• Introduction, and history of Eocene modelling• Atmosphere-Ocean coupled models• ‘EoMIP’• Solutions? – sensitivities to uncertainties

Modelling Eocene Climates: Can any model get it ‘right’?!

Dan LuntM. Heinemann, M. Huber, A. Legrande, A. Ridgwell, P. Valdes, A. Winguth

Zachos et al, Nature, 2001

Orientation…

Bijl et al., Nature 2009

Reduced latitudinal temperature gradients compared to modern…

‘Warm’ mid-high latitudes also supported by a wealth of terrestrial evidence, e.g. Croccodillians, pollen data, etc.

BUT….

‘early’ models, e.g. Huber and Sloan, Bice, etc etc. UNABLE to reproduce, given palaeogeography and CO2 changes.

Is this still the case in 2010?!

General Circulation Models (GCMs)

2001 2007

2007

2001

1995

1995

1990

1990

History of GCMs

Surface Temperature: observations

Surface Temperature: HadCM3

How good are GCMs?(1) temperature

Precipitation: observations

Precipitation: HadCM3

Seaice: observations vs models

How good are GCMs?(2) Precip and seaice

Topography

Veg

Eocene: boundary conditions

(1) Palaeogeography

Eocene: boundary conditions

(2) CO2

Zachos et al, Nature, 2008

Eocene Model Intercomparison Project (‘EoMIP’) results

Lunt et al, Geology, 2010

Heinemann et al, Climate of the Past, 2009

Winguth et al, Journal of Climate, 2010

Huber et al, PPP, 2006

Roberts et al, EPSL, 2009

Panchuk et al, Geology, 2008

New SST/terrestrial data compilation soon, led by Tom Dunkley Jones

280 560 1120 2240 4480 ppmv

What are the reasons for the differences…?

Boundary conditions: 2 x CO2

0.4% decrease in solar constant palaeogeography uniform vegetation/soil everything else modern

Sensitivity to uncertainties:(1) Internal model parameters and clouds

“Control” climate (after 1000 years):

• Perform 100 simulations for the Eocene, varying some key model parameters.

• Do any of these simulations result in a good (i.e. warm pole) simulation?

• Select 10 poorly defined parameters• Select reasonable possible ranges for each parameter• Vary them together (using a ‘latin-hypercube’ sampling method)

• Clouds: Threshold of relative humidity for cloud formation (RHcrit) Precipitation ice fall out speed (VF1) Conversion rate of cloud liquid water droplets to precipitation (CT) Threshold value of cloud liquid water for formation of precip. (CW)

• Convection : Convective roughness length over the sea (Z0FSEA)

• Gravity wave parameters (WAVE)

• Sea ice low albedo (ALPHAM)

• Diffusion in ocean and atmosphere

• 100 simulations performed, each simulation set for 1000 years.– 59 simulations failed within 100 years!– 4 further simulations failed to complete 1000

years.– Hence only 37 simulations completed to 1000

years.• Of these, 19 failed to complete 4000 years• But 18 have completed 10,000 years

Sensitivity uncertainties:(2) Effects of Open Arctic:

Change in climate due to opening Arctic connections to rest of ocean

DJF

JJA ANN

See also Roberts et al, 2009, EPSL

Change in surface air temperature due to orbital parameter changes

Orbital parameters similar to 9kyr BP Obliquity = 25.5o (c.f. 23.5o)

Sensitivity to uncertainties:(3) Orbits:

• EoMIP – comparison of ‘current’ Eocene model simulations

• ‘State of the art’ models still unable to reproduce Eocene climates

• Model uncertainty, Arctic gateways and orbital variations could combine to reconcile models and data

• ……What about issues with the data??!

Warm Climates of the Past – a lesson for the future?10 - 11 October 2011The Royal Society, London