Reconstructing climate fields with terrestrial climate archives, isotope-enabled GCMs and Data Assimilation

Data Assimilation in paleoclimatology (PaleoDA) is a method that has been used in several climate reconstructions for the last millennium. By fusing information from both climate proxies and general circulation models (GCMs), PaleoDA provides statistical estimates of climate fields that are dynamically consistent. However, existing reconstructions mostly rely on calibrated tree ring data and assimilate proxy records on a single, annual time scale. Ice cores and speleothems, which record past variations in the oxygen isotope ratio of precipitation, often have a lower and irregular time resolution, but reliably record climate variations on decadal to centennial time scales. Here, we implemented a computationally efficient DA algorithm that enables the assimilation of proxy records on multiple timescales. The algorithm has been applied to speleothem and ice core records from the SISALv2 and Iso2k database and five isotope-enabled GCMs. Reconstructions of global mean temperature changes during the last millennium compare well in both amplitude and uncertainty to recent studies. The potential of incorporating speleothems is shown with a reconstruction of hydroclimatic changes in tropical South America, where speleothems represent the most abundant type of hydroclimate archive. The experiments performed suggest an increased reconstructed decadal to centennial variability by using proxy records on multiple timescales. Making use of different climate models shows the influence of model biases on the reconstructions. Future PaleoDA reconstructions could be improved from more proxy records and the multiple time scale approach to provide a globally complete picture of past climate changes.