IUGG 2003 
Session MC14: G. Hegerl and P. Stott, Conveners
Detecting and Attributing the Early Signs of Human Influence on the Climate (ICC1)

Invited talk:

MULTI-RESOLUTION METHODS FOR EXTRACTING FINGERPRINTS OF EXTERNAL
FORCING DURING THE LAST MILLENNIUM FROM MODEL AND PROXY DATA

Caspar M Ammann, Philippe Naveau, Hee-Seok Oh, Fortunat Joos, 
Stefan Gerber, Dave S Schimel, Bette L Otto-Bliesner

National Center for Atmospheric Research, Boulder CO, University of
Colorado, Boulder CO, University of Alberta, Edmonton, Canada,
University of Bern, Switzerland

Confidence in the detection of human influence on climate would be
increased if the natural climate variability could be explained to a
better degree. This endeavour is challenging because indirect data has
to be used to estimate both the major forcing factors as well as
measures of past climate variations themselves. The instrumental
record is too short to verify our ability to capture the magnitude and
temporal evolution of externally forced and internal climate
variations. Extension back in time is also critical for the detection
process because the early 20th century warming seems to be due to
natural warming out of one of the colder Little Ice Age episodes and
thus needs to be put into perspective of the past.

External forcing factors, such as solar irradiation changes and
influences from explosive volcanism, can significantly perturb the
earth radiation balance. To a first order, this effect can be
simulated and studied with a range of different climate models. In
recent years, additional emphasis has been given to a host of
dynamical feedbacks of the coupled climate system to these external
forcings. These secondary effects can not only strengthen or weaken
the overall radiative effects on the global to hemispheric scale, they
also result in specific temporal and spatial variations. One striking
example is a common winter warming over northern high-latitude
continents after large volcanic eruptions. Such temporally and
spatially inhomogenous expressions of the natural variability are part
of the proxy record and careful separation of these signals is
desirable. When forced appropriately, coupled Ocean-Atmosphere General
Circulation Models can reproduce a number of these spatial patterns
and thus offer an additional way to analyse and compare past climate
reconstructions and simulations.

We present new statistical techniques designed to specifically focus
on detecting and quantifying externally forced variations in proxy
climate data. We compare forcing to impact relationships found in the
climate reconstructions with those extracted from coupled model
simulations with the NCAR-Climate System Model (Version 1.4) for the
time interval of last Millennium. Additionally, to provide some
constraints on the magnitude of estimated and simulated low frequency
variations due to solar irradiance changes, simulations with the
simplified climate-carbon cycle model from Bern (Bern CC) are
presented. Finally, the results are presented in the context of the
20th century climate, and unexplained residuals from the viewpoint of
natural climate variability are isolated.