Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

How accurate are satellite ‘thermometers’?


We believe that lower-tropospheric temperatures measured directly by satellites have excellent long-term accuracy, as seen by comparisons with independent atmospheric measurements from weather balloons. Our results contradict indirect measurements by Hurrell and Trenberth1 who claimed that the satellite data have significant discontinuities.


There are two independent direct methods currently used to measure the temperature of the atmospheric layer known as the lower troposphere (surface to 7 km). One method uses balloon-borne instrument packages (radiosondes) that rise through the atmosphere and the other uses satellite-measured intensities of microwave emissions from atmospheric oxygen, which are proportional to temperature (T2R; ref. 2). The latter method relies on polar-orbiting microwave sounding units.

Recently, Hurrell and Trenberth estimated the lower-tropospheric temperature indirectly using variations in sea surface temperatures (SSTs)1. When comparing their SST-based estimates with T2R in the tropics (20° S-20° N), they found two relative temperature jumps, one in September 1981 (0.25°C) and one in late 1991 (0.10°C). They claimed the jumps were due to problems with the satellite measurements.

To check the claim for 1981, we compared T2R against temperatures measured independently at tropical radiosonde stations. We computed the average tropospheric temperature at each location for both radiosondes and satellites for the two years before and after September 1981. Using only four years of radiosonde data assures that instrumental inconsistencies are essentially non-existent. For stations in three latitude bands (±15°, ±20° and ±25° with 14, 30 and 38 locations, respectively) the two-year average temperatures after September 1981 minus those before are −0.04, 0.00 and −0.02 °C, respectively. The same differences for T2R are −0.04, −0.03 and −0.03 °C. These two independent, direct measurements of tropical tropospheric temperatures before and after the alleged jump agree within ±0.03 °C, indicating no “spurious” jump of 0.25 °C. Similar results are obtained for the “spurious” 1991 jump.

If these jumps described by Hurrell and Trenberth are real, trends over the period since 1979 should show clear differences of 0.1 to 0.2 °C per decade between T2R and radiosondes. Trends from several multi-stationradiosonde data sets (one of which used over 300 stations worldwide and three data sets that had no changes in instrumentation) all agreed with T2R to within 0.03 °C per decade, the level of uncertainty we had previously ascribed to the satellite data3,4,5,6.

Hurrell and Trenberth proposed the idea that merging NOAA-07 satellite data into the time series caused the “spurious” jump in late 1981. We disprove this claim by displaying the independent anomalies of tropical T2R from NOAA-06 and NOAA-07 separately (Fig. 1). A displacement of 0.25 °C should be clear, but we found nothingof that magnitude. The agreement between the satellites' anomalies, calculated directly, is 5 to 10 times better than Hurrell and Trenberth's approximation.

Figure 1

Monthly temperature anomalies (°C) for the tropics (20° S-20° N) computed independently from two satellites, NOAA.-06 and NOAA-07, for July 1981 to April 1983.

Additional speculations about the “spurious” jumps (surface emissivity variations or disagreements with the rigid SST/troposphere linkage found in climate models) are not supported by any observational evidence. Other studies have shown that surface emission variations do not significantly affect tropical averages (A. Basist, personal communication) and that multi-decadal variations in the surface/troposphere relationship do occur378.

In summary, claims of large and spurious jumps in the T2R satellite data set are not supported by either of two direct and independent measures of tropospheric temperatures.


  1. 1

    Hurrell, J. W. & Trenberth, K. E. Nature 386, 164–167 (1997).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Spencer, R. W. & Christy, J. R. J. Clim. 5, 858–866 (1992).

    Google Scholar 

  3. 3

    Angell, J. K. J. Clim. 1, 1296–1313 (1988).

    Google Scholar 

  4. 4

    Parker, D. E. et al. Geophys. Res. Lett. 24, 1499–1502 (1997).

    Google Scholar 

  5. 5

    Christy, J. R. Clim. Change 31, 455–474 (1995).

    Google Scholar 

  6. 6

    Christy, J. R., Spencer, R. W. & Lobl, E. J. Clim. (submitted).

  7. 7

    Reid, G. C., Gage, K. S. & McAfee, J. R. J. Geophys. Res. 94, 14705–14716 (1989).

    Google Scholar 

  8. 8

    Ross, R. J. et al. Geophys. Res. Lett. 23, 3179–3182 (1996).

    Google Scholar 

  9. 9

    Waliser, D. E. & Zhou, W. J. Clim. (in the press).

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Christy, J., Spencer, R. & Braswell, W. How accurate are satellite ‘thermometers’?. Nature 389, 342 (1997).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing