80
and three quarters of the CMIP5 GCMs exhibit warming that exceeds the highest
plausible value for AAWR that we infer from the climate record. This is rather dis-
concerting, given the prominence of the CMIP5 GCMs in the discussion of climate
policy (e.g., Rogelj et al. 2016 and references therein).
The most likely reason for the shortcoming of CMIP5 GCMs illustrated in Fig. 2.13
is that climate feedback within these models is too large. Although tabulations of λ
from CMIP5 GCMs exist (i.e., Table 9.5 of IPCC 2013 ), comparison to values of λ
found using the EM-GC framework is complicated by the sensitivity of λ to the ΔRF
of climate due to aerosols as well as ocean heat export. Most studies of GCM output
(Shindell et al. 2013 ; Andrews et al. 2012 ; Vial et al. 2013 ) do not examine all three
of these parameters. For meaningful comparison of GCMs to climate feedback from
our simulations, it would be particularly helpful if future GCM tabulations of λ pro-
vided ΔRF due to aerosols and the ocean heat uptake efficiency coefficient (Raper
et al. 2002 ) that best describes the rise ocean heat content within each GCM simula-
tion. While the discussion of Fig. 9.17 of IPCC ( 2013 ) emphasizes good agreement
between the observed rise in ocean heat content (OHC) and the CMIP5 multi-model
mean rise in OHC since the early 1960s, there is an enormous range in the actual
increase of OHC among the 27 CMIP5 GCMs used in their analysis.
Cloud feedback tends to be positive in nearly all GCMs; i.e., simulated changes in
the properties and distribution of clouds tends to amplify ΔRF of climate due to rising
Fig. 2.13 Attributable Anthropogenic Warming Rate from the EM-GC and CMIP5 GCMs.
Diamonds, triangles, and squares show the best estimate of AAWR, 1979–2010, found using ΔT
from the CRU (Jones et al. 2012 ), GISS (Hansen et al. 2010 ), and NCEI (Karl et al. 2015 ) data
centers, for various data records of OHC denoted by color. Error bars on these points represent the
upper and lower limits of AAWR computed based on consideration of 15 possible time series for
ΔRF of aerosols shown in Fig. 2.7a. Values of AAWR over 1979–2010 from the 41 GCMs that
submitted RCP 4.5 simulations to the CMIP5 archive are shown by the box and whisker (BW)
symbol. The middle line of the BW symbol shows the median value of AAWR from the 41 GCMs;
the boxes denote the 25th and 75th percentile of the distribution, and the whiskers show maximum
and minimum values of AAWR. See Methods for further information
2 Forecasting Global Warming