Reconstructions of palaeoclimate allow us to extend the climate record back into eras where the climate was different from today, and link these changes to external influences (sometimes referred to as boundary conditions) on climate. This is crucial to our understanding of future climate, when the boundaries will also be different from today. Palaeoclimate reconstructions also allow us to validate our climate models 'out-of-sample', since they have been extensively tuned on current climate data. However, reconstructing palaeoclimate is a very challenging process. At a given site (i) the sensor (eg a plant) responds to several aspects of climate jointly, such as temperature, precipitation, and seasonality; (ii) the sensor also responds to non-climate features, some of which are known (eg gross elevation) and some of which are not (eg aspect, local forestation); (iii) the proxy is very imperfectly recorded in the archive (eg pollen in lake sediments, tree-ring density or width). There are additional challenges in combining results from different sites, and over time, and in using them to reconstruct spatial patterns or large or hemispheric-scale averages of climate. Ideally the result of the reconstruction process would be a joint distribution for, say, temperature, in which the uncertainty represents the accumulation of the uncertainties introduced at the various stages. Interpretation of the resulting records, for example in attribution studies will then need to account very carefully for the uncertainties in either data.
We are currently quite far from achieving this. Progress will require the close co-operation of palaeoclimate scientists and statisticians. Palaeoclimate scientists are fully aware of the uncertainties in climate reconstructions, but lack the tools and the expertise to account for these uncertainties. The objectives of this workshop are to raise the awareness of palaeoclimate scientists to statistical methods for reconstruction (eg Bayesian inversion of forward models), to interest statisticians in the challenges of palaeoclimate reconstruction, and to foster collaborations between the two groups.