4. The combined discharge rates

are shown in Fig. 5. An accumulation-balancing rate of 107 Gt/yr is given by Rignot et al. (2008). The effect of increased snow accumulation on Antarctica during the immediate future (as indicated by observations Church et al., 2013) would mean a larger potential value for D. Measurements from Rignot and Kanagaratnam, 2006 and Rignot et al., 2008 are shown as well in Fig. 5. More recent overviews ( Shepherd and Wingham, 2007 and Shepherd et al., 2012) show considerable variation in the Greenland and Antarctic mass balance measurements. Because the sampling was performed during different periods and does not include all ice sheets, we have left these from further consideration. The progression of D   in Fig. 4 shows the collapse of the West-Antarctic

ice sheet. The discharge rate Adriamycin increases dramatically with this event. With the ice sheet gone, calved icebergs drift more easily. We expect basal melt to decrease then. On the other SCR7 datasheet hand, more land ice is in contact with the ocean, which should increase the absolute amount of melt taking place. Without any way of quantifying either effect, we suggest that after a collapse event the basal melt amount returns to pre-collapse levels. The expression becomes equation(14) Nsi(t)=μi·Dsi(t)t⩽30μi·Dsi(30)t>30Gt/yrfor the WAIS (region i), where μW=0.30μW=0.30. Similar considerations to those above lead us to keep the amount of basal melt steady at the 2030 levels for the other two regions, which then give the exact same form as Eq. (14) with the appropriate μμ values ( Table 2). Far deposition is allocated to all mass loss not already claimed by basal melt. The expression for Antarctic

F   is then simply equation(15) Fs(t)=(1-μs)·Ds(t)t⩽30Ds(t)-μs·Ds(30)t>30Gt/yr.for all three regions with μsμs replaced by the appropriate basal melt fraction and rsrs the corresponding discharge rate. Table 4 gives a summary of the melt scenario features on which our projections are based. In Table 5 a break-down of mass loss expressed as sea-level equivalent is given. We can compare with some other severe scenarios, see Fig. 6. The most recent scenarios are by Pfeffer et al., 2008 and Katsman et al., 2011. A projection close to Palmatine the values given by Pfeffer et al. (2008) as upper bounds would tax the rate of retreat of the tidewater glacier to nonphysical limits. The lower bound from Fettweis et al. (2013) only takes meltwater into account. The projections for ice discharge dominate this by an order of magnitude. To illustrate the effect of the freshwater protocol outlined above, we ran a RCP8.5 experiment with the CCM EC-Earth (Hazeleger et al., 2010). One simulation was run without the extra freshwater forcing applied (control) and one with additional freshwater forcing included (forced) to allow for a sensitivity experiment. The control run is part of the CMIP5 archive and both runs use the RCP8.