Transient simulations of the last 30,000 years, within the GENIE earth-system framework D.J. Lunt (1) M.Williamson (2) A. Price (3) P.J. Valdes (1)

1
Transient simulations of the last 30,000 years,
within the GENIE earth-system frameworkD.J.
Lunt (1) M.Williamson (2) A. Price (3) P.J.
Valdes (1)

• What is GENIE?
• Energy-balance-atmosphere transient coupled
  simulations
• Fully-dynamic-atmosphere transient slab
  simulations
• Conclusions and future work

• (1) BRIDGE, School of Geographical Sciences,
  University of Bristol, UK
• (2) ESM, School of Environmental Sciences,
  University of East Anglia, UK
• (3) E-science centre, University of Southampton,
  UK

2
What is GENIE?

• Earth-system modelling framework
• Focus on fast models glacial/interglacial
  cycles
• Modularity plug-in / plug-out
• Objective tuning
• GRID computational resource

http//www.genie.ac.uk
3
What is GENIE?
Atmosphere IGCM EMBM Prescribed
Land surface MOSES/ TRIFFID ENTS Prescribed
Seaice Goldstein Slab Prescribed
Ocean Goldstein Slab Prescribed
Ice-sheet GLIMMER (pdd) Prescribed
Biogeochemistry BIOGEM Prescribed
More complex
4
EMBM-GOLDSTEINtransient simulations

• 30,000 year transient simulation 30kyrBP -
  present
• Boundary conditions
• 21kyrBP present ICE4G ice-core CO2.
• 30kyrBP 21kyrBP linear interpolation using
  30kyBP ice- sheet reconstruction (STAGE3
  project).

5
EMBM-GOLDSTEINtransient simulations

• Global mean surface temperature evolution.
• Noise is due to integer ice-sheet mask (now
  fixed!).
• Spinup approx 4000 years from 20oC isothermal
  ocean.

6
EMBM-GOLDSTEINtransient simulations
HADSM3, SURFTEMP, LGM-MOD
EMBM-GOLDSTEIN, SURFTEMP, LGM-MOD
EMBM-GOLDSTEIN Global mean change,
LGM-Pre-industrial 4.1oC HadSM3

3.8oC
EMBM shows good agreement with HadSM3, but has a
much weaker response over seaice and Eurasian
ice-sheet.
7
EMBM-GOLDSTEINtransient simulations
Temp, LGMtr-LGMeq
Transient LGMeq

• How close is the LGM to equilibrium, anyway?!
• This implies very! Good news for equilibrium
  simulations such as in .

8
EMBM-GOLDSTEINtransient simulations

• Greenland is too warm and too insensitive in this
  model.
• Bolling-allerod warming is most significant
  departure from fixed model prediction.

GISP2 Model Model fix!
9
EMBM-GOLDSTEINtransient simulations

• Warm periods after H2 and H1 and Younger Dryas
  are not reproduced in the model.
• Model can aid with volume and duration of
  freshwater pulses of Heinrich events and duration
  of 8.2kyr event.\

GISP2 Model fix hosing
10
IGCM-SLABtransient simulations
Initial atmosphere-only tuning experiment
Genetic algorithm, 30 tunable paramters, targets
are modern (NCEP) surface heat fluxes.
11
IGCM-SLABtransient simulations

• 22,000 year simulation
• 2kyr LGM equilibrium spinup 21kyrBP - 1kyBP
  transient.
• Boundary conditions
• As for EMBM, but include orbital parameters.
• Acceleration x10.
• 120 year calibration run for implied ocean heat
  fluxes.
• Seaice flux limited to -50Wm-2

12
IGCM-SLABtransient simulations

• Variability seems quite constant, but be wary
  because of acceleration.
• In a colder simulation (pre-tuning), more
  variability in glacial due to seaice.

13
IGCM-SLABtransient simulations
HADSM3, SURFTEMP, LGM-MOD
EMBM-GOLDSTEIN, SURFTEMP, LGM-MOD
IGCM-SLAB, SURFTEMP, LGM-MOD
Global mean change, LGM-Pre-industrial EMBM-GOLDS
TEIN 4.1oC HadSM3
3.8oC IGCM-SLAB 2.0oC IGCM is in
insensitive region of parameter space. Need to
tune both modern and LGM. Seaice flux.
14
IGCM-SLABtransient simulations

• Relatively insensitive to initial conditions.
  Need a larger ensemble!

15
IGCM-SLABtransient simulations

• 2 extra runs - constant pre-industrial and LGM
  atmospheric CO2.
• Sensitivity of model to CO2 change from 195ppmv
  to 279ppmv is approx constant 0.75oC.

16
Conclusions

• EMBM-GOLDSTEIN has good snapshot agreement with
  HadSM3.
• LGM appears to be very close to equilibrium.
• Transient simulation shows reasonable agreement
  with ice-core record but is insensitive over
  Greenland.
• Model can be used to constrain volume/duration of
  freshwater pulses, but related warmings not
  simulated.
• IGCM-SLAB has good agreement with modern climate
  (GA tuning)
• Appears to have been over-tuned LGM too warm
  relative to EMBM and HadSM3.
• Sensitivity to CO2 forcing constant through
  glaciation.

17
Future Work
EMBM-GOLDSTEIN

• Tune modern and LGM with EnKF. Non-ENTS version
  already tuned using this technique (for modern
  only).
• Ensemble study of paleo-hosing experiments
  compare with data.
• Ensemble of accelerations.

IGCM-SLAB

• Re-tune modern and LGM.
• Compare to ice-core records.
• Ensemble of physicical mechanisms (akin to CO2
  presented here)

GENIE

• Run identical deglaciation experiments with
  different versions all tuned to the same
  targets.
• Ice-sheet coupling (see poster Z024),
  TRIFFID/MOSES coupling, EVP sea-ice coupling,
  BIOGEM coupling.