Partial Results - Climate Change: Identification and Projections - page 76

highlight hypothetical behavior, or to conceal it. Hence the explanations
which follow.

Start of the period
All databases used do not go back to the year 1. Moreover, the oldest
archives lack reliability: they present a dispersion which would lead to
incoherent results if they were taken into account. Conversely, since the
climatic past is sparse in terms of significant events, valuable information
should not be sacrificed. In particular, it is important to fully cover the
Medieval Warm Period, meaning a start date T 0 a reasonable time before the
year 1000. We noted that overall the results fit together, all of them T 0 being
situated between the years 500 and 1000. Two of the archives on irradiance
happen to start in 695, which is why we have chosen this date for T 0 .

End of the period
T 1 will be the current year in order to incorporate the latest available data.
We will also carry out identifications stopped at T , followed by

blind short-term predictions in order to assess the predictive capacity of the
models.

= 1999

1

Figure 7.1. An initial input/output set