Reconciling Surface and Satellite Temperature Measurements

Considerable debate exists regarding the ability of both surface (SFC) and Microwave Sounding Unit channel 2 (MSU2) satellite observations to accurately monitor the global variations in atmospheric temperature and to detect long-term trends. In contrast to the surface record, the MSU channel 2 measurements are sensitive to the temperature averaged over a layer of the atmosphere several kilometers thick, centered in the lower troposphere. This difference in vertical sampling is a key source of uncertainty that has complicated previous comparisons of these measurements. In an attempt to understand the differences between SFC and MSU2 global temperature trends, both records are compared against an objectively-analyzed global radiosonde network (RAOB). The radiosonde measurements offer the unique ability to resolve the vertical structure of the atmospheric temperature variability and are therefore directly comparable with both the deep-layer satellite measurements and the near-surface air temperature records. An initial comparison for the period of 1979-1994 indicates that the global-mean MSU2 temperature anomalies agree remarkably well with the corresponding channel 2 temperatures computed from the RAOB network. The high level of consistency between the two records (~0.03 K rms difference) provides confidence in both records. Likewise, the global-mean SFC temperature anomalies also agree well with temperature anomalies derived from the 1000 hPa level of the RAOB analysis.

Thus, the RAOB temperature record was consistent with both the MSU2 and SFC temperature records over this period. However, the SFC and MSU2 temperature anomalies differ from each other by more than 0.25 K at two periods in the record (near 1980 and 1993), resulting in long-term trends which differ by more than an order of magnitude. Moreover, these differences in global temperature anomalies are also noted between the RAOB-calculated channel 2 temperature anomalies and near surface (1000 hPa) temperature anomalies, suggesting that the largest discrepancies in between the MSU2 and SFC temperature anomalies results from differences in the vertical sampling of the instruments and not from measurement or sampling errors within the individual temperature records themselves. Finally, the observed SFC, RAOB and MSU2 temperature records are compared to those obtained from a GCM forced with observed SSTs. It is shown that the GCM simulates a stronger coupling between the SFC and MSU2 records than is observed. The implications of this comparison for monitoring and prediction of global warming will be presented at the conference.