User Tools

Site Tools


PMIP-CMIP6 Experimental Design Discussion: Last Glacial Maximum

The PMIP4 experimental design available on the PMIP3 wiki has been superseded by the experimental design detailed on the PMIP4 web site

Please submit your comments in the Discussion section at the end of this page

The last glacial maximum (lgm, ~21000 years ago) is the last global cold extreme in which greenhouse gas concentrations were at their minimum and continental ice-sheet at their maximum size, covering large areas of northern North America and northwestern Eurasia. Like the mid-Holocene, the lgm has been a key PMIP experiment since PMIP started, and as such has been the focus for paleo-data syntheses.

The reference experiment to which the lgm is compared to is the pre-industrial control, which is part of the DECK.

For the LGM, our proposed experimental set-up for PMIP-CMIP6 is based on the PMIP3-CMIP5 set-up:

Orbital parameters

eccentricity = 0.018994

obliquity = 22.949°

perihelion-180° = 114.42°

Date of vernal equinox = March 21 at noon

solar constant = same as PI

Concentration of atmospheric trace gases

CO2 = 185 ppm

CH4 = 350 ppb

N2O = 200 ppb

CFC = 0

O3 = same as in CMIP5 PI

Topography/bathymetry/coastlines/ice sheets

This is one of the major changes compared to pre-industrial. Prescribing the lgm ice sheets means implementing new coastlines, a new topography/bathymetry and a new ice sheet extent.

  • Modelling groups planning to run the lgm experiment but not the deglaciation are advised to use the PMIP3 ice sheets/coastlines, so that PMIP3-CMIP5 and CMIP6 can be rigorously compared.
  • Modelling groups planning to run lgm as the start of the the deglaciation experiment (cf PMIP3 deglaciation working group) can use the new boundary conditions provided by Dick Peltier and Lev Tarasov.

Vegetation and land surface

  • Computed using a dynamical vegetation module,
  • or prescribed as in PI, with phenology computed for models with active carbon cycle
  • or prescribed from data

Carbon cycle

Interactive, with atmospheric concentration prescribed and ocean and land carbon fluxes diagnosed as recommended in CMIP5.

For PCMIP: fully interactive with atmospheric concentration computed by the model.

Dust cycle

We invite the modelling groups planning to run simulations with an interactive dust cycle to comment on the most appropriate methodology for their model, in particular as to their ability to use a interactive vegetation description.

Note on the fresh water budget

Modelling groups are advised to carefully check the fresh water budget in their lgm experiments in order to avoid unnecessary drifts of the ocean salinity. It can be necessary to route the snow which has fallen in excess on the ice sheets to the ocean. Given the change in coastlines, it is also sometimes necessary relocate the large rivers' estuaries on the coast.

Global ocean salinity

Add 1 PSU.

[ PMIP3 Wiki Home ] - [ Help! ] - [ Wiki syntax ]


Bette Otto-Bliesner, 2015/04/06 20:47

It seems that we are now giving modeling groups three ice sheet options for the CMIP6 LGM simulations. I am not sure the reason. If modeling groups are only running the LGM, then they are advised to use the PMIP3 ice sheet. The implication is that the CMIP5 and CMIP6 experiments can be more rigorously compared. That implies that those models that use the ICE6G_C or ANU reconstructions in their CMIP6 LGM cannot reliably be used for CMIP6.

Bette Otto-Bliesner, 2015/04/09 15:47

Typo in my comment. Last sentence should read:

That implies that those models that use the ICE6G_C or Tarasov's reconstructions in their CMIP6 LGM cannot reliably be used for CMIP6.

Lauren Gregoire, 2015/04/09 09:29

I think that we should prioritise using the latest data and reconstructions for our experiments rather than keeping compatibility with previous setups.

Since Ice-6G_C and Lev Tarasov’s reconstructions are newer (and more robustly built) than the PMIP3 one, I think they should be the default and the PMIP3 one should be used for sensitivity experiments and linking up with CMIP5. Moreover, if I remember Masa’s results, the PMIP3 ice sheet does produce differences in the Labrador sea climate compared to the Ice-6G and Tarasov reconstructions (which both produce similar climates), so the choice is not unimportant.

Another important point is that the 21 ka CO2 and CH4 values have changed due to improvements in the ice core chronologies :

  • CO2 = 188 ppm (previously 185 ppm)
  • CH4 = 375 ppb (previously 350 ppb)

For compatibility with the deglacial experiments it would be best to update these values in the LGM setup as we have suggested on the deglacial wiki


Ruza Ivanovic, 2015/05/19 10:26

I am in agreement with Lauren on this.

With regard to assessing model improvement (CMIP6 vs. CMIP5), I don’t see how this is possible using the old boundary conditions (especially the old PMIP3 ice sheet), which we know to be out of date. i.e. if the boundary conditions are ‘wrong’, then the models are hardly likely to reproduce the ‘observed’ climate anyway even if the models have been improved. Therefore using the most recent boundary conditions (e.g. ice sheet reconstructions) seems like the only option for even being able to evaluate the models against the palaeo data. It would also be a shame to disengage with the ‘data’ community on the boundary conditions and validation data.

For the same reasons, we should be using the most up to date greenhouse gas concentrations. It may only be a small change in CO2 and CH4, but the PMIP3 values are out of date (188 ppm vs. 185 ppm and 375 ppb vs. 350 ppb, respectively).


Andreas Schmittner, 2015/06/23 19:06

We've just shown that considering the effects of sea level on LGM tidal mixing can make a big difference in MOC simulations.

Would it be useful to include sensitivity experiments with LGM tidal mixing?

Allegra, 2015/09/21 16:12

Yes – update boundary conditions to be the best possible. It is what separates P-MIP from Aqua-MIP.

Why should we allow multiple ice sheet reconstructions? Because it gives us a numerical estimate of the impact of this uncertainty on our simulations.

e.g., for the Laurentide Ice Sheet:

So, for things that are much more difficult to assign boundary conditions, I think multiple reconstructions are okay (especially in groups that can do both). Perhaps a hierarchy? This one first, then if resources, an alternate?

LGM vegetation : My vote is to get pmip-sanctioned prescribed since we don't have dynamic veg at GISS.

For doing LGM dust sensitivity, we need to have dust emissions. Also, in talk with the dust folks at GISS, the single-scattering albedo is still uncertain for today – and it can have a big impact. Is there consistency across groups in this value? Should we do sensitivity experiments with this value for any paleo experiment? And… this value actually could be different in the LGM than today if the provenance of the dust is much changed.

Our freshwater budget is 'closed' to avoid drifts in salinity for GISS. Two things that made this possible (1) accumulation of mass on ice sheets is balanced by ice berg calving in specified boxes around the edges. (2) Variable lakes that can expand to cover multiple grid boxes or contract to get better local fw balance.

Also, since we will hopefully have a few groups using water isotopes, please prescribe SMOW (permil) H2 18O +1. ; HDO +8. (Or, any better ideas for dD since it is harder to know for the LGM? 8 is probably not quite right.)

Enter your comment. Wiki syntax is allowed:
pmip3/cmip6/design/21k/index.txt · Last modified: 2016/06/20 15:34 by jypeter