Global temperatures

Please note, this page has been archived as of Feb 2011 and will not be updated. 

There are four principal global temperature datasets. Three use surface temperature measurements taken at land and sea, and record a clear warming trend over the past century. One is collected by the Met Office Hadley Centre jointly with the Climatic Research Unit (HadCRU) in the United Kingdom. Two datasets are collected by organisations in the United States: the NASA's Goddard Institute of Space Studies (GISS) and NOAA's National Climatic Data Center (NCDC).

The final dataset uses satellites to measure the troposphere - the lower part of the Earth's Atmosphere. It is collected by the University of Alabama in Huntsville (UAH), started in 1978, and also records a clear warming trend.

The World Meteorological Organisation (WMO) of the United Nations uses the datasets kept by GISS, NCDC and HadCRU to calculate a single world average. These datasets are also quoted in the IPCC's Fourth Assessment Report (AR4).

Each year GISS, HadCRU and NCDC each announce a change in the average global temperature. Each organisation uses subtly different methodologies to analyse data, which means this announced statistic can differ. For example, GISS ranks 2005 and 2010 as the joint hottest years but HadCRU classifies 1998 as the warmest. The three datasets record an increase in temperatures over the past century of between about 0.6 and 0.8 degrees celcius.


Surface temperatures on land and at sea have been collated since 1880 by HadCRU, GISS and NCDC. For the data used in the most recent IPCC report, HadCRU used temperatures from 4,349 measuring stations, while NCDC and GISS used more than 7,200. Papers by GISS's Hansen and Lebedeff and HadCRU's Phil Jones et al provide blueprints for analysing data.

The basic technique for deriving temperatures involves mapping the world into 5x5 degree grid boxes. Data from temperature stations within the box is used to calculate an average for the area.

However, stations are not evenly distributed around the world. The Arctic, the Antarctic and the African interior have very few stations and many grid boxes in these regions have no stations at all. The different datasets deal with the lack of data from these remote regions differently. GISS assumes that regions with no stations have the same temperatures as the areas closest by that do have stations. HadCRU classifies remote areas with no stations as missing data. The NCDC extrapolates data, but not to areas in the Arctic with sea ice.

The differences in GISS and HadCRU's published datasets as a consequence of their extrapolation techniques is illustrated below:

Temp -datasets


GISS on the left, HadCRU on the right. Source: NASA Earth Observatory

The GISS methodology is more sensitive to unusual conditions in areas with few temperature stations. Because HadCRU does not extrapolate data to areas where there are no stations it can underestimate the warming effect of climate change in the Arctic, for example.

The rate of Arctic sea ice loss suggests temperatures are higher then currently stated in the HadCRU data. Because temperatures were cool in the Arctic during 1998 the GISS value was lower than that given by HadCRU. During 2005 and 2010 the Arctic experienced warmer conditions and thus GISS gave a higher global temperature than HadCRU. This probably explains why HadCRU states that 1998 is the warmest year on record while GISS states 2005 and 2010 are the warmest years.

Temperature Anomalies

Scientists each year announce the change in global temperature compared to a historic long term average. They describe this as the "temperature anomaly" from a "base" period. We cannot calculate an accurate absolute average temperature. This is because, among many reasons, temperature stations are placed at different altitudes and record data at different times, while some are placed in urban and others in rural areas. Using anomalies allows scientists to extrapolate data across geographical regions more accurately.

GISS, HadCRU and the NCDC each use different base periods: GISS uses 1951-80, HadCRU uses 1961-90 and NCDC uses the whole of the 20th Century. This means that each organisation's annual figures are different.

However, the annual changes in temperature recorded by the different organisations are very similar.

How important is one year?

The difference between the 'warmest' or 'second warmest' year is often just a few hundredths of a degree. Meanwhile, year on year temperatures can fluctuate because of specific weather events (like El Nino) and natural climate variability. Because of this, scientists look at longer periods of time to track changes in the climate.

James Hansen, the director of GISS, has said: "It's not particularly important whether 2010, 2005, or 1998 was the hottest year on record… It is the underlying trend that is important."

Temp -datasets -hadcru -profile

Long term temperature trends. Source: Nasa Earth Observatory

A glance at the long-term temperature trends shows a clear agreement between the datasets, with the warming trend also apparent. Reto Ruedy, also of GISS, has said: "Official records vary slightly because of subtle differences in the way we analyze the data, but they agree extraordinarily well."

So despite their methodological differences, data from the world's most respected temperature monitoring organisations shows, with surprising harmony, that the world is warming.


Hansen, J.E., and S. Lebedeff, 1987: Global trends of measured surface air temperature. J. Geophys. Res., 92, 13345-13372, doi:10.1029/JD092iD11p13345.

Jones, P.D., New, M., Parker, D.E., Martin, S. and Rigor, I.G., 1999: Surface air temperature and its variations over the last 150 years. Reviews of Geophysics 37, 173-199.

IPCC AR4 Working Group I Chapter 3.2.2 - Temperature in the Instrumental Record for Land and Oceans

NASA Press release - NASA Research Finds 2010 Tied for Warmest Year on Record

NASA Earth Observatory - Different Records, Same Warming Trend

Please note - This material was last updated on 10/3/11 and has since been archived.