Long-term Climate Projections - Climate Change: Identification and Projections - page 109

Indeed, if we simply multiply the RCPs in Figure 9.4 by the assumed
range of uncertainty (

− 2 − 2

C G , see section 5.4) of
the global sensitivity to radiative forcing S , we obtain - almost exactly -
the projections in Figure 10.1. However, these detailed RCPs help to
describe in the most accurate way the range of phenomena which would
accompany the presupposed global climate evolution: rising oceans, melting
ice caps, desertification, distribution of precipitation, and so on. The IPCC
could talk endlessly on this point, but these simulations are only as valuable
as the a priori knowledge entered into the RCPs, especially the assumed
radiative forcing from the greenhouse effect and aerosols, and the lack of
modeling of solar activity other than the direct energy factor. At the output
of the models, we simply find what had been entered initially 1 .

0 27 ° / Wm < S < 1 62 ° C / Wm

We shall now move on to “free” projections, where neither the models
nor their input data are predetermined according to what message the IPCC
want them to communicate.

10.2. EBM compatible scenarios

EBM simulations are only useful if we are free to carry out projections
based on all of the parameters identified, including the radiative forcing
coefficient α i . To do so, it is fortunately possible to recover the
corresponding concentrations of CO 2 available with the RCP provisions. For
the scenario RCP 8.5, for instance, the final stabilized concentration of CO 2
reaches up to 7.4 times the pre-industrial concentration, hence the log-
concentration line in Figure 10.2 (top graph, where: log(7.4) ~2).

To carry out projections, it is also necessary to anticipate solar activity.
After the peaks of activity reached towards the end of the 20 th Century, solar
physicists expect a return to normal activity before the end of the 21 st
Century (if not a steep decline). Using standard signal processing techniques,
it is easy to extrapolate a narrow-band signal. Therefore, we modeled solar
irradiance via a Markov process, slightly damped around the 11 year period,
and set with a return towards the millennium's average value (100 year time
constant). A Kalman filter carries out a state estimation, and the

1 Garbage input, garbage output.