The way the future ocean forcing was applied to the ISMs means that the projections are unlikely to exhibit a non-linear dynamic response over the next century. The majority of the ISMIP6 Greenland ice sheet model projections use a parameterization based on a simple linear function that determines a change in terminus position given subglacial discharge (estimated from modeled surface runoff) and ocean thermal forcing (Slater et al., 2019). The sector-averaged strength of the empirical relationship is used to determine terminus positions of each marine terminating glacier in the future, given projected subglacial discharge and ocean thermal forcing, which are then applied as a mask to the modeled ice sheet extent. As a result of this simple approach, non-linearities in future dynamic behavior, for example due to local glacier and bedrock geometry, are not directly accounted for in the standard Greenland projections. In contrast, (Choi et al., 2021) use a calibrated calving law in their Greenland simulations and find that the regions dominated by marine terminating glaciers in Northern Greenland exhibit a stronger dynamic response over the 21stCentury, compared to the ISMIP6 projections that use the empirical retreat parameterization. We infer from this, and our results that the dynamic response obtained in the ISMIP6 simulations is too weak.
The TSLS derived from the GrIS projections are consistent with the historic data since 1971. The 1901-1970 estimate appears anomalous compared to later periods. We see several possible explanations for this anomaly. The early estimate was derived from trimline elevations associated with the Little Ice Age maximum extent of the ice sheet (Kjeldsen et al., 2015). However, trimlines only provide observational constraints on changes in marginal geometry. In reality it is possible that interior mass gains could have partially offset marginal losses, which seems conceivable considering this pattern is seen in present-day satellite observations (e.g.(Helm et al., 2014)). Further, the Little Ice Age maximum does not have a well defined end date, and period length will influence the estimated rate. One explanation could therefore be that the 1901-1970 rate is biased high because there are no observations that constrain interior mass change, or due to uncertainty in timing. Additionally, the ice sheet responds to local climate change, which in turn is linked to global climate. The ice sheet response at the end of Little Ice Age is linked to a change in regional air temperature (Box, 2013; Box and Colgan, 2013), which is not captured by changes in GMST unless it is a global effect.
4.3 West Antarctic Ice Sheet (WAIS)
The WAIS shows no clear sensitivity to average temperature, which suggests that the marine ice sheet instability is not initiated, or is not of sufficient amplitude, during the 21st century for any of the climate forcing scenarios including SSP5-8.5 (Fig. 4). It is worth noting that this scenario has an averaged global temperature anomaly of +2.6°C over the latter half of the 21stcentury, but which is amplified at high latitudes. The lack of scenario dependence in mass loss through ice dynamics is demonstrated by the subset of ISMIP6 simulations shown in Figure 5. These results also suggest that the timescale for the emergence of dependence on scenario in dynamic processes extends beyond the current century (Lowry et al., 2021).