Here they speak of 5 sources of uncertainty in ice core data...
this is one I found interesting.
"Quantitatively representing the effects of possible changes in the relationships between
the proxy variables measured, and the climate variable they are supposed to represent,
is far more problematic. A case in point is the relationship between sea ice and MSA,
which is often supposed to arise from the greater production of MSA in water stratified
by melting sea ice. Abram et al. (2007) concluded that at least in one location, the
ââ¦relationship is most likely due to variations in the strength of cold offshore wind
anomalies ... which act to synergistically increase sea ice extent while decreasing MSA
delivery to the ice core sites.â Beside purely empirical studies (e.g. statistical
calibration/verification residuals), the only way to address this kind of uncertainty is
with process-based studies â in particular, coordinated measurements in the atmosphere
and snow to better understand air-snow transfer relationships â combined with the
incorporation of ice core proxy variables into atmospheric models. In particular, the
inclusion of ice core proxies, in addition to the water isotopes, should be encouraged in
the modeling community (e.g. Field et al., 2006, for
10
Be). To the extent that ice core
observations can be shown to validate these model predictions (an exercise that is just
beginning to be realized), uncertainties can be quantified by conducting model runs to
examine the sensitivity of the resulting proxies to various changes in climate; Schmidt et
al. (2007) provides a useful example."
http://www.ncdc.noaa.gov/paleo/reports/trieste2008/ice-cores.pdf