Arctic tropospheric warming amplification?

Relative rates of temperature change between the troposphere and surface, and the mechanisms that produce these changes, have long been a contentious issue. Graversen et al., predicated upon the ERA-40 reanalysis, report polar tropospheric amplification of surface warming and attempt to explain this finding dynamically. Here we show (1) that data from satellites and weather balloons indicate that the ERA-40 trends are increasingly unrealistic polewards of 62uN; (2) that the two other reanalyses considered exhibit very different polar trends; and (3) that the vertical profile of polar trends in ERA-40 is unrealistic, particularly above the troposphere. These quasi-independent strands of evidence imply that the pattern of warming in the Arctic troposphere is highly unlikely to be as given in ERA-40 and as reported by Graversen et al.. Reanalyses are numerical weather-prediction systems run in hindcast mode considering all globally available observations. Strenuous efforts are made to take account of both time-varying biases in the data and the impacts of the very substantially changing mix and coverage of observations. However, many aspects of the long-term behaviour of reanalyses remain unreliable and their suitability for use in monitoring atmospheric temperature trends has been questioned by a recent expert panel. Comparing ERA-40 with several observational ‘lower tropospheric’ retrievals (corresponding most closely with the original analysis, peaking at about 725 hPa) over the 62.5uN to 82.5uN latitude range (Fig. 1, left-hand panels) yields good month-to-month