Uncertainties in observationally based estimates of temperature change in the free atmosphere

Uncertainties are quantified in atmospheric temperature changes derived fro m satellites, radiosondes, and the reanalyses of the National Center fur En vironmental prediction and European Centre for Medium-Range Weather Forecas ts (NCEP and ERA). To facilitate intercomparison, we compute from the reana lyses and radiosonde data deep layer temperatures equivalent to those estim ated from the satellite-based Microwave Sounding Unit (MSU). Equivalent MSU temperatures generated using global mean weighting functions and a radiati ve transfer code give similar results. NCEP's pre-1979 global mean lower st ratospheric temperature anomalies diverge markedly from radiosonde data. A smaller offset occurs in the midtroposphere. These differences are attribut ed to a likely warm bias in the tropical lower stratosphere in the temperat ure retrievals used by NCEP from November 1978 onward, and changes in the e rror characteristics of the assimilation model's simulation of the lower st ratosphere. In the lower troposphere, ERA and NCEP show different global me an trends due to differences in assimilation strategy, input observational data, quality control procedures, and model physics. Over 1979-1993, ERA wa rms by 0.106 degrees C/decade, while NCEP cools by 0.028 degrees C/ decade. Applying the HadRT1.1 (radiosonde) data availability mask to NCEP improves the agreement between these data sets. Neglecting coverage differences can yield misleading results in MSU-radiosonde trend comparisons. Substantial trend uncertainties also arise from coverage differences between various ra diosonde data sets. Version c of the MSU lower tropospheric temperature ret rieval fails to adjust explicitly fur orbital decay. If this were applied w ithout any additional adjustments, it would resolve an important discrepanc y: in MSUc the lower troposphere has cooled in relation to the midtroposphe re, while the reverse is the case for both reanalyses and for the radiosond e data examined here.