Geoscience Reference
In-Depth Information
Figure 7.1 displays this data over the selected period: 695-2013. This is
the raw, unsmoothed data as it is to be processed by the identification
algorithms.
The noise which affects the modern measurements (after 1850) differ
from earlier reconstructed temperatures. Nevertheless, they are accepted as
they are without any arbitrary pre-processing. Likewise, irradiance
reconstructions before 1700 filter the solar cycles. However, we can take
advantage of the ones which are there. One of the benefits of OE is that it is
robust in this type of heterogeneity.
7.2. Free identification
By
free
identification, it is understood that the parameters to be estimated
are free of all constraints. The identification method (OE) also has no degree
of manipulable freedom. The data therefore speaks freely, without being
forced by any arbitrary assumption or
a priori
knowledge
1
.
Simulations using the identified model
Anthropic contribution
Free model simulation
0.5
Ljungqvist
0.5
0
-0.5
0
1000
1500
2000
TSI c ontr ibution
-0.5
1000
1500
2000
0.5
0
Output error
-0.5
1000
1500
2000
0.5
Volcanic contributions
0
0.5
0
-0.5
-0.5
1000
1500
2000
1000
1500
2000
Figure 7.2.
Simulations from the free model
1 Only the range of time constants [1, 10, 100
year
] is somewhat arbitrary.
Search WWH ::
Custom Search