Download Earth system models of intermediate complexity

Earth system models of intermediate complexity:
Examining the past to understand the future
Andrew Weaver
School of Earth & Ocean Sciences
University of Victoria
Future Challenges for Marine Sciences in Canada
40th Anniversary Symposium
Bedford Institute of Oceanography
24-25 October 2002
Outline
• Background
• The UVic earth system climate model
• Climate change over the last glacial cycle
–
–
–
–
The Last Glacial Maximum
Dansgaard-Oeschger Oscillations & Heinrich events
Labrador Sea Water formation over the last glacial cycle
The future of LSW formation?
• Summary and Conclusions
• Future directions and challenges
Background
Scientific Reasons
There are lots of unsolved puzzles buried within the paleo proxy record
that can only be pieced together through the use of climate models
including a wide range of climate feedbacks.
Funding Agencies:
The use of climate models to examine past climate events is an
important avenue of investigation if one is to have confidence in their
use for future climate change projections.
Personal Reasons
Its fun! One gets to interact with others in a wide variety of
disciplines.
What are earth system models of intermediate complexity?
The old vision of
climate models:
The EMIC vision of
climate models:
What EMICs exist:
1. Bern 2.5D
2. CLIMBER-2
3. EcBilt
4. EcBilt-CLIO
5. IAP RAS
6. MPM
7. MIT
8. MoBidiC
9. PUMA
10. UVic
11. IMAGE 2
The UVic Coupled Model
Ocean Component
• 3-Dimensional Ocean General Circulation Model
• Based on the GFDL Modular Ocean Model (Pacanowski, 1995; Weaver and Hughes, 1996)
• 3.6° (zonal) 1.8° (meridional) resolution; 19 vertical levels
• Several locally-developed subgrid scale parametrisations
Land elevation and ocean depths in UVic coupled model
The UVic Coupled Model
Inorganic Carbon Cycle Component
• Dissolved inorganic carbon (DIC) modelled as a passive tracer
• Atmospheric pCO2 is free to evolve
• Closely follows the protocol set up by the Ocean Carbon-Cycle Model Intercomparison
Project (OCMIP) — (Orr et al., 1999)
Air sea flux of carbon dioxide at equilibrium present-day climate. Blue: carbon uptake; Red: carbon outgassing
The UVic Coupled Model
Sea Ice Component
• Dynamics uses elastic-viscous-plastic rheology (Hunke and Dukowicz, 1997)
• Multi-level thermodynamics (Bitz and Lipscomb, 1999; Bitz et al. 2000)
• Multi-category thickness distribution (Bitz et al., 2000)
The model uses a rotated coordinate system
UVic model grid showing lines of geographic latitude and longitude in the rotated coordinate system
The UVic Coupled Model
Continental Ice Sheet Component
• Continental Ice Dynamics Model (Marshall and Clarke, 1997)
• Models internal deformation of ice (creep)
(a) no basal sliding
(b) no subglacial sediment (bed) deformation (ice stream)
• Vertically-integrated mass balance equation
• 3-D momentum equations; ice rheology (Glen 1855, 1958)
• Includes a local response to isostatic adjustment (Peltier and Marshall, 1995)
• Internal thermodynamics and floating ice shelves usually not included
The UVic Coupled Model
Continental Ice Sheet Component
Northern
Hemisphere
MODEL
O BSERVED
Southern
Hemisphere
Comparison between observed and simulated present day northern and southern hemisphere ice sheets
The UVic Coupled Model
Land Surface Component
• Two versions:
1) Simple bucket Model (Manabe, 1969) modified to allow evaporation to
depend on surface roughness
2) Leaky bucket version of Hadley Centre MOSES scheme (Cox et al., 1999)
UVic model present day soil moisture when a simple modified bucket land surface model is used
The UVic Coupled Model
Terrestrial Dynamic Vegetation Component
• Hadley Centre TRIFFID dynamic global vegetation model (Cox et al. 2000, 2001)
• 5 plant functional types (PFT): broadleaf & needle leaf trees, C3 & C4 grass, shrub) in each grid
box.
• Vegetation dynamics (areal coverage, leaf area index and canopy height of each PFT) driven by
net primary productivity, which is a function of climate and CO2 (Foley et al 1996)
• Carbon fluxes derived using a coupled photosynthesis-stomatal conductance model (Cox et al.
1998)
Preindustrial (1850) UVic model needle leaf distribution (left) and from present-day IGBP observations (right). The model has not
included the consequences of human activity (deforestation into cropland/pastures etc.) whereas the IGBP data does.
The UVic Coupled Model
Atmospheric Component — A dynamic energy-moisture balance model
• Topography on land with specified lapse rate
• Snow models for land and ice
• Water vapour feedback (Thompson and Warren, 1982)
• CO2 radiative forcing specified
• Horizontal heat transport through diffusion
• Horizontal moisture transport through advection
• Precipitation occurs when relative humidity > 85%
MODEL
NCEP REANALYSIS
Preindustrial (1850) UVic model precipitation (left) and from the present-day NCEP reanalysis (right).
The needle leaf tree distribution associated with the model was shown earlier
The UVic Coupled Model
Atmospheric Component — A dynamic energy-moisture balance model
The 32 river drainage basins used in the UVic model.
The UVic Coupled Model
The wind feedback
ECMWF Reanalysis (45°N):
Near-surface air density vs
near surface air temperature
ECMWF Reanalysis (thick lines);
NCEP Reanalysis (thin lines):
Latitudinal variation of parameters
a and b in r = a + b t
The UVic Coupled Model
Evaluation of the wind feedback using CCCma AGCM fields
Zonal mean zonal wind velocity
Zonal mean meridional wind velocity
Geostrophic (frictional near equator) wind anomalies for 2 x CO2.
Red: ECMWF Reanalysis
Green: NCEP Reanalysis
Black: CCCma AGCM
The UVic Coupled Model
Coupling of the subcomponent models
• Through the exchange of latent, sensible and radiative heat fluxes
• Through the exchange of water at the air/sea, air/land or air/sea ice interface
• Through brine rejection/ice melt and heat exchange at the sea/sea ice interface
 includes a parametrisation for local convection due to brine
rejection under multi-category sea ice
• Through a wind feedback
The UVic Coupled Model — Climatology
Surface Air Temperature
Sea Surface Temperature
Sea Surface Salinity
The UVic Coupled Model — Climatology
Atlantic
Meridional
Overturning
(Sv; 1Sv=106m3s–1)
Global
Meridional
Overturning
(Sv; 1Sv=106m3s–1)
The UVic Coupled Model
Sea Ice Climatology
Northern
Hemisphere
SUMMER
Southern
Hemisphere
WINTER
The UVic Coupled Model
Age Tracer Water Mass Analysis
Location of tracer release
27°
W
Release in North Atlantic
Release in North Pacific
Release in Weddell and Ross Seas
The Climate of the 20th Century
Climate Change over the Last Glacial Cycle
Dansgaard-Oeschger Oscillations and Heinrich Events
Focus of rest of talk
Climate Change over the Last Glacial Cycle
The Last Glacial Maximum
UVic Model
CLIMBER-2
CCCma CGCM2
What NADW overturning rate are LGM proxy reconstructions consistent with?
Schmittner, Meissner, Eby, Weaver, 2002: Paleoceanography, in press
Seidov et al. (1996)
De Vernal et al. (2000)
Present day overturning rate: 21 Sv
Seidov et al. (1996)
LGM rate: 11 Sv
LGM_ADV_1A rate: 14 Sv
De Vernal et al. (2000)
LGM PD_FWF rate: 25 Sv
LGM_ADV_1 rate: 4 Sv
WHY?
SST reconstruction:
Warmer than CLIMAP;
All model results colder;
 Strongest overturning
SSS reconstruction
Fresher than Duplessy/Seidov
All model results saltier;
 Weakest overturning
Correlation and basin average error for several
realisations of NADW formation in UVic model
Meissner, Schmittner, Weaver, Adkins, 2002: Paleoceanography, in press
The Hysteresis of the North Atlantic
Overturning
Experimental set up for Analysis
Where would one expect to see the largest impact on glacial top-to-bottom age
differences due to changes in NADW formation?
• Use radiocarbon (C14) as a passive tracer with a half life of 5730 years
Difference between extreme states (weak overturning minus strong
overturning) of the top to bottom age difference. The locations of ocean
sediment cores and deep–sea coral are also shown
What NADW overturning rate are LGM proxy reconstructions consistent with?
Meissner, Schmittner, Weaver, Adkins, 2002: Paleoceanography, in press
Proxy record
+ error bars
Intersection:
model/proxy
NADW: 12.7–18.3 SV
Interpolation between equilibria
Climate Change over the Last Glacial Cycle
Dansgaard-Oeschger Oscillations and Heinrich Events
Schematic diagram taken from: Alley, 1998: Nature, 392, 335 - 337
Stability of the thermohaline circulation using fully interactive ice sheet
Schmittner, Yoshimori, Weaver, 2002: Science, 295, 1489–1493
Investigate Hysteresis behaviour:
Apply linearly varying freshwater perturbation
Sample
metastable
stadial
state
Present Day: Red
2 Stable equilibria
Glacial: Black
2 Stable equilibria
+ metastable regime
with < 10 Sv
Interstadial
state
•If interstadial forcing is turned off, LSW formation shuts down
and system moves to state with ~10Sv overturning
•Without interactive ice sheet this equilibrium is stable
• With interactive ice sheet this equilibrium is metastable
and drifts towards the off state with no NADW
formation.
• Extracting freshwater from stadial regime can cause
rapid transition to interstadial mode.
• Glacial THC more sensitive than modern THC to
freshwater perturbations.
Dansgaard-Oeschger Oscillations
A. Glacial state ice sheet
height and calving rate
B. Calving rate and mass balance
difference immediately after transition
from cold stadial to warm interstadial
Melt of
ice sheet
Increased
calving
Growth of ice sheet
Dansgaard–Oeschger Oscillations
Force the transition from a stadial to an interstadial state
Surface mass balance
(dashed) and calving
rate (solid)
Weak sustained evaporation
Strong sustained evaporation
Response of NADW
formation
Positive values indicate positive
feedback between THC
and ice sheet mass balance
Difference between
mass balance and
calving rates (solid);
Anomalies in Atlantic
surface freshwater
balance (dashed)
Dansgaard–Oeschger Oscillations
A salt oscillator:
Increased THC :
 Increased Ice sheet growth
 Increased calving (after lag of several hundred years)
 Increased freshwater discharge into North Atlantic
 Reduced THC
Essentially the EXACT OPPOSITE of the mechanism of:
Brocker et al., 1990: Paleoceanography, 5, 469–477.
The global response to changes in NADW formation rate
Surface Temperature Anomaly
QuickTime™ and a
GIF decompressor
are needed to see this picture.
500m Temperature Anomaly
QuickTime™ and a
GIF decompressor
are needed to see this picture.
Heinrich Events
Why an abrupt warming after a Heinrich event?
Assess the effects of shutting off ice berg calving following
a Heinrich event at t=0
Sea surface temperatures in Irminger Sea:
Model vs composite of proxy data
Surface air temperatures at GRIP:
Model vs composite of proxy data
Sea surface salinities in Irminger Sea:
Model vs composite of proxy data
Labrador Sea Water formation over the last glacial cycle
Cottet-Puinel, Weaver, Hillaire-Marcel, de Vernal, Clark, 2002: Quaternary Science Reviews, submitted
Hillaire-Marcel, de Vernal, Bilodeau, Weaver, 2001: Nature, 410, 1073–1077
• Micropalaeontological data and stable isotope measurements from
planktonic and benthic foraminifera in deep Labrador Sea sediment cores
• Active deep water formation in the Labrador Sea started around 7 kyr BP
• No analogue during last glacial maximum
Wood, Keen, Mitchell, Gregory, 1999: Nature, 399, 572–575
• Coupled modelling study using Hadley Centre model
• Shutdown of LSW formation as a consequence of global warming
Is there a link between these? Can we understand why?
Labrador Sea Water formation over the last glacial cycle
Cottet-Puinel, Weaver, Hillaire-Marcel, de Vernal, Clark, 2002: Quaternary Science Reviews, submitted
Preindustrial
125 kyr BP
LGM
Labrador Sea Water formation over the last glacial cycle
Cottet-Puinel, Weaver, Hillaire-Marcel, de Vernal, Clark, 2002: Quaternary Science Reviews, submitted
125 Kyr BP
Preindustrial
Labrador Sea Water formation over the last glacial cycle
Cottet-Puinel, Weaver, Hillaire-Marcel, de Vernal, Clark, 2002: Quaternary Science Reviews, submitted
Last Glacial Maximum
February
August
Labrador Sea Water formation over the last glacial cycle
Cottet-Puinel, Weaver, Hillaire-Marcel, de Vernal, Clark, 2002: Quaternary Science Reviews, submitted
6 & 12 kyr BP equilibria looked
like preindustrial equilibrium
NADW
at ~6 kyr BP
A long 21,000 year integration
The future of Labrador Sea Water formation?
Cottet-Puinel, Weaver, Hillaire-Marcel, de Vernal, Clark, 2002: Quaternary Science Reviews, submitted
What about the future?
The future of Labrador Sea Water formation?
The transient response at year 2100
Warm start at 2100
Cold start 1%/year at 2100
Cold start 2%/year at 2100
The future of Labrador Sea Water formation?
The quasi-equilibrium response at year 2700
Warm start at 2700
Cold start 1%/year at 2700
Cold start 2%/year at 2700
Summary and Conclusions
The UVic ESCM represents a versatile tool too:
1) develop intuition / test parametrisation of physical processes
2) develop and test subcomponent models
3) investigate intricacies of interacting climate feedbacks
4) explore puzzles embedded within the paleo proxy record
Summary and Conclusions
In particular we have been able to:
1) explore internal consistency of various proxy reconstructions
2) determine that an overturning 40% weaker than today is the best fit at the LGM
3) suggest that different C14 top-to-bottom age difference measurements may be inconsistent
and suffer from uncertainty in knowledge of paleo atmospheric C14/C12 history
4) determine quantitatively that the glacial THC was more unstable than the modern THC
5) quantitatively capture the salt oscillator mechanism of Broecker et al. (1990) as an
explanation of Dansgaard-Oeschger events
6) offer a plausible trigger for the abrupt warming following a Heinrich event (cessation or
significant reduction of iceberg calving)
7) determine that the Hillaire-Marcel et al. (2001) hypothesis concerning the absence of
LSW formation during the last glacial cycle and its mid-Holocene commencement is
consistent with inferences from the UVic ESCM
8) LSW formation may temporarily cease as a consequence of global warming
Future Directions and Challenges
The Fourth IPCC Scientific Assessment (~5 years)
The UVic ESCM will include:
Interactive terrestrial and ocean carbon cycle models
 Anthropogenic emissions will be specified instead of atmospheric
concentration levels
Several dynamic vegetation models (IBIS, LPJ, TRIFFIF D)
Statistical-dynamical atmospheric model
The Fifth IPCC Scientific Assessment (~10 years)
The UVic ESCM will include:
Socioeconomic model
 Policy options, technology paths, etc will be specified instead of
anthropogenic emissions
Future Directions and Challenges
One overarching conclusion throughout all of our research:
The Labrador Sea represents an extremely sensitive region to climate
change over the last glacial cycle.
It is an ideal location to concentrate observational efforts aimed at
monitoring the oceanic response to anthropogenic climate change.
A 21st century challenge to BIO, through
sustained and increased funding from the
federal government, is to ensure that this
happens.