The spread in dynamic response to global mean temperature in the EAIS is
greater than in the WAIS from 2051-2100, especially for the highest
global mean temperatures (Figure 7). However, the ice sheet model
simulations at the high end of the temperature range have all been
forced with a single CMIP6 model, and the spatial pattern of warming can
vary substantially between models. As with WAIS, the EAIS SMB remains
positive (negative sea level contribution) across all simulations driven
by the available CMIP6 models and future scenarios.
For both EAIS and WAIS, the early 21st century time
period tends to produce lower rates of sea level contribution, with a
smaller spread, than the late 21st century time period
for a comparable temperature change. This is indicative of the diverging
responses of the ice sheet model ensemble members over time, as some
continue with a linear trend in the sea level contribution for the
duration of the century, whereas others, particularly at the high tail
end of the distribution have a super-linear response (Seroussi et al.,
2020).
4.5 Glaciers and Ice Caps (GIC)
GIC display a bimodal behavior dependent on the time period (Fig. 8).
The TSLS is higher for the first half of the 21stcentury because, as they melt, the GIC area declines and hence has less
potential to contribute mass to sea-level change. This reduces their
sensitivity to further temperature increases in the second half of the
century. Nonetheless, over a 50 year period a linear sensitivity to
average temperature change is a reasonable approximation, but over
longer time periods the sensitivity changes. This may also explain why
the observational records do not fall on a straight line, although
another reason for that is likely due to the disappearance of small
glaciers during the 20th century and missing glaciers
from the global inventory that are below a minimum size threshold
(Parkes and Marzeion, 2018). The anomalous early 20thcentury records could also be partly due to the post Little Ice Age
response of GIC. The Little Ice Age was predominantly a Northern
Hemisphere signal, which is also where the GIC are mostly situated,
rather than a global mean temperature anomaly. This regional versus
global difference could result in the 1901-1970 sea level contribution
biasing high (see discussion for the GrIS). Post-2100, we would expect
the TSLS to continue to decrease in line with diminishing glacier area.
We note, however, that a recent study using a deep learning approach and
comparing a model with a linear response to climate forcing (as used in,
for example, most GlacierMIP and hence AR6 simulations) with a non
linear version found that linear models tend to overestimate the
response for the high end scenarios in particular but also for glaciers
with a longer response time (Bolibar et al, 2022). This finding implies
that the TSLS is likely too steep when assuming a linear response. If
correct, this would result in the 21st century TSLS
lying closer to the historic value.