The period of the Last Glacial Maximum (19-23ka) and the subsequent deglaciation covers huge changes to the climate system, comparable in magnitude to those anticipated over the next century. A wide variety of data covering this interval are available from cores drilled on land, ice sheets and ocean floor, but these data are limited in time and space, and often their interpretation is rather uncertain. Thus, while they provide important clues, they do not give us a comprehensive description of the global spatiotemporal evolution of the deglaciation and the causal relationships that underpin the changes. Recent growth in computational power is also starting to enable simulations of this interval with fully coupled GCMs, although as yet very few such simulations have been completed. While such model simulations could provide a coherent picture of climate changes underpinned by physical laws, their accuracy is limited by numerous uncertainties and approximations in the models.
Thus, we plan to bring together models and data to generate a new “reanalysis” (i.e. model-data synthesis) of the past deglaciation. We have already performed a similar synthesis focussing on the Last Glacial Maximum, a period for which much more comprehensive data sets have been compiled and which has already been simulated by a number of different coupled GCMs within the PMIP projects. Standard data assimilation techniques cannot realistically be applied to this type of problem, so we have adapted a simple scaling approach which we show to work well in tests. In this talk we will present results from the LGM reconstruction (Annan and Hargreaves 2013), and some (very!) preliminary results from extending this approach to the full transient deglaciation.