Using CCSM3 to investigate future abrupt Arctic sea ice change

Marika Holland

NCAR

“A mechanism that might lead to abrupt climate change would need to have the following characteristics:

• A trigger or, alternatively, a chaotic perturbation, with either one causing a threshold crossing (something that initiates the event).

• An amplifier and globalizer to intensify and spread the influence of small or local changes.

• A source of persistence, allowing the altered climate state to last ...”

From Abrupt Climate Change: Inevitable Surprises (2002)

“an abrupt climate change occurs when the climate system is forced to cross some threshold, triggering a transition to a new state at a rate … faster than the cause.”

Role of sea ice as an “amplifier”

• Surface albedo feedback amplifies climate perturbations• Models have been used to explore/quantify these effects.

(From Hall, 2004)

VA=variable albedoFA=fixed albedo

(DJF SAT)

Observed Arctic Conditions

Sea ice concentration

(NSIDC, 2005)

(Perovich, 2000)

Fowler, 2003

Observed thicknessLaxon et al., 2003

The observed Arctic sea ice

June 6, 2005

Observations indicate large changes in Arctic summer

sea ice cover

From Stroeve et al., 2005

2002 2003 2004

1980 2000

Sept Ice Extent

Trend = 7.7% per decade

Suggestions that ice has thinned…

Rothrock et al., 1999

Ice draft change 1990s minus (1958-1976)

Indications that Arctic Ocean is warming

Polyakov et al., 2005 1900 2000

Atlantic Layer Temperature

• “Pulse-like” warming events entering and tracked around the Arctic

• General warming of the Atlantic layer

North Atlantic Oscillation Positive Phase

From University of

Reading webpage

JFM NAO Index1950-1992

Timeseries of JFM NAO Index

Maybe it is not all the NAO/AO?

Have led to suggestions that:

“Researchers estimate that in as little as 15 years, the Arctic could be ice free in the summer”

J Climate, 2005

Overpeck et al., 2005

“There is no paleoclimate evidence for a seasonally ice free Arctic during the last 800 millennia”

Overpeck et al.

Future ProjectionsWhat can models tell us?

Future climate scenarios

Meehl et al, 2005Wigley, 2000

• Relatively gradual forcing. • Relatively gradual response in global air temperature

September Sea Ice Conditions

ObservationsSimulated5-year running mean

• Gradual forcing results in abrupt Sept ice transitions

• Extent from 80 to 20% coverage in 10 years.

• Winter maximum shows • Smaller, gradual decreases• Largely due to decreases in

the north atlantic/pacific

“Abrupt”transition

Forcing of the Abrupt Change

• Change thermodynamically driven

• Dynamics plays a small stabilizing role

Change in ice area over melt season

Thermodynamic

Dynamic

• Ice melt rates directly modify the ice thickness

• Ice thickness shows large drop associated w/abrupt event

• However, change is not “remarkable”

MarchIce Thickness

Processes contributing to abrupt change

Increased efficiency of OW production for a given ice melt rate

• As ice thins, vertical melting is more efficient at producing open water

• Relationship with ice thickness is non-linear

% O

W f

orm

atio

n pe

r cm

ice

mel

t

March Arctic Avg Ice Thickness (m)

Basal Melt

Surface Melt

Total Melt

Processes contributing to abrupt change

Albedo Feedback

• Increases in basal melt occur during transitions

• Driven in part by increases in solar radiation absorbed in the ocean as the ice recedes

cm/d

ayW

m-2

SW Absorbed in OML5 Year Running Mean

Processes contributing to abrupt changeIncreasing ocean heat transport to the Arctic

Ocean Heat Transport to

Arctic

Increases in ocean heat transport occur during the abrupt transition.

Contributes to increased melting and provides a “trigger” for the event.

FramStrait

Sib

eria

n S

helf

Arctic

Arctic Ocean Circulation Changes

2040-2049Minus

1980-1999

Sib

eria

n S

helf

FramStrait Arctic

Evidence that ocean circulation changes are related to changing ice/ocean freshwater exchange

(Bitz et al., 2006)

Both trend and shorter-timescale variations in OHT appear important

OHT “natural” variations partially wind driven.

Correlated to an NAO-type pattern in SLP

Ocean Heat Transport to

Arctic

Mechanisms Driving Abrupt Transition

1. Transition to a more vulnerable state

• thinning of the ice cover

2. A Trigger

• rapid increases in ocean heat transport.

• Other “natural” variations could potentially play the same “triggering” role?

3. Positive feedbacks that accelerate the retreat

• Surface albedo feedback

• OHT feedbacks associated with changing ice conditions

Effects of transition on atmospheric conditions

• Winter air temperature increases rapidly during abrupt ice change, with a >5C warming in 10 yrs

• Precipitation shows general increasing trend with largest rate of change over abrupt ice event

Projections of Near-surface Permafrost

Courtesy of Dave Lawrence, NCAR

(Lawrence and Slater, 2005)

Ice E

xte

nt

10

6 k

m2

Permafrost (CCSM)Sept. sea-ice (CCSM)Sept. sea-ice (Observed)

Some Cautions in Using Models to Examine these (and other) issues…

Biases in simulated control state can affect feedback strength

Uncertainties in model physics/response

Acknowledgement that model physics matters for simulated feedbacks

Models provide a powerful tool for examining climate feedbacks, mechanisms, etc but…

“Ethical Considerations”

ITD Influence on Albedo Feedback

• Model physics influences simulated feedbacks• Getting the processes by which sea ice amplifies a climate signal

“right” can be important for our ability to simulate abrupt change

ITD (5 cat)1 cat.

1cat tuned“Strength” of albedo

feedback in climate

change runs

(Holland et al., 2006)

Feedbacks contribute to Arctic amplification

But, that amplification varies considerably

among models

(Holland and Bitz, 2003)

Sea ice in fully coupled GCMs

IPCC AR4

1980-1999 ice

thickness

Red line marks

observed extent

Aspects of the Model’s Internal Variability

ModelStandard Deviation

Model 1 1.93

Model 2 1.90

Model 3 1.72

Model 4 1.68

Model 5 0.42

Summary• Sea ice is an effective amplifier of climate perturbations:

• due to surface albedo changes• due to ice/ocean/atm exchange processes

• CCSM3 simulates abrupt transitions in the future ice cover• preconditioning (thinning) • trigger (ocean heat transport changes)• positive feedbacks (surface albedo; oht changes)

• Models provide a useful tool for exploring the mechanisms that result in simulated rapid climate transitions

• Never completely trust the tool • comparisons to other models; sensitivity tests; “digging” into the feedbacks, etc. can increase confidence in simulated processes

Role of sea ice as an “amplifier”

From Li et al., 2005

Insulating effect of sea ice contributes to large atmospheric response to sea ice changes.

Models are a useful tool to quantify these impacts.

SST SST

LGM ReducedIce

SAT Difference

-44

-40

-36

-32

0 10 20 30 40 50 60 70

x1000 years ago

18O

(p

er

mil

SM

OW

)

Heinrich events

Dansgaard/Oeschger oscillations

Younger Dryas

8.2 k event

-30

-40

-50

-60

Te

mp

era

ture

(C

)

GISP2, Greenland

Role of sea ice for abrupt transitions in a paleoclimate context?

(slide courtesy of Carrie Morrill)

Simulated abrupt transitions in sea iceabrupt forcing (freshwater hosing) can result in abrupt ice changes

• Sea ice changes amplify climate response• Global teleconnections can result• Longevity of these changes are an issue

Sea ice change SAT Change

(From Vellinga and Wood, 2002; Vellinga et al, 2002)

SAT Change at end of 21st century

From A1B scenario

Processes Involving ice/ocean FW exchange

In warmer climate, increased ice growth due to loss of insulating ice cover results in

• Increased ocean ventilation

• Ocean circulation changes

• Transient response

Change in Ice growth rates at 2XCO2

Change in Ideal age at 2XCO2

From Bitz et al., 2006

Change in Ocean Circulation

Yr: 40-60

Change in Ideal Age at 2XCO2

cm

How common are abrupt transitions?

Transitions defined as years when ice loss exceeds 0.5 million km2 in a year

ObsSimulated5yr running mean

September Ice Extent

“Abrupt”transition

How common are forcing mechanisms?

How common are effects?Lagged composites relative to initiation of abrupt sea-ice retreat event

Courtesy of David Lawrence, NCAR

Arctic Land Area

20th Century

21st Century

• Increased Arctic Ocean heat transport occurs even while the Atlantic MOC weakens

Do other models have abrupt transitions?Some do…

Data from IPCC AR4 Archive at PCMDI

Climate models as a useful tool for addressing ACC

As a tool to flesh out/test hypotheses or processes

• How is a climate signal amplified by sea ice interactions

• What processes influence threshold behavior in the sea ice

• How does the control climate state modify the persistence of anomalies

• How are teleconnections between high latitudes and tropics realized

Precipitation Changes

• Precipitation generally increases over the 20th-21st centuries

• Rate of increase is largest during the abrupt sea ice transition

2040-2049 minus 1990-1999

OHT and polar amplification

Change in poleward ocean heat transport at 2XCO2 conditions

Both control state and change in OHT are correlated to polar

amplification

OHT

(From Holland and Bitz, 2003)

Importance of sea ice state for location of warming

• Models with more extensive ice cover obtain warming at lower latitudes

• The location of warming can modify the influence of changes on remote locations

Importance of sea ice state for the magnitude of polar amplification

• Magnitude of polar amplification is related to initial ice thickness• With thinner initial ice, melting translates more directly into open

water formation and consequent albedo changes

Complicates paleoclimate issues since “control state” not as well known

(From Holland and Bitz, 2003)

Does it matter?•Sea ice is an important “amplifier” in the system•When a change is made in a coupled model, often the most dramatic response is in the ice covered regions•This often occurs for changes that are not polar specific - e.g. diurnal cycle stuff.•Getting the processes by which sea ice amplifies a climate signal “right” can be quite important for our ability to simulate abrupt change•These will likely include ice/ocean and ice/atmosphere interactions (ice growth/brine rejection - how it changes - seasonally, etc.; how changes influence the ocean; •Threshold behavior of the ice cover - examples: on/off of the initial ice growth (freezing temp); perennial to seasonal ice cover; perennial to seasonal snow cover;