Miren sympo17oct13

1 Challenge the future

Miren Vizcaino Dep. Geoscience and Remote Sensing, Delft University of Technology

"Icebergs from Jakobshavn Isbrae”, courtesy of Mark Drinkwater, ESA

Recent breakthrough in including ice sheets in climate models and the challenges ahead


Greenland and Antarctic ice sheets are losing mass

A Shepherd et al. Science 2012;338:1183-1189

360 Gt= 1 mm

Mean since 1992 0.59 ± 0.20 mm yr−1

•  In response to both atmospheric and ocean forcing

•  Several surface melt extremes

over GrIS in the last two decades

•  Ocean melt reduces buttressing

of AIS & GIS grounded ice


Ice sheets also modify climate (interactive) locally & globally

Through changes in •  Snow albedo: couples atmosphere and SMB, specially during melt

season •  Topography:

•  Elevation effect (thermodynamical): temperature decreases with elevation •  Local climate: katabatic winds, clouds, orographic forcing of precipitation

•  Atmospheric circulation: storm-tracks, mean circulation

•  Freshwater fluxes: •  Meridional Overturning Circulation

•  Sea-ice formation (Bintanja et al. 2013)

•  Ocean circulation in fjords and under ice shelves (meltwater feedback)

•  Land cover (ice sheet area retreat/advance): affects albedo, roughness, turbulent fluxes


Modeling climate ice interaction

Ice Flow Model

Topography & Area

Precipitation Temperature, radiation, wind,

moisture

Meltwater

Ocean melt

Snow albedo SMB

Global climate

Total mass budget= Surface Mass Balance – ice discharge to ocean


State of the art

• Very few AOGCMs have been coupled to ice sheet models (Ridley et al. 2005, Vizcaino et al. 2008,2010)

• Why should we include them? •  Integration of several components of climate system and their

interaction (e.g. ice-ocean, ice-clouds, ice-storm track, land ice-sea ice)

•  Representing feedbacks (e.g. albedo, elevation) •  Integration of different time-scales:

•  Relating past, present and future processes in the climate system

•  This helps to model validation (for instance, with “black swans”

from paleo and recent observations to identify missing or

misrepresented processes)


Challenges in coupling ice sheet and global climate models

•  Resolution •  Spatial: AOGCMs have resolution of ~100 km. Ice sheet models need higher

resolution (few kms and below) to resolve:

•  Fast ice flow

•  Atmospheric and ocean forcing: ocean melt in fjords, ocean circulation under ice

shelves, orographic forcing of precipitation, melt gradients at ice sheet margins,…

•  Temporal: long time scales involved in ice sheet build-up and complete decay

•  Ice sheet initialization: obtaining a realistic present-day ice sheet •  well adjusted to the simulated climate (to avoid non-physical background trend)

•  with memory of past climate (ice column temperature, viscosity, basal

conditions)


The SMB challenge •  High Surface Mass Balance gradients at

ice sheet margins (steep topography)

Until very recently: •  Climate models considered unsuitable:

coarse resolution (~tens of km needed), climate biases

•  Regional climate models preferred. But:

•  Lateral forcing from GCMs needed •  No direct global climate-ice coupling (e.g. no meltwater-ocean feedbacks) •  Un-suited for multi-century studies

(e.g. no elevation feedback) Surface mass balance (precip-su-melt) as

modeled by RACMO2

kg m-2 yr-1


Ice climate in RCMs vs. GCMs

Global climate

Regional climate

Ice sheet processes

Global climate

Ice sheet processes

Global climate models

Regional climate models

Ice sheet flow & surface models

albedo, elevation, meltwater feedbacks

RCMs)

GCMs)

Lateral Forcing

Snow albedo

feedback


Recent break-through: first realistic SMB from global climate model

•  Model is the Community Earth System Model (CESM) •  Results on model validation and 21st-century

projections of GrIS SMB published in J. Climate


CESM compares well (r=0.80) with in-situ observations from N=475 stations

Vizcaino et al., 2013, J. Clim.

1960-2005 SMB from CESM downscaled at 5 km & from N=475 stations (kg m-2 yr-1)

•  Accumulation rates overestimated in N interior

•  Good match in the southern part, except in the wet SE

•  67 °N, west margin (“k-transect”) •  Modeled equilibrium line

altitude (~1500 m) is close to observations

•  Small differences over 1000 m •  Gradient is underestimated:

•  Local terrain not resolved (narrow fjord framed by tundra)

•  Local anomaly in bare ice albedo (“dark zone”)


CESM compares well with RCMs

CESM RACMO2 Other RCMs (MAR/PMM5/ERA40-d)

Net SMB 359 (120) 376 (117) 288/356/287

PREC 866 (88) 723 (74) 600/696/610

MELT 568 (112) 504 (111)

Refreezing 242 (25) 245 (38)

RUN-OFF 457(95) 306 (86)

SU 54 40 5/108/38

CESM 5 km

RACMO2~11 km

SMB = PREC-RU-SU RU=MELT+RAIN-REF

r= 0.79

Gt yr-1

RACMO2, 5 km

CESM

, 5

km

1960-2005 SMB (kg m-2 yr-1)

SMB<0

SMB>0

•  Bands of precip. maxima are well reproduced •  Higher precip. in the interior & lower in SE •  Major ablation zones well captured •  Narrow SE ablation areas in both models •  Refreezing: 35% of available liquid water

(standard deviation in parenthesis)


Intra-annual mass evolution compares well with gravimetry data (GRACE) 55

995

FIG. 9. Upper) CESM seasonal cycle of snow melt, bare ice melt, refreezing, snowfall, 996

rain and sublimation. Units are Gt. Lower) Comparison of the seasonal cycle of mass 997

anomalies between CESM (blue) and GRACE (black). Units are Gt. Thick lines represent 998

nine-year averages and thin lines represent the maximum and minimum monthly 999

anomalies. Selected periods are 1996-2004 (CESM) and 2003-2011 (GRACE). 1000

1001

0 90 180 270 360day

0

1

2

3

4

5

6

7

mas

s (G

t)

1 2 3 4 5 6 7 8 9 10 11 12month

-300

-200

-100

0

100

200

300

cum

. mas

s an

omal

y (G

t)

Snow Melt

Snowfall

RainfallSU

Ice Melt

Snow Melt

RE

CESM

GRACE

CESM seasonal cycle of snowmelt, bare ice melt, refreezing, snowfall, rain, and sublimation (Gt).

Mean (thick lines) and range (thin lines) of de-trended monthly cumulative mass anomalies (Gt) for CESM (1996-2004, blue) and GRACE (2003-2011, black).

•  Similar maximum, minimum & amplitude, regardless of influence of climate variability & “different” periods (GRACE data starts later, in 2003 vs. 1996 of CESM)


The recipe for a good SMB

1.  Realistic atmospheric forcing (radiation, wind, humidity) 2.  Explicit simulation of snow processes: albedo, compaction,

refreezing. •  SMB in CESM is calculated in land component (giving “immediate”

ice-atmospheric coupling) •  Albedo depends on snow grain size, solar zenit angle, spectral

band, snow impurities (SNICAR model) 3.  Sub-grid representation of elevation dependency of SMB ①  Atmospheric forcing (temperature, humidity) is downscaled to ice

sheet grid ②  SMB is re-calculated at several fixed elevations ③  SMB maps are interpolated to ice sheet grid (horizontally &

vertically)


Explicit albedo simulation 57

1008

FIG. 11. Map of mean April (a) and July (b) albedos for 1960-2005 as simulated by 1009

CESM. (c) July albedo from RACMO2 for the same years. Elevation contours are plotted 1010

every 1000 m. Red lines delimit 50% ice sheet cover. 1011

1012

0.9

0.825

0.8

0.7

0.6

0.5

0.2

a b c

Before melt season (April)

At melt season peak (July)

CESM (global model) RACMO2 (regional model)

1960-2005 mean

albedos


GrIS SMB Projections up to 2100 with CESM

Following high-forcing scenario RCP8.5


Temperature change under RCP8.5

a!

b!

Annual, 2080-99 minus 1980-99 (K) Summer, 2080-99 minus 1980-99 (K)

Global: +3.7 K 60-90°N: +7.9 K

Greenland: +4.7 K

Greenland: +4.1 K

•  MOC reduction reduces warming SE of Greenland •  JJA increase is highest

•  In ice-free regions to N & E, due to stronger sea ice losses (>40%) along the coast

•  In the interior of the ice sheet, which remains below melting point


Change in surface fluxes (RCP8.5)

SWd! LWd! albedo!

Rnet! SHF! LHF!

Surface fluxes (Wm-2) and albedo (right color bar) anomalies 2080-99 minus 1980-99. Non-

significant values excluded

Surface fluxes (Wm-2) and albedo (right color bar) anomalies 2080-99 minus 1980-99. Non-

significant values excluded

2080-99 minus 1980-99 (Only significant anomalies) Wm-2 albedo

Vizcaino et al., 2013, J. Clim., Part II

•  More incoming LW •  Less incoming SW due to

increased cloudiness •  Albedo decreases •  Net radiation increases •  Turbulent flux increases


Change in integrated SMB terms (Gt yr-1)

1980-99 2080-99

SMB 372 (100) -78 (143)

PRECIPITATION 855 (70) 1158 (74) +35%

Snowfall 728 (59) 857 (47) +18%

SURFACE MELT 552 (119) 1186 (155) +215%

Refreezing 240 (25) 318 (25) +33%

RUN-OFF 438 (98) 1168 (168) +266%

SUBLIMATION 54 (3) 60 (4) +11%

1980 2000 2020 2040 2060 2080 2100year

-500

0

500

1000

1500

Gt p

er y

ear

MeltMelt

Snowfall

SMB

Run-off

Rain

SMB = PREC-RUNOFF-SUBLIMATION RUNOFF=MELT+RAIN-Refreezing

•  SMB becomes negative •  Snowfall increases by 18% •  Melt doubles •  Refreezing capacity decreases •  5.5 cm SLE by 2100

Vizcaino et al., 2013, J. Clim., Part II

(standard deviation in parenthesis) % indicates increase 2080-99 wrt 1980-99


New equilibrium line ~500 m higher

1980-99! 2080-99!

•  Ablation area increases from 9% to 28% of ice sheet •  Max. increase of eq. line in NE (~1000 m higher) •  SMB increases over 2000 m •  Map is similar to projections from regional models

kg m-2 yr-1

SMB<0

SMB>0


Summary

•  Ice sheets are not yet interactive components of most global climate models

•  Major challenges come from differences in spatial & temporal resolutions

•  Until recently, global climate models considered unsuited to reproduce high SMB gradients, due to climate biases and insufficient resolution

•  Now the first global climate model is able to simulate realistic SMB, due to realistic atmospheric simulation coupled with explicit representation of snow processes, and sub-grid representation of vertical gradients