Average surface air temperature monthly maps (2024)

The maps and graphs in this section are based on pre-release data from ECMWF's ERA-Interim reanalysis and therefore are subject to change should a significant production problem be found to have occurred. The release of the maps and charts usually happens a few days after the last observations of each month have been made but the underlying data products are not madepublicly availableuntil they have been checked.

Standard ERA-Interim values have been adjusted to compensate for two production issues:

1. Values over sea are taken from the background forecast model not the analysis, to avoid a detrimental effect of analysing biased air-temperature observations from ships.
2. Values over sea prior to 2002 are further adjusted by subtracting 0.1°C. This accounts for a change in bias that arose from changing the source of sea-surface temperature analysis.

In addition, the pre-release ERA-Interim data used in the monthly summaries for March and April 2016 were reduced by 0.15°C for the month of March to offset an error in the SST analyses used in the standard production of ERA-Interim. Pre-release values for February 2016 were also affected slightly by this error. Final ERA-Interim data for public release have subsequently been produced by rerunning the period in question using corrected SST data. The final values for February and March 2016 are used in the monthly summaries from May 2016 onwards.

All maps use the same colour scale for departures of temperature from climatological values in the range from -6°C to +6°C. Temperature anomalies larger than this can occur locally, especially at high latitudes in winter associated with anomalous sea-ice extent. The scale used for anomalies that are larger than 6°C in magnitude is variable.

The global averages shown in the graphs are taken over all types of surface. The averages for Europe are taken over all land areas between 20°W and 40°E, and 80°N and 35°N.

From theOctober 2017summary onward, a new software suite is used to create all charts. This leads to minor differences in the displayed values, but does not affect the resulting analysis and statements.

Map demonstrations

About ERA-Interim

The monthly information is fromERA-Interim, a reanalysis for the period from 1979 to the present. ERA-Interim combines information from meteorological observations with background information from a forecast model, using the data assimilation approach developed for numerical weather prediction. The atmospheric observing system underwent several improvements leading up to 1979.

Over land, values of surface air temperature from ERA-Interim are in effect determined quite directly from observational records for regions where plentiful observations of surface air temperature were made. Elsewhere, the background forecast model plays a stronger role, helping values of surface air temperature to be derived from other types of observation, such as of sea-surface temperatures and winds. Satellite data on the extent of sea-ice cover are important in winter, as surface air temperatures tend to be much warmer over open sea than over ice. Observations of conditions higher in the atmosphere provide some additional information.

Differences in surface temperature datasets

Comparing global temperature datasets helps to identify where and when they provide reliable information. It may suggest the causes of differences and the datasets in which more confidence can be placed. An open-access peer-reviewedscientific papercompares the values provided by ERA-Interim up to July 2016 with corresponding values from the Japanese reanalysis JRA-55 and from several widely used datasets that differ in the types of observation used and in the way the observational data are processed. Uncertainties arise from regions where there are few direct temperature measurements, especially over the Arctic and Antarctic where variability from year to year is high. They also arise from the adjustments needed to estimate sea surface temperature frommeasurements made at different depths and with different biases. The comparisons indicate that ERA-Interim is a reasonable choice of prime dataset for regularly monitoring global temperature.

The table below shows monthly global average temperature anomalies (°C) for 2016 relative to 1981-2010, from the latest versions of the various datasets available on 18 January 2017. The large spread in values later in the year arises from differences in the extent to which the datasets represent the anomalously warm conditions at high latitudes associated with exceptionally low sea-ice cover in both the Arctic and the Antarctic. Consistent with the published comparison up to July 2016, the spatially more-complete GISTEMP and Had4_UAH_v2 datasets agree much better with ERA-Interim and JRA-55 in October and November than NOAAGlobalTemp and HadCRUT4 do.

The following table shows corresponding annual averages from 2005 onwards, and the spread in these values. The spread varies from 0.03°C in 2013 and 2014 to 0.13°C in 2005 and 2016. Its average over the period 1979-2016 is 0.06°C. The increase in average temperature from 2015 to 2016 ranges from close to 0.2°C for the reanalyses to close to zero for HadCRUT4.

Change over the industrial era

The Paris Agreement established the aim of "holding the increase in the global average temperature to well below 2°C above pre-industrial levels and pursuing efforts to limit the temperature increase to 1.5°C", but did not specify precisely what was meant by "pre-industrial levels". This has now been assessed in ascientific paperindependent of C3S that proposes a baseline of 1720-1800 for estimating subsequent change over the industrial era. The paper also estimates that the atmosphere from 1986-2005 was on average between 0.55 and 0.80°C warmer than it was over the baseline period.

The average ERA-Interim temperature for 1981-2010 is almost the same as that for 1986-2005. The 2016 ERA-Interim anomaly of 0.62°C relative to 1981-2010 thus translates to a temperature increase of 1.3°C above the 1720-1800 pre-industrial level, with an uncertainty of more than ±0.1°C. The same is found for the JRA-55 reanalysis. Using the coldest of the estimates for 2016, that of HadCRUT4, the median rise above the 1720-1800 level is 1.2°C for 2016. Choosing instead the first fifty years for which HadCRUT4 provides values, 1850-1899, as the "pre-industrial" reference period gives a median estimate for 2016 that is 1.1°C above this "pre-industrial" level. Corresponding values for the NOAAGlobalTemp and GISTEMP datasets are 1.1°C and 1.2°C respectively, when the "pre-industrial" reference period is taken to be 1880-1899, the first twenty years for which they provide data.

Central estimates of the temperature increase of the year 2016 over the industrial era thus vary from 1.1°C to 1.3°C for the datasets considered, with further uncertainty of more than ±0.1°C.

Length of data record

The ERA-Interim reanalysis starts in 1979, so statements made in the monthly temperature summaries relating to “the warmest month on record” and so forth formally refer to the period from 1979 onwards. In practice, most such statements about warm months and years are very likely to be valid for the whole of the industrial era, especially for global averages and annual European averages. This is based on ourexaminationof the pre- and post-1979 values provided by the well-established conventional datasets, supplemented by theestimateby others that the temperature for the period 1720-1800 was if anything colder than that for the latter half of the 19thcentury.

For global-mean temperature, no month prior to 1979 in the HadCRUT4 dataset has an anomaly as large as 0.1°C relative to 1981-2010, whereas record warm anomalies range from 0.4°C for May to 0.8°C for December. For ERA-Interim, the range of record anomalies is from 0.4°C for June to 0.9°C for February. All record warm months occur during or after 2015 in both datasets. These recent record values exceed pre-1979 record values by much more than the estimated uncertainties in the data values. The NOAAGlobalTemp and GISTEMP datasets tend to have colder values relative to 1981-2010 than HadCRUT4 prior to 1979.

Record annual anomalies in average European temperatures of around 1.2°C occurred in 2014 and 2015. These years were about 0.9°C warmer than 1934, the warmest year prior to 1979 according to values from the GISTEMP, HadCRUT4 and NOAAGlobalTemp datasets. The datasets differ by less than 0.1°C in their estimates of average European temperatures for each of these three years.

The length of data record becomes a limitation for ERA-Interim when it comes to ranking European temperatures for specific months of the year. In HadCRUT4, the warmest months over Europe all occur after 1989, but the years with the second warmest January, October and November are 1975, 1967 and 1938 respectively. The differences in temperature between the warmest and second-warmest months are also small (0.03°C and 0.12°C respectively) for October and November.