Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding...

23
Understanding sea level rise and variability Current projections and key uncertainties Jonathan Gregory CGAM, Department of Meteorology, University of Reading, UK Met Office Hadley Centre, UK

Transcript of Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding...

Page 1: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Understanding sea level rise and variabilityCurrent projections and key uncertainties

Jonathan GregoryCGAM, Department of Meteorology, University of Reading, UK

Met Office Hadley Centre, UK

Page 2: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Simulated global average temperature change

1900 1920 1940 1960 1980 2000-0.5

0.0

0.5

1.0

1.5

K

BCCR-BCM2.0CCSM3CGCM3.1(T47)CGCM3.1(T63)CNRM-CM3CSIRO-Mk3.0ECHAM/MPI-OMECHO-GFGOALS-g1.0GFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EHGISS-ERINM-CM3.0IPSL-CM4MIROC3.2(hires)MIROC3.2(medres)MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1

AOGCM results courtesy of PCMDI and modelling groups

Page 3: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Simulated thermal expansion

1850 1900 1950 2000-0.02

0.00

0.02

0.04

0.06

0.08

Sea

leve

l ris

e (m

)

CCSM3CGCM3.1(T47)CNRM-CM3CSIRO-Mk3.0ECHAM/MPI-OMECHO-GGFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EHGISS-ERINM-CM3.0MIROC3.2(hires)MIROC3.2(medres)MRI-CGCM2.3.2PCMUKMO-HadCM3

AOGCM results courtesy of PCMDI and modelling groups

Page 4: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Krakatoa explains some of the spread0.2

1860 1880 1900 1920 1940 1960 1980 2000Years

− 0

0

20

2

0

60

8

4

Hea

t0

onte

Ct(

10n

2J)

C

2

CMA_C GCM3C 1

CNR

_

_CM3M

SIROC MK3__

GISS

0

AOM

I

_

P_FGA ALS1O 0_G

U

_

MO_HK DCM3A

FDL_G M2_0C V)

GI

(

S_MOS EL_ED H (V)_

ISS_G ODELM E_R (_ )

MIR

V

C3_2O MEDR_ S (V)E

CAR_N CSM3C 0 (V)_

KMO_U ADCMH _V (V3

BBBB

)

BBB

Gleckler et al. (submitted MS)

Page 5: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

AOGCMs have a range of trends in recent decades

1950 1960 1970 1980 1990 2000-0.020

-0.010

0.000

0.010

0.020

0.030

0.040

Sea

leve

l ris

e (m

)

CCSM3CGCM3.1(T47)CNRM-CM3CSIRO-Mk3.0ECHAM/MPI-OMECHO-GGFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EHGISS-ERINM-CM3.0MIROC3.2(hires)MIROC3.2(medres)MRI-CGCM2.3.2PCMUKMO-HadCM3Antonov et al. (2005)

Antonov et al. (2005) 0.40±0.05 mm yr−1, AOGCMs 0.54±0.26

Page 6: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

AOGCMs have less variability than ocean temperature analyses

1950 1960 1970 1980 1990 2000-0.006

-0.004

-0.002

0.000

0.002

0.004

0.006

0.008

Sea

leve

l ris

e (m

)

CCSM3CGCM3.1(T47)CNRM-CM3CSIRO-Mk3.0ECHAM/MPI-OMECHO-GGFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EHGISS-ERINM-CM3.0MIROC3.2(hires)MIROC3.2(medres)MRI-CGCM2.3.2PCMUKMO-HadCM3Antonov et al. (2005)

Page 7: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Ocean temperature variability as a function of depth

HadCM3, there is a tendency to have weak variability

at depth, and overly strong variability in the surface

layers, especially in the South Atlantic, South Pacific,

and North Indian basins.

b. Analysis of standard deviation

Gregory et al. (2 004) compared observed low-

frequency temperature variability to the HadCM3 con-

trol run (natural internal variability only) by computing

the standard deviation of the volume-averaged, 5-yr

running mean temperature at each level from the de-

trended data (their Fig. 4). We will refer to this quantity

as SD[Avg(T)]; it is shown for observed temperatures

(detrended by removing the best fit line) as the broken

line with triangles in Fig. B4. There is a peak in ob-

served variability at 400 m, which neither model control

run reproduced (Fig. B5, right panel).

The subsurface peak in observed SD[Avg(T)] gives

the impression that were a ship to occupy a station,

there would be greater temperature variability mea-

sured at 400 m than above or below. This is not true, as

can be seen in the global average of individual standard

deviations at each point, Avg([SD(T)]. Generally,

SD[Avg(T)] � Avg[SD(T)] because significant cancel-

lation occurs when the volume-average temperature is

taken. How much cancellation depends on the spatial

coherence of temperature fluctuations, so an indication

FIG. B4. Std dev of detrended 5-yr running mean globally av-

eraged temperature as a function of depth (at dd-sampled points

only). Both curves are from observations, but differ in the way

they were computed: black triangles, std dev of globally averaged

temperature; white squares, global average of the std dev of tem-

perature at each point. Curves are normalized by their maximum

value.

FIG. B5. Std dev of detrended (removal of best-fit line) 5-yr running mean globally averaged

temperature as a function of depth, comparing observations (black dots), HadCM3 control

run (gray-shaded region shows the ensemble range), and PCM control run (cross-hatched

region shows ensemble range): (left) computed as the global average of the std dev of tem-

perature at each point and (right) computed as the std dev of globally averaged temperature.

All data is at dd-sampled points only.

1898 J O U R N A L O F C L I M A T E V OLUME 19

Pierce et al. (2006)

Page 8: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Simulated temperature change under scenario A1B

2000 2020 2040 2060 2080 2100

0

1

2

3

K

CCSM3CGCM3.1(T47)CGCM3.1(T63)CNRM-CM3CSIRO-Mk3.0ECHAM/MPI-OMECHO-GFGOALS-g1.0GFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EHGISS-ERINM-CM3.0IPSL-CM4MIROC3.2(hires)MIROC3.2(medres)MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1

AOGCM results courtesy of PCMDI and modelling groups

Page 9: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Simulated thermal expansion under scenario A1B

2000 2020 2040 2060 2080 21000.0

0.1

0.2

0.3

0.4

Sea

leve

l ris

e (m

)

CCSM3CGCM3.1(T47)CNRM-CM3CSIRO-Mk3.0ECHAM/MPI-OMECHO-GGFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EHGISS-ERINM-CM3.0MIROC3.2(hires)MIROC3.2(medres)MRI-CGCM2.3.2PCMUKMO-HadCM3

AOGCM results courtesy of PCMDI and modelling groupsModel spread is due to differences in climate sensitivity and ocean heat uptakeefficiency.

Page 10: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Geographical pattern of simulated sea level change

180 150W 120W 90W 60W 30W 0 30E 60E 90E 120E 150E 18090S

75S

60S

45S

30S

15S

0

15N

30N

45N

60N

75N

90N

-0.2 -0.1 -0.05 -0.02 0.02 0.05 0.1 0.2

0.5

0.5

0.50.5

0.5

0.50.5

0.5

0.50.5

0.5

0.50.5

0.5

0.50.5

0.5

0.5

0.5

0.5

1

1

1

1 11 1

1 11

111

1

1

Colours in m AOGCM results courtesy of PCMDI and modelling groups

Page 11: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Simulation of glacier surface mass balance change

0

20

40

60

80

Gla

ciat

edar

ea/1

03km

2

Glacier contribution to rate of sealevel rise is calculated as

− 1

Ao

∑i

biAi∆Ti

(Ao ocean area, bi, Ai, ∆Ti mass bal-ance sensitivity, area and tempera-ture anomaly for region i). If

∆Ti = Ri∆T − θi

(∆T global temperature anomaly),the rate can be written

bg(∆T − θ).

Zuo and Oerlemans (1997)Gregory and Oerlemans (1998)

Page 12: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Global glacier contribution to sea level rise

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

0

5

10

15

Cum

ulat

ive

glac

ier

mas

s lo

ss (

SLE

, mm

)

Dyurgerov and Meier (2005) (excluding Antarctica and Greenland)GISTEMP + 0.15 KHadCM3 all-forcings ensemble meanIPSL-CM4 anthropogenic forcings

Calculated from observed (GISTEMP) and AOGCM-simulated temperatures, com-pared with an observational estimate.

Page 13: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Global glacier mass balance as a function of ∆T

-0.5 0.0 0.5 1.0Global average surface air temperature anomaly (K)

0.0

0.2

0.4

0.6

0.8

1.0

Rat

e of

gla

cier

mas

s lo

ss (

SLE

, mm

yr-1

) 0.44 mm yr-1 K-1 calculated from GISTEMP + 0.15 K 0.54 mm yr-1 K-1 Dyurgerov and Meier (2005)

Page 14: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Effect of glacier hypsometry

0.5 m

Time

Sea

leve

l ris

e

evolving hypsometry(e.g. Raper and Braithwaite, 2006)

(e.g. Van de Wal and Wild, 2001)

area−volume scalingconstant area

(e.g. Zuo and Oerlemans, 1997)

Page 15: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Simulation of ice-sheet surface mass balance changeHuybrechts et al. (2004) scaled geographical patterns of temperature and pre-cipitation change from high-resolution climate models according to the ice-sheetarea-average changes from AOGCMs.Gregory and Huybrechts (in press) repeated this with five high-resolution models.The high-resolution climate change is input to a 20-km ice-sheet mass balancecalculation.Thus we obtain ice-sheet surface mass balance perturbation (SLE) as a functionof ice-sheet area-average temperature and precipitation change.

Page 16: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Ensemble-mean high-resolution GCM patterns

Annual precipitation change (%) Summer temperature change (K)(10% area-average) (1 K area-average)

0 1000 2000 3000 4000 5000km

0

1000

2000

3000

4000

5000

km

-2 0 2 4 6 8 10 12 14 16 18

2

2

22

2

2

22

2

4

4

44

4

4

46

68

0 500 1000 1500km

0

500

1000

1500

2000

2500

km

0.4 0.6 0.8 1 1.2 1.4

0.2

0.2

0.2

0.2

Page 17: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Precipitation rises linearly with temperature

0 2 4 6Antarctica temperature change (K)

0

10

20

30

40

Ant

arct

ica

prec

ipita

tion

chan

ge (

%)

cccma_cgcm3_1_t47ncar_pcm1ipsl_cm4

Page 18: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Warming threshold for negative surface mass balance in Greenland

0 2 4 6 8K

0

5

10

15

20

25

Fre

quen

cy (

in 0

.5 K

bin

s)

GreenlandGlobal

Page 19: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Future evolution of the Greenland ice sheetYear 0

Volume 100%Year 270

Volume 80%Year 710

Volume 60%Year 1130

Volume 40%Year 1760

Volume 20%

0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500 3000Bedrock altitude (m) Ice thickness (m)

Ridley et al. (2005)Not including any effects tending to produce rapid dynamical acceleration.

Page 20: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Dynamical changes in the West Antarctic ice sheet

zone as the ‘steepening’); and the main ‘trunk’ of the icestream in which basal and lateral drag are equally important(Figure 2, middle). The stress regime in the trunk is similarto that operating in the Siple Coast ice streams implying asubstrate of similarly weak, subglacial sediments. Thesteepening, however, is characterized by very much higherbasal drag implying that the substrate in this area is stronger(either hard bedrock or over-consolidated sediment).[11] We now perturb the calibrated flow model in two

grounding-line retreat scenarios, and in a third scenario inwhich the basal drag coefficient in the ice plain is reduced(simulating the effect of the reduced bed contact consequenton thinning and floatation). The snapshot results imply thatthe immediate mechanical effect of these perturbations isfelt up to 100 km from the grounding line (Figure 2,bottom). This direct effect is brought about by accelerationas a reduction in basal/ lateral drag is transmitted vialongitudinal stress gradients upglacier. The changes invelocity produced in these experiments are consistent(�30% ) with observations. The predicted accelerationscannot, however, explain the observed thinning over theentire extent of the ice stream (up to �200 km upstreamof the grounding line) and, therefore, imply that an ocean-based trigger is unlikely.

4. Results: Delayed Response

[12] Faced with this apparent anomaly, we now investi-gate the effects of the perturbed velocity field on PIG’sthickness. We do this in a time-dependent fashion by

allowing the calculated stress and velocity distributions toaffect the temporal evolution of ice thickness within themodel, as well as allowing changing ice-surface geometry toaffect gravitational driving in the stress model. Results fromthe three-dimensional model suggest that PIG experiencesplug flow (no vertical shear) over the majority of its length(vertical shear accounts for a maximum of �17% of totalflow at the steepening, Figure 2, top). This observationjustifies the use (primarily for reasons of numerical effi-ciency and stability) of a vertically-integrated stress model(which assumes minimal vertical shear [e.g., MacA yeal,1989]) in the time-dependent calculations described here.Further, the vertically-integrated version of the model gen-erates results that are consistent with the three-dimensionalresults for the snapshot perturbation experiments discussedabove.[13] In the context of the long response times traditionally

accorded to ice sheets, the results are surprising becausethey indicate that the initial anomaly can propagate farinland on very short, decadal timescales (Figure 3). Furthernumerical experiments indicate that the speed of this up-stream propagation is a robust feature of our model and, wesuggest, of the real PIG system.[14] The transmission of a perturbation in ice thickness

through an ice mass can be represented as the sum ofsource, kinematic-wave and diffusive responses [ N ye,1963]. Based on the observations for PIG, we estimate thatthe equivalent response times for the kinematic-wave anddiffusion terms are �130 and �20 yr, respectively (using alength scale of 200 km, a downstream kinematic wavespeed of 1.5 km yr�1 and a diffusivity of 2000 km2 yr�1;

Figure 2. (top) Centreline geometry of the PIG modelshowing ice-surface velocity in calibrated model (in pinkwith interferometry observations as red dots and modelledbasal velocity in light blue). Increased surface velocitiesgenerated in the perturbation experiments are also shown(red reduced ice plain basal drag, blue and green groundingline retreat). V ertical dashed line indicates the position ofthe grounding line. (middle) Modelled force balance in thecalibrated model. (bottom) Instantaneous thinning ratesproduced by the perturbations shown with the observedrates (dotted line with error bars from altimetry).

Figure 3. Cumulative changes in ice thickness generatedby a reduction in basal drag in the ice plain. (top) In the first20 years after the perturbation, showing a diffusive thinningin the grounded ice (blue) and an kinematic thickening inthe ice shelf (red). (bottom) Evolution towards a nearlysteady-state condition of �8 m thinning in the trunk of PIGafter 150 yr (coloured lines refers to times shown in legend,dashed lines spaced one decade apart for interveningperiods). Plots refer to the response at PIG’s centrelineand dashed vertical line refers to the grounding line. Above260 km, the response is muted because this point marks theboundary between the fast-flowing PIG and slow-flowinginland ice.

L23401 PAYNE ET AL.: THINNING OF LARGEST WEST ANTARCTIC ICE STREAM L23401

3 of 4

zone as the ‘steepening’); and the main ‘trunk’ of the icestream in which basal and lateral drag are equally important(Figure 2, middle). The stress regime in the trunk is similarto that operating in the Siple Coast ice streams implying asubstrate of similarly weak, subglacial sediments. Thesteepening, however, is characterized by very much higherbasal drag implying that the substrate in this area is stronger(either hard bedrock or over-consolidated sediment).[11] We now perturb the calibrated flow model in two

grounding-line retreat scenarios, and in a third scenario inwhich the basal drag coefficient in the ice plain is reduced(simulating the effect of the reduced bed contact consequenton thinning and floatation). The snapshot results imply thatthe immediate mechanical effect of these perturbations isfelt up to 100 km from the grounding line (Figure 2,bottom). This direct effect is brought about by accelerationas a reduction in basal/ lateral drag is transmitted vialongitudinal stress gradients upglacier. The changes invelocity produced in these experiments are consistent(�30% ) with observations. The predicted accelerationscannot, however, explain the observed thinning over theentire extent of the ice stream (up to �200 km upstreamof the grounding line) and, therefore, imply that an ocean-based trigger is unlikely.

4. Results: Delayed Response

[12] Faced with this apparent anomaly, we now investi-gate the effects of the perturbed velocity field on PIG’sthickness. We do this in a time-dependent fashion by

allowing the calculated stress and velocity distributions toaffect the temporal evolution of ice thickness within themodel, as well as allowing changing ice-surface geometry toaffect gravitational driving in the stress model. Results fromthe three-dimensional model suggest that PIG experiencesplug flow (no vertical shear) over the majority of its length(vertical shear accounts for a maximum of �17% of totalflow at the steepening, Figure 2, top). This observationjustifies the use (primarily for reasons of numerical effi-ciency and stability) of a vertically-integrated stress model(which assumes minimal vertical shear [e.g., MacA yeal,1989]) in the time-dependent calculations described here.Further, the vertically-integrated version of the model gen-erates results that are consistent with the three-dimensionalresults for the snapshot perturbation experiments discussedabove.[13] In the context of the long response times traditionally

accorded to ice sheets, the results are surprising becausethey indicate that the initial anomaly can propagate farinland on very short, decadal timescales (Figure 3). Furthernumerical experiments indicate that the speed of this up-stream propagation is a robust feature of our model and, wesuggest, of the real PIG system.[14] The transmission of a perturbation in ice thickness

through an ice mass can be represented as the sum ofsource, kinematic-wave and diffusive responses [ N ye,1963]. Based on the observations for PIG, we estimate thatthe equivalent response times for the kinematic-wave anddiffusion terms are �130 and �20 yr, respectively (using alength scale of 200 km, a downstream kinematic wavespeed of 1.5 km yr�1 and a diffusivity of 2000 km2 yr�1;

Figure 2. (top) Centreline geometry of the PIG modelshowing ice-surface velocity in calibrated model (in pinkwith interferometry observations as red dots and modelledbasal velocity in light blue). Increased surface velocitiesgenerated in the perturbation experiments are also shown(red reduced ice plain basal drag, blue and green groundingline retreat). V ertical dashed line indicates the position ofthe grounding line. (middle) Modelled force balance in thecalibrated model. (bottom) Instantaneous thinning ratesproduced by the perturbations shown with the observedrates (dotted line with error bars from altimetry).

Figure 3. Cumulative changes in ice thickness generatedby a reduction in basal drag in the ice plain. (top) In the first20 years after the perturbation, showing a diffusive thinningin the grounded ice (blue) and an kinematic thickening inthe ice shelf (red). (bottom) Evolution towards a nearlysteady-state condition of �8 m thinning in the trunk of PIGafter 150 yr (coloured lines refers to times shown in legend,dashed lines spaced one decade apart for interveningperiods). Plots refer to the response at PIG’s centrelineand dashed vertical line refers to the grounding line. Above260 km, the response is muted because this point marks theboundary between the fast-flowing PIG and slow-flowinginland ice.

L23401 PAYNE ET AL.: THINNING OF LARGEST WEST ANTARCTIC ICE STREAM L23401

3 of 4

Payne et al. (2004)

Page 21: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

SummaryAOGCM output can be used to simulate sea level rise due to thermal expansionand land ice changeScenario and climate sensitivity are major uncertainties for making projections

Uncertainties—thermal expansionGlobal ocean heat uptake varies among models, reflecting differences in interiortransport processesGeographical patterns of sea level change show substantial differencesSimulated decadal variability of ocean heat content is smaller than in observa-tional analyses

Uncertainties—glaciers and ice capsTreatment of the evolution of hypsometry and the “unperturbed” stateMass balance sensitivity to temperature (in general, to climate change)Geographical patterns of temperature and precipitation changeRole of rapid dynamical response to climate change

Page 22: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Uncertainties—ice sheetsMagnitude of increase in precipitationGeographical and seasonal pattern of change in temperature over the ice sheetsPossibility of lubrication of ice flow caused by surface runoffDependence of basal melting of ice shelves on ocean climate changeVulnerability of ice shelves to surface climate changeDynamical response of ice streams to melting at the grounding line and toremoval of ice shelvesRetreat of grounding line as a feedback on dynamical changePossibility of developing ice streams in slow-moving areas of ice sheets

Uncertainties—totalModels of the major terms (thermal expansion and glaciers) give results agreeingreasonably well with corresponding observational estimates for the 20th century—but they don’t add up to ∼1.5 mm yr−1!

Page 23: Understanding sea level rise and variability Current ...jonathan/talks/paris0606.pdfUnderstanding sea level rise and variability Current projections and key uncertainties Jonathan

Simulation of historical sea level change

1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000-0.030

-0.020

-0.010

-0.000

0.010

0.020

0.030

0.040

0.050

0.060

Sea

leve

l cha

nge

(m)

ALL250NAT500Glacier contribution

Gregory et al. (in press)