The Earth System at the Last Glacial Maximum

 

This figure shows the relationship between tropical cooling for the LGM and equilibrium climate sensitivity in PMIP2 and CMIP5/PMIP3 models. From Hopcroft & Valdes (2015).

Relevant publications:

P.O. Hopcroft, P.J. Valdes, S. Woodward and M. Joshi (2015). Last glacial maximum radiative forcing from mineral dust aerosols in an Earth System model, Journal of Geophysical Research, in press, doi:10.1002/2015JD023742. Supporting Information: pdf

P.O. Hopcroft and P.J. Valdes (2015). How well do simulated last glacial maximum tropical temperatures constrain equilibrium climate sensitivity?, Geophysical Research Letters, 42, doi:10.1002/2015GL064903. Supporting Information: pdf

P.O. Hopcroft and P.J. Valdes (2014).
Last Glacial Maximum constraints on the Earth System Model HadGEM2-ES, Climate Dynamics, online early, doi:10.1007/s00382-014-2421-0.

The last glacial maximum (LGM) is generally defined as a period of maximum ice volume during the last ice-age (e.g. Mix et al 2001). This occurred 21,000 years ago when sea-level was approximately 120m lower than now. This sea-level drop was caused by an expansion of the all of the main ice-sheets present today, and the formation of ice-sheets in North America, Europe and Patagonia in the Andes. Several other aspects of the Earth System were modified at the LGM in comparison to the pre-industrial or present day. Together they signify a fundamentally different global climatic state. The 3 major greenhouse gases have been measured in ice-cores from Antarctica and were all reduced during the ice-age with near-minimum values during the LGM. Carbon dioxide was reduced to around 185ppmv (parts per million volume) compared to 280ppmv in the pre-industrial era. Understanding the causes of the reduced CO2 at the LGM is a major research topic which is reviewed in detail by Kohfeld & Ridgwell (2001). Similarly methane (CH4) and nitrous oxide (N2O) were reduced.


A recent compilation of climate reconstructions gives a global temperature change at the LGM of around 4C (Shakun et al 2012) which is comparable in magnitude to the warming projected by the year 2100 in business-as-usual type future scenarios. The LGM has been a focus time-period for palaeoclimate modelling for a number of years, with simulations co-ordinated between different modelling centres through the Palaeoclimate Model Intercomparison Project (PMIP). Models predict around the right magnitude of global temperature cooling at the LGM, but tend to overestimate the tropical cooling (Braconnot et al 2007, Schmidt et al 2014, Annan & Hargreaves, 2015).


Hargreaves et al 2012, showed that the LGM tropical temperature change may provide a useful constraint on climate sensitivity. Climate sensitivity is generally defined as the global mean warming in response to a doubling of CO2. So constraints on the value derived from past climate states could help to reduce uncertainty in future climate change projections. A new ensemble of LGM simulations performed with the most recent models as used in the IPCC's 5th Assessment report, gave the opportunity to expand the number of models and perhaps arrive at a more statistically robust correlation of climate sensitivity and LGM temperature changes. The results (Schmidt et al 2014, Hopcroft & Valdes, 2015) show that this correlation breaks down in the newer models. This appears to be mostly driven by additional Earth System feedbacks now incorporated in some of the models, such as changes in the global vegetation distribution and changes in the loading of dust in the atmosphere. For example MIROC-ESM which shows the strongest deviation in the figure below includes dynamic vegetation and an interactive mineral dust cycle. GISS-E2-R which has a low climate sensitivity value but shows a relative large cooling signal at the LGM, uses prescribed vegetation from palaeo-vegetation reconstructions. These results highlight that these extra factors can lead to relatively small temperature changes in the tropics, but that these are enough to perturb the relationship with climate sensitivity between the different models.


If all of the models were to include all of the same processes (i.e. dynamic vegetation, aerosols), then perhaps there would be established a new correlation between these two variables. Alternatively, the uncertainty in these processes may dominate the resultant temperature changes precluding us from making inferences about climate sensitivity from LGM climate model simulations.