Geoscience Reference
In-Depth Information
past may increase the accuracy of sensitivity estimates owing to the possibly
larger sizes of those changes (K
hler et al.
2010
; Masson-Delmotte et al.
2013
).
This approach meets two obstacles. First, the paleo analyses regard longer timescales
than from
ö
1850 to present, therefore additional feedback processes may act
(Masson-Delmotte et al.
2013
). Second, S may not be a constant but depend on the
mean climatic state (K
*
hler et al.
2010
; Masson-Delmotte et al.
2013
).
To tap the potential of the paleoclimate archives (including model output) whilst
bypassing both obstacles, the ClimSens project concentrated on the IG sections of
various paleoclimatic records and
ö
fitted statistical change-point regression models
to time-dependent temperature and forcing. This approach rests on the assumption
of including in the analysis large enough changes in both variables while avoiding
glacial intervals, which could exhibit different climate sensitivities. The ratio of the
slopes (temperature over forcing) is used to estimate climate sensitivity on within-
IG timescales. This method deviates from visually comparing both variables
(K
ö
hler et al.
2010
) or from
fitting an errors-in-variables regression to both vari-
ables (Mudelsee
2014a
).
ClimSens contributes to INTERDYNAMIK via quantifying trends and studying
feedback mechanisms also for older IGs (i.e., before MIS 1) utilizing data from ice
cores, marine archives, and climate modeling.
2 Materials and Methods
The database (Mudelsee
2014b
) comprises (1) one long [i.e., back to
800 thou-
sand years (ka) before present (BP)] radiative forcing anomaly time series (
*
R)
based on EDC data (greenhouse gases) and modeling, (2) the classic long EDC
Antarctic temperature series, (3) one long Northern Hemisphere (NH) surface-air
temperature series based on
ʔ
18
O data and modeling, (4) two long low-resolution
δ
SST series, (5)
400 ka BP) SST series, and (6) 21 high-
resolution Holocene SST series. One wishes to have many records from distributed
locations, on the other one needs a certain length and a high resolution to perform
statistical analyses. The dates for the IG boundaries (R
ve shorter (back up to
*
thlisberger et al.
2008
;
NEEM Community Members
2013
) are augmented by the date for the end of
Marine Isotope Stage (MIS) 11, which was set to 395 ka BP (when EDC has the
same temperature as at MIS 11 start 426 ka BP).
We
ö
(Mudelsee
2014a
) to the records to infer changes in temperature and forcing. The model allows
for the observed evidence that even within an IG, the climate needs neither be
constant nor display a monotonic trend; rather, IG climate may show trend changes,
as is seen for the Holocene and its
fit a simple change-point regression model called the
“
break
”
“
”
Holocene climate optimum
(Wanner et al.
2008
). By systematically
fitting the break model to a large set of records, we learn
about changes in time, the signatures of
, and generally about the
dynamics of IG climates. Despite that, the farther back in time, the fewer are the
suitable records and the lower is the spatial coverage.
“
IG optima
”
Search WWH ::
Custom Search