Wind stress fields are available from reanalyses, satellite-based data sets, and in situ observations. Basin scale wind stress trends at decadal to centennial time scales have been reported for the Southern Ocean, the North Atlantic and the Tropical Pacific as detailed below. However, these results are based largely on atmospheric reanalyses, in some cases a single product, and consequently the confidence level is low to medium depending on region and time scale considered.
In the Southern Ocean, the majority of reanalyses in the most comprehensive study available show an increase in the annual mean zonal wind stress (Swart and Fyfe, 2012; Figure 3.8). They find an increase in annual mean wind stress strength in four (NCEP1, NCEP2, ERAI and 20CRv2) of the six reanalyses considered (Figure 3.8). The mean of all reanalyses available at a given time (Figure 3.8, black line) also shows an upward trend from about 0.15 N m–2 in the early 1950s to 0.20 N m–2 in the early 2010s. An earlier study, covering 1979–2009, found a wind stress increase in two of four reanalyses considered (Xue et al., 2010). A positive trend of zonal wind stress from 1980 to 2000 was also reported by Yang et al. (2007) using a single reanalysis (ERA40) and found to be consistent with increases in wind speed observations made on Macquarie Island (54.5°S, 158.9°E) and by the SSM/I satellite (data from 1987 onwards). The wind stress strengthening is found by Yang et al. (2007) to have a seasonal dependence, with strongest trends in January, and has been linked by them to changes in the Southern Annular Mode (SAM, Box 2.5), which has continued to show an upward trend since AR4 (Section 2.7.8). Taken as a whole, these studies provide medium confidence that Southern Ocean wind stress has strengthened since the early 1980s. A strengthening of the related wind speed field in the Southern Ocean, consistent with the increasing trend in the SAM, has also been noted in Section 2.7.2 from satellite- based analyses and atmospheric reanalyses.
In the Tropical Pacific, a reanalysis based study found a strengthening of the trade wind associated wind stress for 1990–2009, but for the earlier period 1959–1989 there is no clear trend (Merrifield, 2011). Strengthening of the related Tropical Pacific Ocean wind speed field in recent decades is evident in reanalysis and satellite based data sets. Taken together with evidence for rates of sea level rise in the western Pacific larger than the global mean (Section 3.7.3) these studies provide medium confidence that Tropical Pacific wind stress has increased since 1990. This increase may be related to the Pacific Decadal Oscillation (Merrifield et al., 2012). At centennial time scales, attempts have been made to reconstruct the wind stress field in the Tropical Pacific by making use of the relationship between wind stress and SLP in combination with historic SLP data. Vecchi et al. (2006), using this approach, found a reduction of 7% in zonal mean wind stress across the Equatorial Pacific from the 1860s to the 1990s and related it to a possible weakening of the tropical Walker circulation. Observations discussed in Section 2.7.5 indicate that this weakening has largely been offset by a stronger Walker circulation since the 1990s.
Changes in winter season wind stress curl over the North Atlantic from 1950 to early 2000s from NCEP1 and ERA40 have leading modes that are highly correlated with the NAO and East Atlantic circulation patterns; each of these patterns demonstrates a trend towards more positive index values superimposed on pronounced decadal variability over the period from the early 1960s to the late 1990s (Sugimoto and Hanawa, 2010). Wu et al. (2012) find a poleward shift over the past century of the zero wind stress curl line by 2.5° [1.5° to 3.5°] in the North Atlantic and 3.0° [1.6° to 4.4°] in the North Pacific from 20CRv2. Confidence in these results is low as they are based on a single product, 20CRv2 (the only century time scale reanalysis), which may be affected by temporal inhomogeneity in the number of observations assimilated (Krueger et al., 2013).