Europe ERA5 Air and Sea Temperature Tracker

Domain and source data

Dataset

ECMWF Reanalysis v5 (ERA5)

Spatial domain

30°–72° N, 25° W–40° E

CDS grid

0.25° × 0.25°; 44,109 grid points

Data coverage

Loading coverage…

The tracker uses ERA5 hourly data on single levels from the Copernicus Climate Change Service (C3S) Climate Data Store. ERA5 is a global reanalysis: a historical reconstruction of the atmosphere produced by repeatedly running a modern weather model over past decades while continuously incorporating observations from satellites, weather stations, balloons, aircraft, ships and other observing systems. This provides a spatially and temporally complete estimate of how atmospheric conditions evolved, including in places and periods where direct observations are sparse. ERA5 is produced with the ECMWF Integrated Forecasting System (IFS) and four-dimensional variational data assimilation (4D-Var), which combines these observations with the numerical model to reconstruct a physically consistent atmospheric state through time.

2 m air temperature. This is the estimated air temperature two metres above the model surface and is available across both land and sea. In the IFS it is a diagnostic screen-level quantity rather than the temperature of a model level: it is derived from the temperature at the lowest atmospheric model level and the underlying surface skin temperature using boundary-layer stability information. The hourly 2 m temperature distributed through the ERA5 Climate Data Store is an analysed parameter, so it is consistent with the observation-constrained atmospheric state produced by the reanalysis.

Sea surface temperature (SST). ERA5 SST is a prescribed foundation sea surface temperature rather than a prognostic temperature generated by the atmospheric model. Before September 2007 it is based on HadISST2; from September 2007 onward it is based on the Met Office OSTIA analysis. The ERA5 SST represents a foundation temperature without a resolved daytime solar-warming cycle. Although the field is available at hourly timestamps, its SST content is updated once per day.

ERA5 values are reanalysis estimates rather than direct weather-station observations. Apparent local records or other extremes that are important for a specific application should therefore be checked against observations and, where appropriate, other climate datasets.

Data availability. Recent data are initially provided by C3S as preliminary ERA5T and are typically available about five days behind real time, with no guaranteed release time. This tracker updates daily and normally operates about six days behind real time. C3S subsequently replaces ERA5T with final ERA5 data, normally about two months later, so values from the most recent months should be regarded as preliminary.

Data citation: Hersbach H, Bell B, Berrisford P, et al. ERA5 hourly data on single levels from 1940 to present. Copernicus Climate Change Service Climate Data Store. doi:10.24381/cds.adbb2d47. For broader background on reanalysis, see Advancing Reanalysis.

Copernicus attribution and disclaimer. Contains modified Copernicus Climate Change Service information 2026. The European Commission and ECMWF are not responsible for uses made of the Copernicus information or data presented on this page.

Figure licence. Figures generated by this tracker may be reused and adapted under the Creative Commons Attribution 4.0 International licence (CC BY 4.0) , including for commercial purposes, provided that XpertScientific is credited and the ERA5/Copernicus data-source attribution is retained or otherwise provided. If a figure is modified, the modification should be indicated.

How the tracker is calculated

Daily temperature. Daily means cover 00:00–23:00 UTC. Historical values were calculated from 24 hourly ERA5 values; updates use C3S daily statistics with the same UTC window and hourly sampling. 2 m air temperature requires all 24 hours; SST is retained only for complete valid days and, although timestamped hourly, is updated once daily.

Map smoothing. The raster is bilinearly interpolated between 0.25° cells for display only. SST interpolation is weighted by ocean fraction from the ERA5 land–sea mask; pixels with less than 50% ocean fraction are transparent. Map clicks snap to the nearest ERA5 grid point, and all reported values and calculations use the original grid data.

Historical comparison. This is calculated separately for each calendar date. For the selected years, the chart shows the median and 10th–90th percentile after applying raw values or a 7- or 15-day trailing mean. Temperature mode also shows the 1991–2020 WMO climatological mean. Trailing means include the plotted date and the preceding 6 or 14 days.

Daily mean anomaly. This is the daily mean minus the mean for the same calendar date in the selected 30-year baseline. The default is 1991–2020; any complete 30-year window from 1940–1969 to 1991–2020 can be selected. Leap day is calculated separately from leap years. The 7- and 15-day curves are trailing means of daily anomalies.

Year-to-date mean anomaly. This is the mean of daily anomalies from 1 January through the plotted date. Rankings use years with at least 90% valid coverage through the same calendar date.

Annual and seasonal mean temperature maps. These maps show the absolute mean temperature for a complete calendar year or season on a fixed, shared colour scale.

Annual and seasonal mean anomalies. Complete annual or seasonal mean temperature is compared with the corresponding mean for the selected baseline. Annual values use calendar years; winter (DJF) is assigned to the year containing January and February; spring (MAM), summer (JJA), and autumn (SON) are assigned to their calendar year. Incomplete years or seasons are excluded.

Standardized mean anomalies. The annual or seasonal mean anomaly is divided by the sample standard deviation (n−1) of the corresponding baseline means and expressed in SD. Maps use the smallest symmetric range among ±3, ±4, ±5, ±6, and ±8 SD that contains the 99th percentile of |z|; because the scale can vary, use the colourbar for numerical comparison. The 1940–1969 DJF baseline contains 29 complete winters because December 1939 is outside the ERA5 archive.

Long-term trends. Trends use complete annual or seasonal means; DJF is assigned to the January–February year and the first available winter is 1941. At least 25 complete years are required. The OLS slope is reported in °C/decade, with Newey–West HAC standard errors using Bartlett weights, L = floor[4(n/100)^(2/9)], and the n/(n−2) correction. The chart shows the fitted trend and pointwise 95% CI; p is the two-sided HAC Wald p-value and R² is the conventional OLS value.

Full-domain area-weighted mean. Grid cells are weighted by physical area. For 2 m air temperature, all valid land and sea cells are weighted by cosine of latitude, so this is not a land-only European mean. SST additionally uses fractional ocean area from the land–sea mask. Calculations use the original grid before interpolation; when no point is selected, derived plots, anomalies, and rankings use this area-weighted series.

Heatmaps. Heatmaps show the full available record at daily or monthly resolution. Daily heatmaps use the daily values directly; monthly values are means of the available daily values in each calendar month. Anomalies are relative to the selected 30-year baseline. Standardized anomalies divide the daily or monthly anomaly by the corresponding sample standard deviation (n−1) within the baseline period. Anomaly colour scales are symmetric around zero and based on the 99th percentile of absolute anomalies, so isolated extremes may exceed the displayed range.