Short-term Predictions - Climate Change: Identification and Projections - page 115

11

Short-term Predictions

11.1. Decadal time scale predictions by GCM

Figure 11.1 below reproduces Figure 11.9 from the AR5. It gives a
summary of the results from the CMIP5, and is based on the scenario
RCP 4.5, which is less extreme than the RCP 8.5. Since we did not
participate in the project, we are not aware of the detail of the protocols for
the development of models, and the exact nature of the constraints to which
the participants were subjected. The striking similarity between the
simulations from the 42 models leads us to believe that these constraints are
strict, starting with the obligatory use of the RCPs, which leaves no freedom
in the evaluation of the radiative forcing coefficients - or the equivalent - in
the GCMs.

The thin lines in graph (a), as well as the uniform shaded areas in (b)
come from simulations initialized in 1986. From 1986 to 2005, they result
from the so-called historic forcing, actually coming from coefficients of
radiative forcing assumed by the IPCC, and after 2006, from the scenario
RCP 4.5, based on the same assumed coefficients. The spread of the results
is partly due to the nature of GCMs themselves, which, like numerical wind-
tunnels, “naturally” simulate atmospheric or oceanic turbulence. These
models are not explicitly based on a specific coefficient of climate reaction
λ

G , but the weak spread amongst the simulated responses, to an imposed
scenario of radiative forcing (RCP 4.5), suggests that the effective global
feedback varies little from one model to the next.

The average of the simulations is indicated by the white median line in
graph (b). The (perhaps overestimated) effect of radiative forcing caused by