Satellite vs. Ground Measurement of Global Mean Temperature

by Loren Cobb, PhD
March 2008

Beginning in the mid-1990s, a controversy arose concerning the reliability of ground-based measurements of global mean temperature. Skeptics of global warming claimed that ground-based measurements were seriously compromised by the "urban heat-island" effect, and by the over-concentration of measurement locations in the developed world. They made the point that, when compared to satellite-based measurements of the temperature of the lower troposphere (the part of the atmosphere closest to the ground), global warming was visible only in ground-based measurements, and not at all in satellite-based measurements.

Figure 1 (below) shows a representative graph from this era. Note that satellite measurement tools only became available in 1979.

— Figure 1 —

While there were indeed a few measurement problems in both ground-based and satellite measurements, the primary cause of the apparent divergence of the two systems turned out to be random fluctuations — the slopes of the trend lines shown in Figure 1 are very uncertain, due to the small number of data on which they are based. After the passage of ten years and the addition of ten further points to the data, the chart now appears as in Figure 2, below.

— Figure 2 —

Observe that the two trend lines are now almost exactly parallel, and that both display the capacity for large swings (the 1998 satellite-based measurement and the 2007 ground-based measurement are two prominent examples).

Skeptics of global warming have now seized upon the drop in global mean temperature that we have recently registered in both ground- and satellite-based sensors. This decline is visible in Figure 3 below, which shows monthly UAH satellite measurements of global mean temperature. They claim that this decline, now 13 months long, is a "sign" of global cooling. The communications director for the Republican minority on the Senate Environment and Public Works Committee was quoted in the New York Times (2 March 2008) as having headlined his summary of the data as follows: "Earth’s ‘Fever’ Breaks: Global COOLING Currently Under Way".

— Figure 3 —

In fact, as Figure 2 shows, the annual mean temperature rose between 2006 and 2007. The recent drop is not even as big as several others that have occurred since 1979, most notably the event in 1998-99. The overall trend remains unchanged: temperatures rising at a rate of about one-third of a degree every 25 years.


All trendlines in this note were drawn by Microsoft Excel, using a simple linear regression against year on all data. In each case all available data were used, and no observations were trimmed or censored.

Data Sources

UAH satellite data: for the online public data source, click here. The dataset comes from the National Space Science & Technology Center, located at the University of Alabama, Huntsville.

  • Annual global mean temperatures: Column 8, every 12th observation (use December for calendar-year averages).
  • Monthly global mean temperatures: Column 3.

Ground-based data: for the online public data source, click here. The dataset is a collaborative product of the Met Office Hadley Centre and the Climatic Research Unit at the University of East Anglia. The data used in this note came from the Global file for HadCRUT3v (the variance-adjusted version of the basic HADCRUT3 dataset).

  • HadCRUT3v Global mean temperatures: Each year consists of 12 monthly measurements and one annual measurement (14th column). Every other row contains percent coverage information; for example, a coverage of 73 means that the corresponding measurement used input from 73% of the earth's surface. The data come from both land and sea measurements of temperature.

Return to main article