The model projections appear vertically offset from the relationship
indicated by the observational data with AR6 being about 1 mm/yr below
the empirical relationship and the ensemble falling ~2
mm/yr lower (Fig. 9). This suggests that models are biased low by a
constant drift. This is not surprising as the emulated land ice
contributions have not been adjusted for drift. The emulated ice sheet
contributions are based on the ISMIP6 which by design results in zero
trend for present-day temperatures. In addition, here we are not
including small quasi-constant components of the sea level budget such
as the effect of glacial isostatic adjustment, the deep steric term or
land water storage. These terms, while important for closing the sea
level budget, have a negligible effect on the TSLS.
The intercept of the GMSL TSLS relationship on the x-axis we term the
balance temperature and represents the value at which SLR is zero. This
value is important for considering attribution of SLR. A reconstruction
of sea level over the last 2500 years shows variations around a mean
close to zero up until the start of the 20th century
(Kopp et al, 2016) and appears closely tied to global temperature
anomalies. Our results are consistent with this, indicating that the
balance temperature of -1.1±0.1˚C is equivalent to pre-industrial
temperature. In other words, SLR since the start of the
20th century can be entirely attributed to
anthropogenic global warming.
4.7 Summary of sensitivities
We summarize the range of estimated TSLS for every model of the major
contributors in Fig. 10. Most contributors to sea level rise appear to
have a near time-invariant TSLS with GIC being a notable exception. This
is in accordance with expectations, as the ice sheets and ocean heat
content have multi-centennial response times, and the response on the
time scales considered in this paper can therefore be considered
transient. Glaciers, however, have a much shorter response time, and
thus GIC has started to equilibrate to the new warmer climate by the end
of a century, and we see that as a reduced TSLS in the late
21st century. This reduces the universality of the
TSLS metric for the GIC contribution over a century time scale.