Each experiment is integrated for five model years with the respective forcing fields applied. Some of these runs approach a new steady state, whereas other simulations—particularly those exhibiting strong inflow of warm water beneath the ice—do not reach a new equilibrium. We chose not to integrate the model for longer time because the ongoing trends in these runs are clear and because the Epacadostat cost applied forcing is relatively extreme in these scenarios and does not represent typical conditions at the present time. We assess the realism of our simulations by comparing the recent observations

below the FIS with synthetic mooring data from the most realistic ANN-100 experiment. Together with other parameters presented later, Fig. 5

shows a time series of simulated temperatures (Fig. 5(a)), interpolated at locations of the upper and lower sensors of M1 and M3, covering the five model years of the ANN-100 experiment and the last six months of the initialization simulation. For comparison, the temperature GSK J4 cell line axes in Fig. 5(a) and Fig. 4(b) are equal. In general, the model shows predominantly low ice shelf cavity temperatures and warmer events due to the intermittent access of ASW and MWDW, yielding a sub-ice shelf water mass distribution that resembles the observations. This can be seen from the θθ–S histograms in Fig. 6, presenting the frequency of occurrence of different water masses at M1 and M2 in the different model experiments. The color shading uses the same scale as for the observations in Fig. 3(b), which for comparison are overlaid as black contours, showing most similarity with the ANN-100 experiment in Fig. 6(b). The model reproduces warm pulses of MWDW at the lower sensor of M1 (red curve in Fig. 5(a)), eltoprazine with similar characteristics as observed by the actual M1 mooring in Fig. 4(b). A wavelet analysis of the synthetic mooring time series (not shown) reveals a similar frequency distribution and intensity of the episodes of increased

current variability, contemporaneous with warm pulses of deep water, in agreement with the pattern described for the observations in Section 2.4. However, with a strictly periodic seasonal forcing applied, the model shows a regular inflow of MWDW at M1 during late winter and spring, while the two available years of observations suggest a greater inter-annual variability for the warm pulses at depth. Also the seasonal access of ASW beneath the FIS is reproduced by the model. This is shown by higher temperatures in the period between January and July at the upper sensors of M1 and M3 (blue curves), while temperatures below the surface freezing point indicate the presence of ISW during the rest of the year.