Greenland temperature and precipitation over the last 20 000 years using data assimilation

Abstract

Abstract. Reconstructions of past temperature and precipitation are fundamental to modeling the Greenland Ice Sheet and assessing its sensitivity to climate. Paleoclimate information is sourced from proxy records and climate-model simulations; however, the former are spatially incomplete while the latter are sensitive to model dynamics and boundary conditions. Efforts to combine these sources of information to reconstruct spatial patterns of Greenland climate over glacial–interglacial cycles have been limited by assumptions of fixed spatial patterns and a restricted use of proxy data. We avoid these limitations by using paleoclimate data assimilation to create independent reconstructions of mean-annual temperature and precipitation for the last 20 000 years. Our method uses oxygen isotope ratios of ice and accumulation rates from long ice-core records and extends this information to all locations across Greenland using spatial relationships derived from a transient climate-model simulation. Standard evaluation metrics for this method show that our results capture climate at locations without ice-core records. Our results differ from previous work in the reconstructed spatial pattern of temperature change during abrupt climate transitions; this indicates a need for additional proxy data and additional transient climate-model simulations. We investigate the relationship between precipitation and temperature, finding that it is frequency dependent and spatially variable, suggesting that thermodynamic scaling methods commonly used in ice-sheet modeling are overly simplistic. Our results demonstrate that paleoclimate data assimilation is a useful tool for reconstructing the spatial and temporal patterns of past climate on timescales relevant to ice sheets.

Cosmogenic isotope measurements from recently deglaciated bedrock as a new tool to decipher changes in Greenland Ice Sheet size

“…The temperature reconstruction relies on oxygen-isotope records from eight ice cores, the precipitation reconstruction uses accumulation records from five ice cores, and both are guided by spatial relationships derived from the transient climate-model simulation TraCE21ka (Liu et al, 2009;He et al, 2013.) The climate reconstruction is in good agreement with published paleoclimate reconstructions (Badgeley et al, 2020 and references therein). The climate reconstruction yields 9 combinations of temperature and precipitation that are used as transient climate boundary conditions to force 9 ice model simulations.…”

“…Holocene ice model simulations use the highest horizontal mesh resolution across our field area to date, and use a range of state-of-the-science gridded climate reconstructions as the input climate (Badgeley et al, 2020;Briner et al, 2020), thus presenting an opportunity for new insights regarding glacier history and ice-sheet modeling. Our goals in exploring the model simulations are to (a) assess the magnitude of recession inboard of the present margin, (b) compare rates of retreat and timing of ice-margin change in both the model and in the observations, and (c) explore avenues for model improvement in a known problem area for ice-sheet modeling.…”

“…Simulations 7 through 9 rely on a temperature history that has a muted early Holocene warming compared to other runs, followed by peak Holocene mean annual temperature anomalies above the 1850-2000 mean from 7 ka to 4 ka BP (Badgeley et al, 2020;Briner et al, 2020;Fig. 17).…”

Can we reconstruct the formation of large open-ocean polynyas in the Southern Ocean using ice core records?

“…Nevertheless, frequent polynya formation during some periods should lead to a higher production of AABW in the Weddell Sea and, potentially, to a modification of its characteristics. It has been speculated that AABW formation because of open-ocean convection was higher during the 1350-1850 period, potentially because of colder conditions at that time, than in the more recent past (Broecker et al, 1999). By contrast, from the analyses of marine records on the southern Chilean margin in the southeast Pacific, Collins et al (2019) suggest that AABW formation was weaker after 1400 compared with the periods before.…”

“…As classically done, we consider that the errors are not correlated and that the observation error covariance matrix is diagonal. Because of the large spatial variability mentioned above, we also assume here that the representation error is much larger than the measurement error (Thomas et al, 2017;Laepple et al, 2018;Cavitte et al, 2020;Badgeley et al, 2020), and we only include the contribution of the former in our estimate of the error. The representation error is due, in particular, to the fact that the model is not able to simulate the small-scale processes that are included in the signal recorded in the archive (the so-called "error of representation due to unresolved scales and processes"; see, for instance, Janjić et al, 2018).…”