Official daily values from DMI (the Danish Meteorological Institute) are available for Denmark from 2011 (mean temperature and precipitation from 1991). For earlier years the daily maximum temperature (TX), minimum temperature (TN) and mean temperature (TG) shown on this site have been reconstructed from DMI's raw observations, which are available from 1953. Reconstructed values are an approximation: they are not quality controlled by DMI and usually differ by a few tenths of a degree from what an official daily value would have been. Official values are never replaced; reconstructed values only fill days where no official value exists.
DMI publishes all raw station observations back to 1953 (metObs bulk archive). Until about 1986 these are synoptic observations, taken every 3 hours (00, 03, 06 ... 21 UTC). At 06 and 18 UTC the stations also report the highest and lowest temperature of the preceding 12 hours. From 1987 many stations report every hour, including the highest, lowest and mean temperature of that hour, so a daily value can be built in the same way DMI does it today.
For five stations with a long record (Vestervig, Nordby/Fanø, Tranebjerg, København and Hammer Odde) DMI also published quality-controlled daily maxima and minima in its Historical Climate Data Collection (DMI Report 20-02). These are used where available.
DMI's current daily values use the local calendar day (00:00 to 24:00 Danish time). The classical Danish climate day, still visible in the long historical series, runs from 06 UTC to 06 UTC: the thermometers were read in the morning, so the highest and lowest temperature cover the 24 hours up to that reading. When only 12-hour reports exist, the 06–06 UTC day is the best that can be built, which is why a small, systematic difference with the local day remains.
Every reconstructed value is built with one of the methods below. For each variable the first method that has enough data is used.
| Method | Variable | How it is built | Accuracy |
| Hourly | TX, TN | Highest of the hourly maxima and lowest of the hourly minima of the local day; at least 22 of 24 hours must be present. | About 0.05–0.1 °C from the official value |
| Hourly mean | TG | Mean of the 24 hourly mean temperatures (or, if those are missing, of the hourly temperatures) of the local day; at least 22 of 24 hours. | About 0.05–0.1 °C |
| DMI long series | TX, TN | Values published in DMI Report 20-02 for the five long-record stations. | Quality controlled by DMI |
| Two reports | TX, TN | TX: highest of the 12-hour maxima reported at 18 UTC and at 06 UTC the next morning. TN: lowest of the 12-hour minima reported at 18 UTC the evening before and at 06 UTC. | Identical to DMI's long series in 97–100 % of the days; about 0.1–0.5 °C from the local-day value |
| One report plus night | TX, TN | Only the 18 UTC report exists. The missing 12 hours are covered by the four 3-hourly temperatures of that period (21, 00, 03 and 06 UTC). | TX about 0.1 °C, TN 0.1–0.3 °C too warm |
| 3-hourly mean | TG | Mean of the 3-hourly temperatures from 03 to 24 UTC; at least 7 of the 8. | About 0.3 °C, no systematic bias |
| 3-hourly maximum | TX | Highest of the 3-hourly temperatures; last resort when no 12-hour report exists. | On average 0.6 °C (summer 0.8 °C) too low |
Order of preference: TX: hourly, DMI long series, two reports, one report plus night, 3-hourly maximum. TN: hourly, DMI long series, two reports (18 UTC the evening before plus 06 UTC), two reports (06 UTC plus 18 UTC the same day), one report plus night. TG: hourly mean, 3-hourly mean.
| Period | Data available | Maximum (TX) | Minimum (TN) | Mean (TG) | Why |
| 1953–1969 | 3-hourly observations at 7–30 stations; the 12-hour extremes are mostly reported at 18 UTC only | DMI long series (5 stations); else one report plus night; else 3-hourly maximum | DMI long series (5 stations); else one report plus night | 3-hourly mean | Without the 06 UTC report a full 24-hour extreme cannot be built, so the other 12 hours are estimated from the 3-hourly temperatures. Values are least accurate in this period. |
| 1970–1986 | 3-hourly observations at 30–40 stations; both 06 and 18 UTC reports on most days | DMI long series; else two reports; else one report plus night; else 3-hourly maximum | DMI long series; else two reports; else one report plus night | 3-hourly mean | Two 12-hour reports together cover 24 hours and reproduce DMI's own long series almost exactly. Months with only the 18 UTC report use the night method. |
| 1987–1990 | Hourly maxima, minima and means at 23–60 stations | Hourly; else as 1970–1986 | Hourly; else as 1970–1986 | Hourly mean; else 3-hourly mean | Hourly values give the same local-day definition DMI uses today, so these years connect to the official series without a break. |
| 1991–2010 | As above; official mean temperature and precipitation exist | Hourly; else as 1970–1986 | Hourly; else as 1970–1986 | Official where it exists; gaps filled with hourly or 3-hourly mean | Official values are kept; reconstruction only fills missing days. |
| 2011–today | Official daily values | Official | Official | Official | Nothing is reconstructed. |
Some days remain empty. The archive does not contain a 12-hour report for every station and day (in January 1982, for example, station 6190 has maxima on only a part of the days), and some stations only observe during the daytime, so no 3-hourly values exist for the night. Other variables, such as precipitation, sunshine, wind and pressure, are not reconstructed before 1991 or 2011; they are simply empty. Daily precipitation from 1991 is official.
The reconstructed values have not been checked by DMI, and the day definition differs between methods (local day for hourly data, 06–06 UTC for the 12-hour reports). A single estimated value can therefore be a few tenths of a degree away from the official one. The long-term averages (the "normals" on the averages pages) use all available years, so they mix official and reconstructed values. Records (highest and lowest values) can come from a reconstructed day; treat records from before 1987 as indicative.