The seasonal access of ASW beneath the FIS was first observed by Hattermann et Epacadostat ic50 al. (2012), and a similar seasonality of basal melting seen in the ANN-100 experiment was suggested by coarser model simulations using isopycnal coordinates (Nicholls et al., 2008). While the smoothed topography in ROMS may lead to an overestimate of upper ocean contribution to melting beneath the

FIS, sensitivity studies with the idealized setup of Zhou et al. (2014) suggest that the inflow of ASW is indeed a realistic feature of the simulations. In their experiments with different ice shelf geometries, the amount of ASW entering the cavity is largely independent of the shape of the ice front, and occurs when the wind-driven deepening of the ASW layer outside the cavity exceeds the depth PD-0332991 nmr of the ice draft. Nevertheless, numerical artifacts associated with the terrain following coordinates cannot be ruled out in this setup. Quantifying the exact contribution

of upper ocean Mode 3-type of melting, and scrutinizing its sensitivity to varying forcing, thus remains subject to future work. The idealized simulations of Zhou et al. (2014) also show that the effect of the ASW on the frontal dynamics is a robust result and not an artifact of the hydrographic nudging at the periodic model boundary, a potential criticism in our model. Their annual experiments reproduce a similar deepening of the ASW and a shallower thermocline near the coast, although the ASW is exclusively introduced at the ocean surface. The realism of our simulations is challenged by the simplifications that are necessary to compromise the resolution of mesoscale eddies in a compact periodic domain, the limited amount of data available to construct the model forcing and boundary conditions, and the desire to limit the model’s complexity for the process-oriented sensitivity studies. Time-varying winds (Graham et al., 2013) and the modulating mechanical

effect of sea ice (Nunez-Riboni and Fahrbach, SPTLC1 2009) are likely to modify the short-term and seasonal variability seen in the ANN-100 experiment. The effects of a reduced momentum transfer during the maximum sea ice extent in winter could possibly be inferred from the experiments with different constant wind forcings, but a main challenge will be to better understand the ambiguous role of sea ice during transition between fully ice-covered and open-water conditions (Lüpkes and Birnbaum, 2005). To investigate the effect of time-varying winds, we conducted an additional test run with the 6-hourly RACMO2 wind stress applied. Compared to the constant-wind scenario, this run shows more variability of the coastal current and enhanced deep and shallow melting by about 10 cm year−1 for the entire ice shelf. But this simulation also features more MWDW inside the ice shelf cavity than shown by the observations.