I was alerted to a new paper on Arctic climate (h/t Erik). The paper is
Brönnimann, et al 2012 : A multi-data set comparison of the vertical structure of temperature variability and change over the Arctic during the past 100 years . Climate Dynamics (in press)
The abstract reads [highlight added]
We compare the daily, interannual, and decadal variability and trends in the thermal structure of the Arctic troposphere using eight observation-based, vertically resolved data sets, four of which have data prior to 1948. Comparisons on the daily scale between historical reanalysis data and historical upper-air observations were performed for Svalbard for the cold winters 1911/1912 and 1988/89, the warm winters 1944/1945 and 2005/2006, and the International Geophysical Year 1957/58. Excellent agreement is found at mid-tropospheric levels. Near the ground and at the tropopause level, however, systematic differences are identified. On the interannual time scale, the correlations between all data sets are high, but there are systematic biases in terms of absolute values as well as discrepancies in the magnitude of the variability. The causes of these differences are discussed. While none of the data sets individually may be suitable for trend analysis, consistent features can be identified from analyzing all data sets together. To illustrate this, we examine trends and 20-yr averages for those regions and seasons that exhibit large sea-ice changes and have enough data for comparison. In the summertime Pacific Arctic and the autumn eastern Canadian Arctic, the lower tropospheric temperature anomalies for the recent two decades are higher than in any previous 20-yr period. In contrast, mid-tropospheric temperatures of the European Arctic in the wintertime of the 1920s and 1930s may have reached values as high as those of the late 20th and early 21st centuries.
Interesting excerpts are
We …. analyse trends and mean values over longer time periods. We first return to the long record of 850 hPa temperature in winter over Svalbard (Fig. 1). Although interannual variability was relatively similar comparing the datasets, there are substantial differences even in the trend of the relatively recent 1980-2002 period from ERA40 (0.85 °C/decade), NNR (0.57°C/decade), JRA25 (0.49 °C/decade), observations from Ny Ålesund and Barentsburg merged(0.35 °C/decade), and 20CR (0.19 °C/decade). These large discrepancies among the data sets underscore the large uncertainties involved with estimates of the trend.
A common feature seen in Figs. 9 and 10 is that the tropospheric warming is especially strong in the 1978-1997 period. The Canadian Arctic (Fig. 12) shows the strongest warming for the most recent period. A further common feature is the cooling trend throughout the troposphere in 1948-1967 in almost seasons and regions (except in 20CR for the Canadian Arctic (Fig. 12)).
Concerning the vertical structure, almost all recent warming trends (1978-1997 and 1988-2007),with the most notable exception of the summer trend in the Pacific Arctic, are stronger near the ground than at 700 hPa.
This is an informative analysis. The lower trends higher in the troposphere, for example, needs an explanation including an assessment as to how well the multi-decadal climate models are at predicting this difference. In addition, the skill at the models at predicting the spatial variations in the vertical structure of the atmosphere needs to be assessed. This is an excellent paper to use in such a model evaluation.