The linear trends for a number of fields obtained from the reanalyses of the National Centers for Environmental Prediction – National Center for Atmospheric Research (NCEP-NCAR) are calculated for the Northern Hemisphere winter months (January, February, March) from the 55-year period 1948-2002. The fields include sea-level pressure (SLP), geopotential height at 500 and 50 hPa, temperature at 500 and 50 hPa, zonally averaged height, temperatures, zonal, meridional and vertical velocities from 1000 to 50 hPa, and surface air temperature (SAT) from Jones (1994). The trend fields are expressed in terms of two alternate expansions: (a) into contributions from the Arctic Oscillation (AO) and Cold Ocean – Warm Land (COWL) patterns, as defined from the leading modes of an empirical orthogonal function (EOF) analysis of sea level pressure; or (b) into contributions from the modified Arctic Oscillation (AO*) and modified COWL (COWL*) patterns, defined from the leading EOFs of 500 hPa height. The residuals in each expansion are considered, and the completeness properties of the expansions discussed.

The AO contribution explains most of the SLP falls over the Arctic and half the SLP rise over the North Atlantic; the COWL pattern explains the entire negative pressure trend over the Pacific and half the rise over the Atlantic. In the expansion into AO* and COWL* patterns the latter explains most of the SLP trend. Similar remarks hold for the height trend at 500 hPa. In each case the residual is a small fraction of the trend. The observed SAT trend (warming over North America and Asia, cooling over northeast Canada and the Pacific) is partitioned nearly equally between contributions from the AO and COWL, although the COWL contribution dominates over North America. In the alternate expansion, the COWL* explains nearly all the warming over North America and Asia. The mid-tropospheric (500 hPa) temperature trend is mostly due to the COWL (or COWL*) patterns, with the AO explaining only the local cooling over Greenland. In the lower stratosphere, the 50 hPa height and temperature trends are not well explained by either set of patterns.

The AO pattern contribution to the zonal mean height trend shows large height falls (modest height rises) in high (mid)-latitudes, which resemble the overall trend except for a significant poleward shift. The residual is therefore a large fraction of the trend. In the AO* / COWL* expansion, the COWL* contribution is much stronger than that due to the AO*, although again shifted. Very similar results are obtained for the zonal average temperature and zonal average circulation trends. However, the strong warming seen in the high latitude stratosphere and near the ground are completely missing in the AO, and only hinted at in the COWL and COWL* contributions.