Stockholm Historical Weather Observations — Daily mean air temperatures since 1756

Anders Moberg

Weather observations of several meteorological variables have been made continuously at the old astronomical observatory in Stockholm since 1756. As a part of a collection of data for selected variables, the dataset provided here contains daily mean air temperatures for the period 1756–2018. Three variants are available: (i) non-homogenized daily mean temperatures, as estimated from the local temperature observations, (ii) homogenized daily mean temperatures, adjusted after 1870 to exclude the urban warming trend in the city, and (iii) homogenized daily mean temperatures having an additional adjustment applied to data for May–August in the period before 1859, to correct for a supposed warm bias of the observed temperatures due to likely poor protection of thermometers against radiation.

Citation

Anders Moberg (2019) Stockholm Historical Weather Observations — Daily mean air temperatures since 1756. Dataset version 1. Bolin Centre Database. https://doi.org/10.17043/stockholm-historical-daily-temperature-1

References

Moberg A, Bergström H, Ruiz Krigsman J, Svanered O. 2002: Daily air temperature and pressure series for Stockholm (1756–1998). Climatic Change 53: 171–212. https://doi.org/10.1023/A:1014966724670

Moberg A, Alexandersson H, Bergström H, Jones PD. 2003: Were Southern Swedish summer temperatures before 1860 as warm as measured? Int. J. Climatol. 23: 1495–1521. https://doi.org/10.1002/joc.945

Moberg A, Tuomenvirta H, Nordli Ø. 2005: Recent climatic trends. In: Physical Geography of Fennoscandia. (Ed: Seppälä M). Oxford Regional Environments Series, Oxford University Press, Oxford: 113–133

Data description

Data are provided in a single file having 7 columns, available in two file formats:

1. plain text file (.txt)

2. tab-separated value file (.tsv)

File content:

Column 1: Year

Column 2: Month

Column 3: Date

Column 4: Non-homogenized daily mean temperature according to the observations.

Column 5: Homogenized daily mean temperature, adjusted after 1870 to account for the urban heat island effect and with some gaps filled in using data from Uppsala.

Column 6: Homogenized daily mean temperature, as in column 5, but with additional adjustment before 1859 to correct for a supposed warm bias of May–August temperatures.

Column 7: Data source ID code, identifying data from: (1) Stockholm, manual observations, (2) Uppsala, adjusted to represent Stockholm, (3) Stockholm, automatic weather station (used from 2013 onwards).

Temperature unit: °C. Code for missing values: -999.

The dataset also includes correction values used to adjust daily mean temperatures in the most recent years (after 1998), in order to derive the homogenized temperature data in columns 5 and 6 from the non-homogenized data in column 4.

Further information about the data file content is found in a README text file.

Comments

The observation site is included in the national station network managed by the Swedish Meteorological and Hydrological Institute (SMHI), with station number 98210 (manual station) and 98230 (automatic station), but data in the current dataset can differ from those in the SMHI database.

The first two variants of daily mean temperatures presented here have been derived from the dataset with thermometer observations as described by Moberg et al. (2002). The first variant (non-homogenized data) consists of daily mean temperatures calculated to account for changes in the actual observation times and number of observations. Thus, they are estimates of daily mean temperatures obtained directly from the observations without any further corrections.

The second variant (homogenized data) includes the following adjustments, relative to the first variant:

1756.04.03–1756.10.31: The observed temperatures during this period are judged to be too high, and of poor quality. Therefore, they have been replaced with contemporaneous data extrapolated from Uppsala through linear regression.
1763.02.25–1763.02.28: No observations were made. Daily temperatures have been extrapolated from Uppsala.
1819.08.01–1825.01.13: Correction by −0.7°C because an incorrect thermometer was probably used.
1870.01.11–2018.12.31: Correction for urban heat island trend and other inhomogeneities.

The adjustment for the urban warming trend, and the effect of other inhomogeneities detected in homogeneity tests against surrounding reference station data, is applied such that the most recent homogenized temperatures are made somewhat colder than the non-homogenized temperatures in order to make the entire series being respresentative of the rural conditions that prevailed before the mid-19th century. The size of the adjustment changes with time and varies over the year. It causes homogenized temperature data after 1967 being on average 0.8°C colder than non-homogenized temperatures.

The third variant (with additional adjustment to data for May–August in the period 1756–1858), has been derived as in Moberg et al. (2005) based on conclusions in Moberg et al. (2003). It includes the following additional adjustments, relative to the second variant:

May: day 1–4: 0.0°C, day 5–9: −0.1°C, day 10–13: −0.2°C, day 14–18: −0.3°C, day 19–22: −0.4°C, day 23–27: −0.5°C, day 28–31: −0.6°C.
June: day 1–30: −0.7°C.
July: day 1–31: −0.7°C.
August: day 1–4: −0.6°C, day 5–9: −0.5°C, day 10–13: −0.4°C, day 14–18: −0.3°C, day 19–22: −0.2°C, day 23–27: −0.1°C, day 28–31: 0.0°C.

This gives an average adjustment by −0.3°C for May and August and −0.7°C for June and July. The motivation for this additional adjustment is based on the available knowledge about the position of the old thermometer in combination with results from homogeneity testing and other statistical modelling. However, it should be seen as an ad hoc judgement rather than a result from a strict statistical analysis. Nevertheless, this variant of daily mean temperatures is judged to be the one that currently best represents the true temperature history for the May–August season. Further research would be needed to better assess the need for adjustments in the early period and to determine their size and how they should be applied. Actually, there is a need also to undertake further analysis of how to better adjust the temperature data for the modern urban heat island effect.

The number of daily observations used to calculate the daily means varies with time. This number is typically:
1756–1760: 2 observations
1761–1858: 3 observations
1859–1946: 3 observations & Tmin in summer months
1947–2018: 3 observations & Tmin & Tmax

Data from before 2013 are based on manual observations. Data from 2013 onward are based on the automatic weather station (SMHI station 98230).

Information about instrument positions and original data sources is available in the comments to the associated dataset with thermometer observations.

Figures that illustrate daily mean air temperatures in each separate year within the period 1756–2018 are available as PNG image files. The daily mean temperatures in this dataset are used to derive the corresponding dataset with monthly mean air temperatures.