WGI AR5 Fig3-4.jpg
Figure 3.4 (a) The 1955–2005 climatological-mean sea surface salinity (World Ocean Atlas 2009 of Antonov et al., 2010) colour contoured at 0.5 PSS78 intervals (black lines). (b) Annual mean evaporation–precipitation averaged over the period 1950–2000 (NCEP) colour contoured at 0.5 m yr –1 intervals (black lines). (c) The 58-year (2008 minus 1950) sea surface salinity change derived from the linear trend (PSS78), with seasonal and El Niño-Southern Oscillation (ENSO) signals removed (Durack and Wijffels, 2010) colour con- toured at 0.116 PSS78 intervals (black lines). (d) The 30-year (2003–2007 average centred at 2005, minus the 1960–1989 average, centred at 1975) sea surface salinity difference (PSS78) (Hosoda et al., 2009) colour contoured at 0.06 PSS78 intervals (black lines). Contour intervals in (c) and (d) are chosen so that the trends can be easily compared, given the different time intervals in the two analyses. White areas in (c) to (d) are marginal seas where the calculations are not carried out. Regions where the change is not significant at the 99% confidence level are stippled in grey.
WGI AR5 Fig3-5.jpg
Figure 3.5 Zonally integrated freshwater content changes (FWCC; km 3 per degree of latitude) in the upper 500 m over one-degree zonal bands and linear trends (1955–2010) of zonally averaged salinity (PSS78; lower panels) in the upper 500 m of the (a) and (c) Atlantic, (b) and (d) Pacific, (e) and (g) Indian and (f) and (h) World Oceans. The FWCC time period is from 1955 to 2010 (Boyer et al., 2005; blue lines) and 1950 to 2008 (Durack and Wijffels, 2010; red lines). Data are updated from Boyer et al. (2005) and calculations of FWCC are done according to the method of Boyer et al. (2007), using 5-year averages of salinity observations and fitting a linear trend to these averages. Error estimates are 95% confidence intervals. The contour interval of salinity trend in the lower panels is 0.01 PSS78 per decade and dashed contours are 0.005 PSS78 per decade. Red shading indicates values equal to or greater than 0.05 PSS78 per decade and blue shading indicates values equal to or less than –0.005 PSS78 per decade.
WGI AR5 Fig3-9.jpg
Upper 2000 dbar zonally-averaged linear trend (1950 to 2000) (colours with white contours) of salinity changes (column 1, PSS-78 per 50 yr), neutral density changes (column 2, kg m -3 per 50 yr), and potential temperature changes (column 3, °C per 50 yr), for the Atlantic Ocean (ATL) in row 1, Indian Ocean (IND), row 2, Pacific Ocean (PAC), row 3, and global ocean (GLO) in row 4. Mean fields are shown as black lines (salinity: thick black contours 0.5 PSS-78, thin contours 0.25 PSS-78; neutral density: thick black contours 1.0 kg m -3 , thin contours 0.25 kg m -3 ; potential temperature: thick black contours 5.0°C, thin contours 2.5°C). Trends are calculated on pressure surfaces (1 dbar pressure is approximately equal to 1 m in depth). Regions where the resolved linear trend is not significant at the 90% confidence level are stippled in grey. Salinity results are republished from Durack and Wijffels (2010) with the unpublished temperature and density results from that study also presented.

Multi-decadal trends in sea surface salinity have been documented in studies published since AR4 (Boyer et al., 2007; Hosoda et al., 2009; Roemmich and Gilson, 2009; Durack and Wijffels, 2010), confirming the trends reported in AR4 based mainly on Boyer et al. (2005). The spatial pattern of surface salinity change is similar to the distribution of surface salinity itself: salinity tends to increase in regions of high mean salinity, where evaporation exceeds precipitation, and tends to decrease in regions of low mean salinity, where precipitation dominates (Figure 3.4). For example, salinity generally increased in the surface salinity maxima formed in the evaporation-dominated subtropical gyres. The surface salinity minima at subpolar latitudes and the intertropical convergence zones have generally freshened. Interbasin salinity differences are also enhanced: the relatively salty Atlantic has become more saline on average, while the relatively fresh Pacific has become fresher (Figures 3.5 and 3.9). No well-defined trend is found in the subpolar North Atlantic, which is dominated by decadal variability from atmospheric modes like the North Atlantic Oscillation (NAO, Box 2.5). The 50-year salinity trends in Figure 3.4c, both positive and negative, are statistically significant at the 99% level over 43.8% of the global ocean surface (Durack and Wijffels, 2010); trends were less significant over the remainder of the surface. The patterns of salinity change in the complementary Hosoda et al. (2009) study of differences between the periods 1960–1989 and 2003–2007 (Figure 3.4d), using a different methodology, have a point-to-point correlation of 0.64 with the Durack and Wijffels (2010) results, with significant differences only in limited locations such as adjacent to the West Indies, Labrador Sea, and some coastlines (Figure 3.4c and d).

It is very likely that the globally averaged contrast between regions of high and low salinity relative to the global mean salinity has increased. The contrast between high and low salinity regions, averaged over the ocean area south of 70°N, increased by 0.13 [0.08 to 0.17] PSS78 from 1950 to 2008 using the data set of Durack and Wijffels (2010) , and by 0.12 [0.10 to 0.15] PSS78 using the data set of Boyer et al. (2009) with the range reported in brackets signifying a 99% confidence interval (Figure 3.21d).

Community content is available under CC-BY-SA unless otherwise noted.