Modelling recent and future sea ice changes
An assessment of CMIP5/COMBINE results
T. Fichefet, F. Massonnet
with contributions from
M. Chevallier, P. J. Hezel, T. Koenigk, O. Lecomte,
D. Salas y Mélia, G. Vergé-Dépré and K. Wyser
September Arctic sea ice extent
million km²
October 3rd, 2013 – COMBINE Final General Assembly
(IPCC WG1 AR5, 2013)
The Arctic sea ice
clock is ticking
Annual Arctic sea ice extent anomalies
(IPCC WG1 AR5, 2013)
Antarctic sea ice variability
is more puzzling than ever
Annual Antarctic sea ice extent anomalies
1. Modelling sea ice : from CMIP3 to CMIP5
2. Sea ice projections
1. Modelling sea ice : from CMIP3 to CMIP5
2. Sea ice projections
3. Developments in sea ice modelling
1. Modelling sea ice : from CMIP3 to CMIP5
2. Sea ice projections
3. Developments in sea ice modelling
1. Modelling sea ice : from CMIP3 to CMIP5
The seasonality of Arctic sea ice extent
is better simulated in CMIP5 than CMIP3
(IPCC WG1 AR5, 2013)
1980‒1999
Average Arctic sea ice extent
OBS
J F M A M J J A S O N D
Month
CMIP3
CMIP5
1986‒2005
CMIP5 mean
NASA
The CMIP5 model spread
around the mean is still large
Average Arctic sea ice extent
(IPCC WG1 AR5, 2013)
1980‒1999
Average Arctic sea ice extent
OBS
J F M A M J J A S O N D
Month
CMIP3
CMIP5
(IPCC WG1 AR5, 2013)
February September
The CMIP5 model spread
around the mean is still large
Number of CMIP5 models simulating at
least 15% of sea ice during 1986-2005
OBS
September Arctic sea ice extent
(IPCC WG1 AR5, 2013)
Trends in September Arctic sea ice extent
are better simulated in CMIP5 than CMIP3
CMIP3
CMIP5
OBS
OBS
(NASA)
Trends in September Arctic sea ice extent
are better simulated in CMIP5 than CMIP3
September Arctic sea ice extent Distribution of CMIP5 September Arctic sea ice
extent trends (1979-2010, 66 realisations)
(IPCC WG1 AR5, 2013)
CMIP3
CMIP5
OBS
Num
ber
of
models
September sea ice extent trend
Anthropogenic influences have very likely
contributed to Arctic sea ice loss since 1979
September Arctic sea ice extent
CMIP5
Pre-industrial forcing
CMIP5
Historical forcing
OBS (NASA)
(IPCC WG1 AR5, 2013)
1986‒2005
CMIP5 mean
OBS(NASA)
1980‒1999
(IPCC WG1 AR5, 2013)
Mean Antarctic sea ice extent: noticeable
improvements, but still very large spread
Average Antarctic sea ice extent Average Antarctic sea ice extent
CMIP3
CMIP5
OBS
J F M A M J J A S O N D
Month
(IPCC WG1 AR5, 2013)
Mismatch between observed and
simulated Antarctic sea ice variability
February Antarctic sea ice extent
(Zunz et al., The Cryosphere, 2013)
CMIP5 mean
OBS (NASA)
1979-2005 standard deviation of the
detrended Antarctic sea ice extent
CMIP3
CMIP5
OBS
Mean state
Trends/
variability
Attribution/
Detection
Improved Improved
Improved Status quo
Changes detectable
and attributable Uncertain
Conclusion 1
CMIP3 CMIP5: improvements
with persistent uncertainties
Arctic Antarctic
(large spread) (large spread)
2. Sea ice projections
3. Developments in sea ice modelling
1. Modelling sea ice : from CMIP3 to CMIP5
Improvements with persistent uncertainties
2. Sea ice projections
3. Developments in sea ice modelling
1. Modelling sea ice : from CMIP3 to CMIP5
Improvements with persistent uncertainties
−8%
−34%
−43%
−94%
February September
Changes in CMIP5 Arctic sea ice extent (reference: 1986-2005)
The Arctic sea ice cover will very likely
continue to shrink as global temperature rises
(IPCC WG1 AR5, 2013)
RCP8.5
RCP4.5
Average 2081-2100 CMIP5 sea ice concentration
February September
OBS 1986-2005
(IPCC WG1 AR5, 2013)
The Arctic sea ice cover will very likely
continue to shrink as global temperature rises
(Massonnet et al., The Cryosphere, 2012)
CMIP5 mean
NASA
CMIP5 mean
NASA
The spread in summer Arctic
sea ice projections remains wide
September Arctic sea ice extent simulated by CMIP5 models
RCP4.5 RCP8.5
near ice-free near ice-free
Proportion of CMIP5 models with September Arctic
sea ice extent < 1 million km² for at least 5 years
The spread in summer Arctic
sea ice projections remains wide
Year of disappearance of summer Arctic
sea ice is linked to baseline sea ice state
(IPCC WG1 AR5, 2013 ; Massonnet et al., The Cryosphere, 2012)
A nearly ice-free Arctic Ocean in September is
likely by mid-century (high-emission scenario)
CMIP5 ensemble
5 selected models
September Arctic sea ice extent (RCP8.5)
million km²
(IPCC WG1 AR5, 2013 ; Massonnet et al., The Cryosphere, 2012)
near ice-free
(Hezel et al., in prep.)
Possible recovery of summer Arctic
sea ice if radiative forcing decreases
W/m²
Million km²
Radiative forcing
September sea ice extent
OBS
RCP8.5 (9 models)
RCP2.6 (9 models)
RCP4.5 (14 models)
March Arctic sea ice extent
Million km²
7 out of 9 CMIP5 models reach ice-free conditions
in winter by 2300 under a high-emission scenario
(Hezel et al., in prep.)
−16%
−67% −8%
−30%
February September
(IPCC WG1 AR5, 2013)
A decrease in Antarctic sea ice extent is expected
during the 21st century, but with low confidence
Changes in CMIP5 Antarctic sea ice extent (reference: 1986-2005)
Arctic Antarctic
Conclusion 2
CMIP5 offers the possibility to investigate Arctic sea
ice projections, caution has to be taken for Antarctic
Projected sea
ice extent very likely
Seasonally
ice-free
by mid-century
if high-emission scenario likely
Recovery of sea
ice in September possible if low-
emission scenario
low confidence
Annually
ice-free
clear possibility by 2300
if high-emission scenario
1. Modelling sea ice : from CMIP3 to CMIP5
Improvements with persistent uncertainties
2. Sea ice projections
CMIP5 offers the possibility to investigate Arctic sea
ice projections, caution has to be taken for Antarctic
3. Developments in sea ice modelling
1. Modelling sea ice : from CMIP3 to CMIP5
Improvements with persistent uncertainties
3. Developments in sea ice modelling
2. Sea ice projections
CMIP5 offers the possibility to investigate Arctic sea
ice projections, caution has to be taken for Antarctic
Seasonality of simulated mean sea ice extent
sensitive to freshwater and salt conservation
CNRM5.1
CNRM 5.2 +
melt ponds
CNRM 5.2
OBS (HadISST)
CNRM5.1
CNRM 5.2 +
melt ponds
CNRM 5.2
1980-1989 Average Arctic sea ice extent 1980- 1989 Average Antarctic sea ice extent
10
6 k
m²
Improved conservation
of salt and freshwater
Surface melt ponds
parameterisation (Holland et al., J. Clim., 2012)
CNRM 5.1 CNRM 5.2
OBS
Antarctic sea ice volume variability enhanced
due to stronger ocean convection variability
CNRM 5.1
CNRM 5.2
CNRM 5.2 +
melt ponds
10³ km³
September Antarctic sea ice volume
1960 1980 2000 2020 2040 2060
10
8
6
4
2
Million km²
September Arctic sea ice extent
OBS
Stream 1
Stream 2
Increased summer Arctic sea ice
sensitivity with realistic surface albedo
Surface melt ponds
parameterisation (Koltzow et al., JGR, 2007)
EC-Earth
Stream 1
EC-Earth
Stream 2
Stream 2
Stream 1
Simulated Arctic sea ice is highly sensitive
to coupling formulation with atmosphere
with LIM2
(1 ice category)
with LIM3
(1 ice category)
with LIM3
(5 ice categories,
equal fluxes)
EC-Earth Arctic sea ice extent
Implementing atmospheric coupling and
process-level developments in NEMO-LIM3
- Coupling of LIM3 (Vancoppenolle et
al., Oc. Modell., 2009) to IPSL-CM5
- Wave-ice interactions in NEMO-
LIM3 (Vancoppenolle et al., in prep.)
- Implementation in NEMO-LIM3 of a
comprehensive snow scheme
(Lecomte et al., JAMES, 2013)
- Detailed represesentation of surface
melt ponds, refreezing meltwater and
blowing snow (poster O. Lecomte)
Distribution of snow
depth on Arctic sea ice
Observed (Kwok et al., JGR, 2013)
Simulated (Lecomte et al., JAMES, 2013)
On first year ice
On multi-year ice
On first year ice
On multi-year ice
Implementing atmospheric coupling and
process-level developments in NEMO-LIM3
- Coupling of LIM3 (Vancoppenolle et
al., Oc. Modell., 2009) to IPSL-CM5
- Wave-ice interactions in NEMO-
LIM3 (Vancoppenolle et al., in prep.)
- Implementation in NEMO-LIM3 of a
comprehensive snow scheme
(Lecomte et al., JAMES, 2013)
- Detailed represesentation of surface
melt ponds, refreezing meltwater and
blowing snow (poster O. Lecomte)
COMBINE
contributions Impacts
Conclusion 3
All in all, improving sea ice physics
generally improves sea ice simulations
Snow processes
Surface processes Melt ponds and albedo
parameterisation
Conservation issues
Space+time varying snow
properties & distribution
Coupling to atmosphere
Strict conservation of salt
& freshwater in sea ice
Flux formulation
by ice category
Realistic snow
depth distribution
Decreased sea ice extent
and increased sensitivity
Increased
sea ice extent
Strong sensitivity to
type of formulation
1. Modelling sea ice : from CMIP3 to CMIP5
Improvements with persistent uncertainties
2. Sea ice projections
CMIP5 offers the possibility to investigate Arctic sea
ice projections, caution has to be taken for Antarctic
3. Developments in sea ice modelling
All in all, improving sea ice physics
generally improves sea ice simulations
~2020
CMIP6
IPCC AR6
COMBINE2?
September Arctic sea ice extent
mill
ion
km
²
Keep working and stay focused – there is a payoff!
(IPCC WG1 AR5, 2013)
NASA1
NASA2
Distribution of CMIP5 February Antarctic sea ice
extent trends (1979-2010, 66 realisations)
(IPCC WG1 AR5, 2013)
(IPCC WG1 AR5, 2013)
A B C D
Mil
lio
n k
m²
1 million km²
September Arctic sea ice extent
The one million
dollar question
A B C D
Mil
lio
n k
m²
Quadratic fit
September Arctic sea ice extent
1 million km²
The one million
dollar question
In climate change science,
never rely on your intuitions
Quadratic fit
A B C D
September Arctic sea ice extent
simulated by a CMIP5 model
Mil
lio
n k
m²
1 million km²