(2012). We expect the additional freshwater to immediately affect local sea-surface height and through barotropic effects to propagate information throughout the world ocean (Stammer et al., 2011 and Lorbacher

et al.). The freshwater might also affect ocean currents. In the forced run the North Atlantic sub-polar gyre remains weakly affected for a considerable time. It is not until 2075 that the mean sea-level rise is comparable to the local rise in the gyre (not shown). The reason for this is that most of added the freshwater is taken away by boundary currents in the Northern Hemisphere. The same can be seen in other experiments of comparable resolution with Greenland freshwater release find more like (Stammer et al., 2011, Kopp et al., 2010, Weijer et al. and Swingedouw et al., 2013). A climate model is a chaotic system and shows sensitivity to small variations in initial conditions. SGI-1776 cost An ensemble of runs can bring out the so called internal variability. We have used such an ensemble of control runs to determine the variance in the SSH. In Fig. 7 the areas where the rise does not exceed 2σσ are

mapped onto the eustatic sea-level, where the whitepoint is centred. The model allows for a free-surface adjustment which shows an increase of SSH with the addition of more freshwater as can be seen in the lower panel. The response to the freshwater forcing is largely advective with the mean subpolar gyre circulation transporting the melt water southward. This can be seen by the comma-shaped feature present in both panels and lying more to the east in the lower one. To the west and south of the sub-polar gyre the sea-surface anomaly is larger than within the gyre, or to the north. The west-to-east gradient in the North Atlantic with a strong anomaly along the northeast coast of North America, as noted in Kopp et al. (2010), can also be seen in the top panel of Fig. 7. The lower panel, which depicts the situation for the last five years of the century, shows an opposite pattern. Here, a positive anomaly on the eastern side of the Atlantic basin can be seen. The formation/inversion of this pattern is also

present in the atmosphere-coupled run discussed in Stammer et al. (2011). A strong signal develops along the American coast and a signal similar to the one in the lower panel of Fig. 7 can be seen after four decades (see also Swingedouw et al., 2013 for a comparison isometheptene between several models showing a similar pattern). The additional freshwater does not impact the Atlantic meridional overturning. In Fig. 8 the annual mean of its maximum value is shown for the RCP8.5 only run (green) and with the freshwater added (blue). The difference (red) indicates little difference between the two. The maximum mixed layer depth (not shown) shows some decrease in the Labrador region and an increase north of Iceland, but this effect is highly variable. We surmise that most of the freshwater does not reach the convection regions and has little impact on dense-water formation.