Partial Results - Climate Change: Identification and Projections - page 79

Parameters identified
Minimizing the output error leads to the following parameter estimations:

S irrnet ( °C/Wm -2 )
Sensitivity to
irradiance

S irr11 ( °C/Wm -2 )
Frequential
sensitivity

S clim (°C)
Climate
Sensitivity

S vol ( °C )
Volcanic
activity

T clim ( an )
Climate time
constant

T E ( °C )
Equilibrium
temperature

1.28 10.7 90.0 0.90 0.027

17.5

N OTE .- It should be recalled that S clim = S 1 log(2). The sensitivity to net
irradiance can be noted equally as S 2 , S irr or S irrnet , and S irr 11 is the sensitivity
at the frequency

f = 1 / 11

(see 4.4).

The climate sensitivity found (1.28 ° C to a doubling of CO 2 ) is compatible
with the IPCC's assumed window (1to6C ). However, the sensitivity to

irradiance calculated (17.7 ° C /Wm -2 ) is ten times higher than the upper limit,
(

°

), which was reported as very likely (i.e. at 90%).

W m −

2

S < 1.6216 C / °

2.

The values identified are in relation to the respective influences of
anthropogenic and natural factors in Figure 7.2.

Static and dynamic sensitivities

The climatic time constant identified is 90 years. Given the thermal
inertia of the oceans, that is understandable. This means that the climate
continues to react to past influences during a significant length of time,
which does not facilitate the visual interpretation of climatic causality. Yet
the model reacts almost instantly to pulse inputs, as is shown by the
response to each volcanic eruption. This behavior characteristic of thermal
processes demands a model structure with several time scales like the one
we have adopted, without limiting ourselves to a simple first order. In
fact, the step response of the model identified provides a similar appearance
to that of Figure 4.2, representative of the GCM simulation. The
difference between this and a simple first order (dotted line) is minor, but it
plays a key role in the immediate reactions of the climate to solar or volcanic
influences.