Paleo-Earth System Modelling

1
Paleo-Earth System Modelling

• Paul Valdes
• School of Geographical Sciences,
• University of Bristol

2
Structure of Talk

• Introduction Why do we need a paleo-perspective
  to Earth System Models?
• Example 1 Palaeoclimate Model Intercomparison
  Project (PMIP)
• Example 2 Paleo-atmospheric composition.
• Example 3 Rapid paleoclimate change.
• Example 4 Palaeoclimate studies of the distant
  past

3
Why Study Paleoclimates

• Test our understanding of Earth system dynamics
• Previous paleoclimate research has highlighted
  missing features of Earth System Models
• Test Earth System Models
• Especially important now that we are including
  slow components of Earth System
• Fundamental science (cultural science)
• E.g. Evolution of life.
• Direct Commercial and Policy Applications for
  long-term changes
• Oil formation strongly linked to climate
• Many other mineral reserves sensitive to climate
• Long term climate change relevant for nuclear
  waste disposal and may also show thresholds for
  CO2 increases.

4
What time periods to study?
(From Frakes, 1979)
5
What time periods to study?

• No direct climate analogues for the future
• Test climate model processes
• Choose periods which have a large signal to
  uncertainty ratio
• Last 2000 years
• Signal/Uncertainty (0.2-0.5 oC/0.2 oC) gt 1
• Last Glacial Cycle (0 125,000 years ago)
• Signal/Uncertainty (0-5 oC / 1-2 oC) gt 1
• Cretaceous period (100 million years ago)
• Signal/Uncertainty (10 - 15 oC/ 5-10 oC) gt1

6
What Climate Model to use?The dynamic hierarchy
of models

• State-of-the-art model
• Rigorous test of these models comparison to paleo
  periods
• Lower resolution version of above
• Uncertainty in boundary conditions
• Intermediate complexity models
• Development of conceptual models, systems
  approach, emergent properties, but NOT the best
  predictive model
• Energy balance/ box models
• Educational, illustrative etc.
• No model is complete. Definitions of each type of
  model will change with time

7
Glacial Interglacial cycles

• Fundamental changes of the Earth System which are
  still poorly understood, especially in a
  quantitative sense
• Amplitude of variability shows us that Earth
  system feedbacks are essential is we are to
  explain these observations
• Physical and biological feedbacks both important

8
Palaoeclimate Model Intercomparison Project PMIP

• understanding the mechanisms of climate change by
  examining such changes in the past, when the
  external forcings were large and relatively well
  known and when various kinds of geological
  evidence provide evidence of what actually
  happened
• providing a framework for the evaluation of
  climate models in order to determine how far they
  are able to reproduce climate states radically
  different from that of the present day

9
PMIP I (1990-2000)

• Last Glacial Maximum (21,000 years ago). Change
  orbit, CO2, land ice sheet. Two simulations
• Prescribed Sea Surface Temperature (CLIMAP)
• Slab ocean model
• Mid-Holocene (6,000 years ago). Change orbit
  only.
• Atmosphere only simulation.

10
Failure of the GCMs in the mid-Holocene
Data from Sandy Harrison
11
Simulated changes in African monsoon Land and
ocean Feedbacks
Picture by Sandy Harrison
12
Last Glacial Maximum Tropics
Ferrara et al., 1999
Rosell-Melé et al., 1998
Harrison, 2000
Large scale patterns not too bad from slab ocean
model, but some gradients wrong and N.Atlantic.
CLIMAP SST wrong
13
LGM Coupled-Atmosphere Ocean models
Annual mean surface air temperature 21ka-0ka
ECBILT
MRI
Courtesy of C D Hewitt
14
PMIP II (2002-?)

• Same goals as PMIP I but emphasis the use of the
  same models being used for future climate change
  predictions
• Hence using atmosphere-ocean-vegetation models
• Problems about spinning-up models for LGM
• Simulations 1000 years long
• LGM and mid-Holocene time periods
• Glacial Inception and Early Holocene
• Fresh water hosing simulations

15
Example 2 Modelling Methane at LGM

• Main natural sources are from wetlands (160 Tg
  CH4 per year) and from termites (27 Tg CH4 per
  year)
• Wetland emissions depend on extent of wetlands
  and the amount of decaying material (i.e. net
  primary productivity of vegetation)
• Main sink of methane is the reaction with the
  hydroxyl radical (e.g. CH4 OH ? CH3 H2O)
• OH concentrations is influenced by reactions with
  many other compounds in the atmosphere (e.g. CO
  OH ? CO2 H)
• These include emissions of NOx (from soils and
  lightning), and organic volatile compounds such
  as isoprene (C5H8) and terpenes (C10H16) (from
  vegetation)

16
Models of the Methane Cycle
Climate (atmosphere/ocean) Model
Land Surface Hydrology
Terrestrial Vegetation Cover
Isoprene/Terpene emission Soil and Lightning NOx
Wetland area, methane emissions, Biomass burning
Terrestrial Carbon Cycle
Terrestrial Nitrogen Cycle
Atmospheric Chemistry
Co-workers David Beerling and Colin Johnson.
17
Results
18
Example 3 Rapid Climate Changes
19
Slab ocean simulations forced by
Examine how important are circulation changes in
the ocean
20
Temperature Change
21
Greenland Climate Change
22
Fresh water hosing and predictability
Sensitivity to time scale of fresh water
pulse AND Initial conditions
Strength of Atlantic THC when fresh water pulse
imposed (size and duration of pulse uncertain)
From Renssen et al. 2002
23
Cretaceous Earth System Models
Move continents High CO2 (4 x pre-ind) Reduced
solar constant. NO permanent ice
Spin-up for 100 years, followed by 5000
ocean-only, followed by another 100 years
24
Warm Cretaceous coupled ocean-atmosphere
simulation
25
Why is model so warm?

• Consider simple global mean, annual mean energy
  balance. The radiative forcing due to
• Increased CO2 8 Wm-2
• Albedo reduction 8 Wm-2
• Cloud cover changes 10 Wm-2
• In contrast, UGAMP model under similar conditions
  suggested cloud forcing of -8Wm-2

26
Runaway Greenhouse?

• Temperature does not stabilise if
• CO2 4 x pre-ind. and no solar constant change
• Temperature does stabilise (at similar to that
  shown) if
• CO2 3 x pre-ind. and solar constant reduced by
  0.6

27
Future Work

• Coupled atmos.-ocean (and carbon cycle) models
  will require long spin-ups (a few 1000 years)
• Chemistry-climate models require comprehensive
  chemistry and multi-decadal simulations.
• Non equilibrium models will require long
  (10000 years) simulations, and multi-member
  ensembles
• New Earth System components will require
  multi-physics (multi-component) ensembles because
  in many cases we are still arguing over the basic
  equations.
• New components will allow for much more rigorous
  model-data comparisons (e.g. isotopes, dust) and
  will require much better collaboration between
  modellers and data

28
Summary

• Paleo climate studies have shown the need to
  incorporate most components of the Earth system.
• Previous paleo modelling studies showed that
  models were missing key processes, but model-data
  comparisons hampered by limited knowledge of
  input boundary conditions
• New generation of Earth system models will not
  require as many input boundary conditions, hence
  testing of the models will be easier, but
  computationally very expensive.
• Many new questions can be addressed, and many
  more time periods now possible. Some are good
  tests of models. ALL are good for testing our
  understanding.
• A truly exciting time to come!