The reliability of the ground-based and the satellite-based record of global temperature anomalies has been recently discussed (1,2,3). The satellite record is supported by the independent global temperature record from radio sonde balloons (4). There is, however, a fourth independent measure of the global temperature variations, let's call it the Carbon Dioxide Thermometer.

Table 1. gives global data for the period 1979-1998. The anthropogenic carbon emissions, Pg/year [petagrams (1015 grams) per year], originate from combustion of fossil fuels and cement production (5). The global satellite temperature anomaly for the lower troposphere is calculated as the 12-month mean value (6). The global surface-based temperature anomaly (7) is recalculated as the difference from the 1979-1998 mean value in order to help a comparison with the satellite record. The atmospheric carbon dioxide concentration, ppmv [volumetric parts per million], is calculated as the 12-month mean value from measured values at Mauna Loa (8).

The warmest years have, according to the surface record, occured during the last decade, or 1998, 1997, 1995 and 1990. According to the satellite record, the warmest years have been 1998, 1988, 1987, 1980, and 1991.

The atmospheric concentration of carbon dioxide shows statistically a linear increasing tend for the period. There are however large differences in the increase rate during different years ranging from 0.70 ppmv (1993) to 2.88 ppmv (1998). The coefficient of variation is 33.5 %. The coefficient of variation for the emissions from fossil fuel and cement production is, however, only 8.5 %.

As 71 % of the global surface is covered by water, the global mean temperature is dominated by the measured temperatures of the ocean regions. The difference between globally averaged anomaly record from surface stations and satellites is thus primarily due to different measured values for these regions. (9,10).

Due to the fact that the solubility of carbon dioxide is dependent of the water temperature, a warm ocean surface and consequently, a high global surface temperature, should give a high atmospheric concentration and vice versa. As a measure of the yearly temperature anomaly of the global ocean surface water in a situation when there is a steady anthropogenic increase of the atmospheric carbon dioxide concentration, we define The Carbon Dioxide Thermometer (CDT) reading that is simply calculated as the yearly increase rate of atmospheric carbon dioxide minus the average increase rate for the period.

As the CDT is defined only as a deviation from an average trend, that partly can be a result of a warming trend of 0.0188º C/year discovered in the surface record, calculation of the direct correlation of the surface temperature anomaly (column 4) with the CDT may lead to wrong conclusions. We therefore create a surface temperature residual record (column 5) instead by removing the trend from the original record. The correlation of this residual to the CDT is then calculated.

A multiple regression analysis with the emission rate and the surface temperature residual as independent x-variables and the CDT as dependent y-variable gave that both the emission rate and the surface temperature residual were eliminated as statistically insignificant. The correlation coefficient between the surface temperature residual and the CDT is only 0.48 or a little higher than a precariously calculated correlation coefficient of 0.46 between the measured surface temperature anomaly and the CDT.

A statistically non-significant trend of 0.0058º C./year was also removed from the satellite record and the above described regression analysis was repeated with the satellite record used instead of the surface residual record. The same results were however obtained even if the trend was not removed. The emission rate was again eliminated as insignificant, but the satellite record could now explain the CDT reading to a surprisingly high and significant extent. The correlation coefficient was 0.72, the calibration constant of the CDT with respect to the satellite record was 2.28 ppmv/ºC. The statistical significance is p = 0.001.

The correlation coefficient between the satellite and surface records is 0.716. Both records thus seem to describe principally the same global temperature anomaly. The difference from perfect correlation is due to the fact that the numbers divergate for some of the years in the record. The weakness of the correlation between the surface temperature residual and the CDT is mainly due to data for the same years for which the satellite record diverges from the surface record.

The statistical analysis of the CDT readings indicates that the satellite temperature anomaly for the lower troposphere seems to be a better measure of the global surface temperature anomaly than the reported global surface temperature anomaly itself. The finding supports the assumption of most GCM modelers of a strong thermodynamical coupling between the lower troposphere and the backmixed ocean water layer. However, it raises doubts about the reliability of the surface record.

Errors may have been induced in the surface record by the replacement of ship observations by buoy measurements. The distances between the oceanic measuring points may simply be big enough to create errors in the mean value for certain circulation patterns. Errors may also be due to the fact that the number and probably, also quality of the ship measurements, has degraded during the last decade (11).

Table 1. Anthropogenic carbon emissions, temperature anomalies, atmospheric carbon dioxide concentrations and Carbon Dioxide Thermometer readings calculated from refs. 5,6 and 7.

References

1. J. Christy, Climatic Change, 31, 455 (1995)

2. R. Spencer, NASA Space Science News, 14 aug. (1998)

3. F. Wentz and M. Schabel, Nature 394, 661 (1998)

4. World Meteorological Organization, WMO-No. 858 (1997)

5. G. Marland, et al., "Global fossil fuel emissions", The Carbon Dioxide Information Analysis Center (CDIAC), http://cdiac.esd.ornl.gov (1999)

6. J. Christy and R. Spencer, "Monthly means of lower troposphere channel 2LT.d02", NASA, ftp://wind.atmos.uah.edu/msu/t2lt/t2ltglhmam.d03 (1999)

7. P.D.. Jones, et al., "Global temperature anomalies", The Carbon Dioxide Information Analysis Center (CDIAC), http://cdiac.esd.ornl.gov (1999)

8. C.D. Keeling and T.P. Whorf, "Atmospheric CO2 concentrations", The Carbon Dioxide Information Analysis Center (CDIAC), http://cdiac.esd.ornl.gov/ftp/maunaloa-co2/maunaloa.co2 (1999)

9. J. Christy, USGCPR National Seminar, 20. May (1995)

10. J. Christy and R.T. Mc Nider, Nature 367, 325 (1994)

11. M. Stendel, et al. "How representative are recent temperature trends ?"

Max Planck Institute for Meteorology http://www.mpimet.mpg.de/Depts/Modell/TRENDS/ERA­_index.html (1998)