Workshop: Using Paleo-Climate Model/Data Comparisons to Constrain Future Projections

For the first time in climate modelling history, there is a substantial component of coordinated paleo-climate modelling being performed with the same models and with the same protocols as the projections of future climate change. We have heard for many years the claim that understanding paleo-climate is the key to understanding future potential changes, and indeed, much work has already been done (via PMIP 1 and 2, and many individual studies). However, there has been a lack of quantitative analysis that truly links the future simulations or forecasts with skill or sensitivity in the paleo-climate simulations or forecasts with skill or sensitivity in the paleo-climate simulations, most often because paleo-simulations were not done with the same models currently being used for future projections and through a lack of suitable paleo-climate skill metrics.

As part of the CMIP5 protocol and archive, the paleo-component (PMIP3) consists of simulations for the Last Glacial Maximum (LGM), the Mid-Holocene (MH - 6000 yrs BP) and the Last Millennium (LM, specifically 850 AD-1850 AD) (Taylor et al., 2011). For this database, these simulations will be true 'out-of-sample' tests in that there will not have been enough time to 'tune' any of these models to produce better paleo simulations. (Note that this is not necessarily unwise, but it would complicate some of the potential analyses.)

Many of the issues highlighted in the IPCC AR4 report (such as sub-tropical rainfall changes, ENSO, the N. Atlantic meridional circulation, the fate of Arctic sea ice, etc.) can potentially be illuminated by targeted analyses of the paleo-climate simulations since many of these subsystems have changed in the past and left decipherable proxy records (Schmidt, 2010). Unfortunately, the majority of skill metrics taken from the historical (20th Century) simulations are not constraining the future projections (i.e. models that are either good or bad at simulating some aspect of modern climate - the climatology, seasonal cycle, or interannual variability - give essentially the same spread of future projections) (Santer et al., 2010; Knutti et al.). Paleo-climate offers a substantially larger signal than climate changes in the 20th Century (albeit with substantial noise and difficulties in interpretation) and so is unique in potentially being able to constrain future change.

The problems in dealing with paleo-climate data are well-know but progress is being made. Issues related to proxy interpretation are being helped by the development of forward models for proxies, and by the using of multi-proxy interpretation are being helped by the development of forward models for proxies, and by the using of multi-proxy approaches and new techniques of Bayesian inference. The increase in scope of coherent databases of paleo-climate information (including uncertainties) is helping push model-data comparisons away from metrics based on single proxies or merely qualitative comparisons towards global, multi-proxy quantitative comparisons.

The next IPCC report (AR5) is due in 2013, with final deadline for submitted papers being July 31, 2012, and 'in press' deadlines for papers to be assessed is March 15, 2013. Many of the studies that will be assessed for AR5 will use the initial contributions to the CMIP5 archive (thought the archive itself will continue to grow until 2015). Thus the early part of 2012 is a key time period for thinking about and producing key work that will feature in the AR5 and set the agenda for similar studies and programmes in the future.