types are provided. Trend fits a model to one city and reports adjusted R² and AIC so competing models can be weighed against each other. Distribution groups the same series by decade, each year a dot, with the decade’s median and range; a threshold bar can be moved to compare how the early decades sit against the recent ones. Compare sets one Old World city against one New World city, with a choice of two comparisons.
Absolute (Full Record) shows each city across its downloaded record, shaded against its own first available 30 years. The darker band spans the middle half of that early window, the lighter band its 10th to 90th percentile. The two baselines cover different calendar years for each city, so the baselines cannot be directly compared.
Indexed (Common) clips both records to their first shared year and normalises each to an index of 100 at that year, so proportional change reads on one scale however far apart the cities’ absolute levels are. Global CO₂ is drawn dashed on the same index, set to 100 at the shared start year or at 1869, whichever is later — the CO₂ series itself begins in 1869, and four pairs in this set reach back further than that. With this chart the relative rise or fall in degree-days between the two cities can be read directly.
The arithmetic is benchmarked against three independent publishers, with the calculations in the Validations download. Environment and Climate Change Canada publish a heating and a cooling degree-day value on every daily row at base 18 °C; against Toronto our annual totals agree to 0.06% on HDD and 0.46% on CDD. NOAA / National Weather Service publish annual tables for Central Park at base 65 °F; we sit 2.5% below on CDD and 1.3% above on HDD, a consistent offset that comes from the NWS convention of rounding each day’s mean to a whole degree before differencing the base, instead of using the exact mean. Eurostat provides a methodology cross-check instead of a precise validation, because their approach uses spatial averaging instead of discrete station data. We implement Eurostat’s two-threshold rule on our De Bilt daily series and compare with their published Netherlands figure, showing strong agreement at 1.84% on HDD despite the different approaches, a 25 km grid averaged over a country versus a single station. No publisher issues degree-days below 65 °F / 18 °C, so the 55 and 60 °F bases are not externally benchmarked; they run through the same summation with one parameter changed.
Central England is a regional composite rather than a single station, and its daily series begins in 1772 — the historical 1659 start is monthly only, and we do not compute degree-days from monthly means, which would bias the result. Stockholm and Amsterdam · De Bilt are homogenised, each by its own national custodian rather than by us; the other eight are not. At Stockholm the thermometer exposure changed several times at the Observatory and the city grew around it, and Moberg’s correction removes both. At De Bilt, KNMI correct for the move of about 300 m to a more open site on 27 August 1951 and for the new screen type introduced on 16 September 1950, deriving the corrections from measurements taken in parallel at the old and new locations. A homogenised series will tend to show less warming than the same record left unadjusted, so the comparison is not perfectly even across all ten. Toronto and Montreal are each built from two records joined across a short overlap, with measured average differences of 0.10 °C and −0.16 °C respectively. Montreal’s usable record begins in 1809; fragments reach back to 1742 but are separated by a gap too long to fit a trend across. Blue Hill is a summit observatory at 190 m and reads cooler than Boston at sea level. Charleston is the downtown site, not the airport. Madrid begins in 1920, confirmed independently by NOAA, ECA&D and AEMET. Latitudes are shown for each city because the pairs are not matched on them: Madrid and New York sit within half a degree of each other, while Stockholm and Charleston are 27 degrees apart. Read a widely separated pair for the change in each city rather than for the difference in level between them.