Session: Glacial Climates (LGM, Last deglaciation, Ice sheet uncertainties, Glacial-interglacial cycles)
Author: Lauren Gregoire / email@example.com / University of Leeds, UK
Co-author: James Salter, University of Exeter, UK;
Daniel Williamson, University of Exeter, UK;
Tamsin Edwards, The Open University, UK;
Ice sheet models fail to reproduce reconstructed patterns of Northern Hemisphere ice sheet retreat through the last deglaciation (21,000-6,000 years ago). This is the main barrier to understanding the role of ice sheets in past abrupt climate and sea level changes and constraining the post-glacial isostatic adjustment of the solid Earth. The primary reason for this failure is the large uncertainty in the boundary conditions to the ice sheet models, which are derived from global climate simulations. Simulations of the transient evolution of climate over the last 21,000 years are computationally expensive to produce and have large biases and uncertainties in their inputs (e.g. ice sheet melt input to the oceans). It is thus impossible to produce large ensembles of climate simulations that would span plausible deglaciation climates.
We developed a statistical method to systematically explore the spatio-temporal uncertainty in climate (temperature and precipitation) through this period, by combining output from transient General Circulation Model (GCM) simulations of the last 21,000 years with surface air and sea surface temperature change proxy records. The method consist of producing basis vectors of climate change through single value decomposition of an ensemble of climate simulations. We then define a set of linear combinations of the basis vectors which match a compilation of proxy records of temperature changes within their uncertainty.
We present a pilot application of this method using a compilation of surface temperature records from Shakun et al. (2012) and output from a perturbed physics ensemble of FAMOUS simulations as well as the Trace-21k simulation produced with CCSM3. A set of 500 plausible spatio-temporal temperature field were thus produced. With this, we ran an ensemble of 500 simulations of the North American ice sheet evolution from 21,000 to 6,000 years ago with the Glimmer-CISM ice sheet model, where climate and ice sheet parameters are varied. An evaluation of the output against reconstructed ice extent is performed to identify plausible ice sheet simulations and their corresponding climate input.