Assessing the robustness of Antarctic temperature reconstructions over the past 2 millennia using pseudoproxy and data assimilation experiments

Martin.Werner [ at ]


The Antarctic temperature changes over the past millennia remain more uncertain than in many other continental regions. This has several origins: (1) the number of high-resolution ice cores is small, in particular on the East Antarctic plateau and in some coastal areas in East Antarctica; (2) the short and spatially sparse instrumental records limit the calibration period for reconstructions and the assessment of the methodologies; (3) the link between isotope records from ice cores and local climate is usually complex and dependent on the spatial scales and timescales investigated. Here, we use climate model results, pseudo-proxy experiments and data assimilation experiments to assess the potential for reconstructing the Antarctic temperature over the last 2 millennia based on a new database of stable oxygen isotopes in ice cores compiled in the frame- work of Antarctica2k (Stenni et al., 2017). The well-known covariance between δ18O and temperature is reproduced in the two isotope-enabled models used (ECHAM5/MPI-OM and ECHAM5-wiso), but is generally weak over the different Antarctic regions, limiting the skill of the reconstructions. Furthermore, the strength of the link displays large variations over the past millennium, further affecting the potential skill of temperature reconstructions based on statistical methods which rely on the assumption that the last decades are a good estimate for longer temperature reconstructions. Using a data assimilation technique allows, in theory, for changes in the δ18O–temperature link through time and space to be taken into account. Pseudoproxy experiments confirm the benefits of using data assimilation methods instead of statistical methods that provide reconstructions with unrealistic variances in some Antarctic subregions. They also confirm that the relatively weak link between both variables leads to a limited potential for reconstructing temperature based on δ18O. However, the reconstruction skill is higher and more uniform among reconstruction methods when the reconstruction target is the Antarctic as a whole rather than smaller Antarctic subregions. This consistency between the methods at the large scale is also observed when reconstructing temperature based on the real δ18O regional composites of Stenni et al. (2017). In this case, temperature reconstructions based on data assimilation confirm the long-term cooling over Antarctica during the last millennium, and the later onset of anthropogenic warming compared with the simulations without data assimilation, which is especially visible in West Antarctica. Data assimilation also allows for models and direct observations to be reconciled by reproducing the east–west contrast in the recent temperature trends. This recent warming pattern is likely mostly driven by internal variability given the large spread of individual Paleoclimate Modelling Intercomparison Project (PMIP)/Coupled Model Intercomparison Project (CMIP) model realizations in simulating it. As in the pseudoproxy framework, the reconstruction methods perform differently at the subregional scale, especially in terms of the variance of the time series produced. While the potential benefits of using a data assimilation method instead of a statistical method have been highlighted in a pseudoproxy framework, the instrumental series are too short to confirm this in a realistic setup.

Item Type
Primary Division
Primary Topic
Peer revision
ISI/Scopus peer-reviewed
Publication Status
Eprint ID
DOI 10.5194/cp-15-661-2019

Cite as
Klein, F. , Abram, N. J. , Curran, M. A. J. , Goosse, H. , Goursaud, S. , Masson-Delmotte, V. , Moy, A. , Neukom, R. , Orsi, A. , Sjolte, J. , Steiger, N. , Stenni, B. and Werner, M. (2019): Assessing the robustness of Antarctic temperature reconstructions over the past 2 millennia using pseudoproxy and data assimilation experiments , Climate of the Past, 15 (2), pp. 661-684 . doi: 10.5194/cp-15-661-2019


Download (11MB) | Preview



Research Platforms


Edit Item Edit Item