Introduction - Climate Change: Identification and Projections - page 8

Method). Given the nature of the data, the model is not statistically optimal,
but there is no reason to assume that it is far from it. Moreover, as the
uncertainty calculation does not result from an estimate of the maximum
likelihood, a method to calculate the uncertainties is specifically developed -
without giving rise to any particular difficulties. Finally, the end product is a
reliable instrument, both in terms of determining parametric estimations and
the associated uncertainties.
Chapter 7 gives a first overview of the results of identification. In our
climatic archive catalogue, we have selected 8 an initial combination out of
sixteen (four temperature reconstructions and the same amount for solar
irradiance). Two identification methods are shown. The first is a “free”
identification, whereby the six parameters of the structure chosen in
Chapter 4 minimize the error between the overall simulated temperature and
the historic temperature data, without any a priori constraints being
imposed. The second is a “forced” identification, where some parameters are
fixed to comply with the assumptions given in Chapter 5. Some of the
parameters in the free identification are located very far outside the IPCC
range, especially with regard to the climate's sensitivity to solar irradiance.
As long as we are dealing with recent warming (end of the 20 th Century), the
visual examination of the simulated output temperature from both equally
reproduce this warming. The difference being that with free identification,
the contribution of solar irradiance is highly predominant over that of CO 2
levels, while the opposite is true for forced identification. However, the
IPCC experts claim that it is physically impossible for solar irradiance to
have a significant impact on the climate. It is therefore important to go
further, and not simply rely on a visual impression. Statistical analysis starts
by assessing the autocorrelation function of the output error, and the cross-
correlation with input signals. This confirms that the constraints of forced
correlation lead to a strong correlation with solar irradiance. This points
towards a cause-effect relationship which is not taken into account. Even
more importantly, areas of related uncertainty show that the IPCC's
assumption of weak sensitivity to solar irradiance must be rejected. There is
a very low likelihood of error in such a result. This rejection is based, not on
considerations from theoretical physics, which are excluded from our study,
but on the statistical processing of observations. If the observations and
processing are correct, one must therefore conclude that the assumptions
given above are false.
8 The one which, with hindsight, appeared to be the most representative.