Page 8

1979. The satellite measurements appear to be substantiated by independent trend estimates for this period based on radiosonde data. Some have interpreted this apparent discrepancy between surface and upper air observations as casting doubt on the overall reliability of the surface temperature record,4 whereas others have concluded that the satellite data (or the algorithms that are being used to convert them into temperatures) must be erroneous. It is also conceivable that temperatures at the earth's surface and aloft have not tracked each other perfectly because they have responded differently to natural and/or human-induced climate forcing during this particular 20-year period. Whether these differing temperature trends can be reconciled has implications for assessing:

• how much the earth has warmed during the past few decades,

• whether observed changes are in accord with the predicted response to the buildup of greenhouse gases in the atmosphere based on model simulations, and

• whether the existing atmospheric observing system is adequate for the purposes of monitoring global-mean temperature.

This report reassesses the apparent differences between the temperature changes recorded by satellites and the surface thermometer network on the basis of the latest available information. It also offers an informed opinion as to how the different temperature records should be interpreted, and recommends actions designed to reduce the remaining uncertainties in these measurements.break

4 Unless specified otherwise, the "surface record" referred to in this report is a combination of the temperature of sea surface water and the temperature of surface air overland.