Pathways and impact of Southern Ocean Pathways and impact of Southern Ocean currents on Antarctic Icesheet melting in currents on Antarctic Icesheet melting in

response to global warmingresponse to global warming

Frank Colberg and Nathan BindoffFrank Colberg and Nathan BindoffFrank.colberg@csiro.auFrank.colberg@csiro.au

Thanks to: Petra Heil, Ben Galton-Fenzi, Helen Philips, TPAC team

TPAC

OutlineOutline

•Introduction and MotivationIntroduction and Motivation•Model setupModel setup

–Ocean, Seaice, IceshelvesOcean, Seaice, Iceshelves–Large scale circulation impacts on Large scale circulation impacts on

seaiceseaice•Climate change runs and resultsClimate change runs and results

–Focus on seaice, freshwater fluxes, Focus on seaice, freshwater fluxes, and bottom water propertiesand bottom water properties

•Summary and future workSummary and future work

IntroductionIntroduction

Global climate models used for the latest IPCC lack adequate representation of icesheet/ shelve physics

These add to uncertainties in future estimates of sea level rise

Velicogna and Wahr (2006) decreasing ice thickness follows

the reduction of ice shelves Thoma et al., 2008 modeled

variations in the influx of CDW on Amundsen Sea Cont Shelf may act as an indicator for the ice thickness variations

J.L. Chen et al, 2009

IntroductionIntroduction

Observations suggest changes in the hydrological cycle occurred and manifest themselves as a freshening signal in the upper ocean.

Mismatch between modelled hydrological cycle and observed (Figure, Helm et al, 2009)

Freshening signal apparent in the bottom water properties (Rintoul, 2007)

Hypothesis: is it plausible that this surplus of observed fresh water may arise from ice shelve melting

Gains support from fact that only 0.5% of additional iceshelf melting is needed (Helm et al, 2009)

Helm et al., 2009

Model Setup Model Setup

• ROMS (Shchepetkin et al, 2004) Shchepetkin et al, 2004)

• 1/8 degree horizontally, circumpolar, 20S-85S

• 25 vertical levels

• ~40.000.000 gridpoints -> largest model of ROMS to date

• No flux correction due to nature of experiments

• ~13Tb data, one model output: 1.2Gb, 5 per month

• Forcing: CORE (Large and Seager, 2006)

• Initialization: WOA (Conkright, 2002)

• Currently running on TPAC cluster: katabatic 256CPU

• 1 Model year simulated in 1 day

Model setup and componentsModel setup and components

AREA OF INTEREST

CDW

• After Budgell (2004)

• Ice dynamics based on Elastic-Viscous-Plastic (EVP) rheology (Hunke and Dukowicz, 1997 and Hunke, 2001)

• Ice thermodynamics based on Mellor and Kanta (1989) and Kakkinen and Mellor (1992)

• Two ice layers and a single snow layer are used in solving the heat conduction equation

• Molecular sub-layer separates bottom and ice-cover from upper ocean

Iceshelf dynamics as specified by Hunter (2006)

Hedstroem, 2009

Iceshelf mask, cavity thicknessIceshelf mask, cavity thickness

• Smith and Sandwell and BEDMAPSmith and Sandwell and BEDMAP (Lythe et al, 2001)(Lythe et al, 2001)

• Blended at 60SBlended at 60S

• Including all major iceshelves + fringesIncluding all major iceshelves + fringes

Cavity thickness

West

Schackleton

Depth of Iceshelf

Mertz

Ross

Filchner-RonneGeorge VI

Larsen

Rieser-Larsen

Abbot

Getz

Amery

Flimbul

Substantial seaice lossSubstantial seaice loss

•Faced with one major problem: loosing seaice•Important as sea ice is a good proxy for goodness

of ocean atmosphere state•2 major reasons: •Error in seaice code fluxes had wrong sign•Related to horizontal mixing in large scale context

Model results, ice concentration

Observations/ SSMI

(1) (2)

Salinity structure and Salinity structure and meridional heat transport meridional heat transport

across 50Sacross 50SYear 1

Year 3 Year 4

Year 2

Red : Ocean model without Smagorinsky type mixing

Blue : After applying Smagorinsky mixing

Eddy kinetic energy and Eddy kinetic energy and transport across Drake Passagetransport across Drake Passage

Maximum seaice concentrationMaximum seaice concentrationObservations/ SSMI Model - September mean over 10 years

•Maximum mean ice extent as observed by •SSMI:~18E7 km2

Climate change runsClimate change runs

•Base run: 12 years Base run: 12 years •Run 1: Whole suit of IPCC anomalies Run 1: Whole suit of IPCC anomalies •Run 2: Wind anomalies onlyRun 2: Wind anomalies only•Run 3: Run 3: All IPCC anomalies but windAll IPCC anomalies but wind•To distinguish between dynamic and buoyancy forcing To distinguish between dynamic and buoyancy forcing

and feedbackand feedback•Results from this/ last week !!Results from this/ last week !!•showing difference plots of: showing difference plots of:

– Ice concentration, thicknessIce concentration, thickness– Melting ratesMelting rates– Bottom water propertiesBottom water properties

Difference in ice concentration and Difference in ice concentration and thickness, timeseriesthickness, timeseries

Black: Base RunBlue: Run 1, IPCC allRed: Run 2, IPCC, windsGreen: Run 3, IPCC, all but winds

Response due to •enhanced upwelling of CDW•increased upper ocean warming due to changes in radiation•Transport of seaice to midlatitudes

Difference in ice concentration spatial Difference in ice concentration spatial patternpattern

BASE RUN

IPCC, all

IPCC, winds

IPCC, all but winds

Meltwater response to changes in Meltwater response to changes in surface forcingsurface forcing

Pos: 93 GT/a Neg: -216 GT/a

Black: Base RunBlue: IPCC (all)Red: IPCC (wind)Green: IPCC (all, but wind)

• (a) Enhanced melting only when buoyancy forcing terms are applied in - forcing-Additional freshwater flux into ocean: ~150Gt/a

• (b) Reduced melting only when anomalous wind forcing is applied Reduction: ~50Gt/a

Annual melting rates for scenarios Melt rates: Run 1 (IPCC, all) – Base RunSpatial pattern

No clear signal

Meltwater response for region

Black: base runBlue: all IPCC forcingRed: wind IPCC forcingGreen: buoyancy IPCC forcing

I

IV

IIIII

VI

V

VII

Temperature anomalies on 1000m depth contour – yearly means

Run 1, Run 2: similar response indicating CDW intrusion onto shelf

Year 1

Year 3

Year 5

Year 7

IPCC, all IPCC, all but winds

Only when we apply anomalous wind forcing are we getting temperature anomalies that put the focus on the WAIS

Melting decoupled from CDW intrusion ?

Bottom layer temperature Bottom layer temperature and salinity changes and salinity changes

0.5

-0.5

Temperature anomaly after 9 yearsof integration

Temperature bottom layer Temperature section along ~150E

0.5

-0.5

All IPCC forcing:

Buoyancy forcing:

Salt anomaly after 9 years of integration

Salinity bottom layer Salinity section along ~150E

All IPCC forcing:

Buoyancy forcing:Strong intrusion of CDW onto shelf

(1)

Bottom water properties and possible connection to iceshelf melting ?

Similar timing for enhanced melting in Mertz shelf and cooling plus freshening of BT

Accelerates after 4-6 years

Blue: IPCC all – Base RunRed: IPCC, winds - BRGreen: IPCC, all but winds - BRMelting rate Temperature

Melting as a function of depth

Buoyancy forcing•Enhanced melting for upper 500m •No change in deeper shelves evident

2 Years 8 Years

Wind forcing•Reduced melting for upper 500m for first 4 years•Enhanced melting for 500-1000m after 4 years

Suggests:• stronger contributions of deep iceshelves for longer integration times•These may act as to change bottom water properties

SummarySummaryDeveloped state of the art ocean – seaice – iceshelf model of ROMSLarge scale circulation greatly influences seaice production and likely affects shelf processes as it controls/ modifies the poleward heat/ salt transportIceshelves overestimate melt rates to some extent when compared with previous modelling studies, however, fluxes not adjusted yetIncluding full suit of IPCC forcing results in a net increased freshwater flux from iceshelvesIncluding anomalous wind forcing results in strong warm temperature anomalies in the WAIS areaBottom water property changes may be related to modifications in melt rates from selected iceshelvesBottom water is sensitive to changes in windforcing onlyInclusion of wind anomalies introduces a 4-5 time delay

Future workFuture workAnalysesExtending model runsModel improvement:

Refine iceshelf mask Include frazil ice dynamics (see GF, 2009) Model drift needs further exploration – note:

no flux corrections were appliedVertical mixing under seaice

Thanks ... !

Shelf MASK 1 Area Mkm2 MASK 2 Area MASK 3 Area Helmer, 2004 Area

Total 1400 12.8 3139 19.02 1725 14.7 906 12.33

Amery 27 0.292 56 0.4 42 0.34 17 0.55

Ross 235 4.4575 303 4.9 263 4.7 180

Filchner-Ronne 130 3.73 274 4 269 3.9 119

Getz 60 0.25 170 0.48 137 0.41 53

Rieser-Larsen 115 0.42 153 0.64 93 0.5 38

Schackleton 57 0.23 93 0.35 44 0.28 47

Abbot 50 0.26 105 0.5 72 0.3 18

Melt rates and sensitivity to shelf Melt rates and sensitivity to shelf maskmask

Iceberg calving amounts to ~2000Gt (Jacobs et al, 1992)No heat conduction into iceshelfNo frazil ice, No tidesMay all act as to enhance melting

Galton-Fenzi, 2009

Sea Ice componentSea Ice component• After Budgell (2004)

• Ice dynamics based on Elastic-Viscous-Plastic (EVP) rheology (Hunke and Dukowicz, 1997 and Hunke, 2001)

• Ice thermodynamics based on Mellor and Kanta (1989) and Kakkinen and Mellor (1992)

• Two ice layers and a single snow layer are used in solving the heat conduction equation

• Molecular sub-layer separates bottom and ice-cover from upper ocean

Wai: melt rate at the upper ice/ snow surfaceWao: freeze rate on the upper ice/ snow surfaceWfr: rate of frazil ice growthWio: freeze rate at the ice/ water interfaceWro: rate of run-off of surface melt water

Diagram of the different location where ice melting and freezing can occur

• Iceshelf dynamics as specified by Hunter Iceshelf dynamics as specified by Hunter (2006)(2006)

• Dont use full set of thermodynamical eqs Dont use full set of thermodynamical eqs (GF, 2009)(GF, 2009)

• Iceshelf in steady stateIceshelf in steady state

• No frazil ice No frazil ice

• No heat conduction into iceshelfNo heat conduction into iceshelf

IceshelvesIceshelves

Holland and Feltham, 2004Holland and Feltham, 2004

Descending plumes of HSSW are generated by sea ice Descending plumes of HSSW are generated by sea ice brine rejection (interact and modulate CDW)brine rejection (interact and modulate CDW) It will melt the base of the iceshelf. This fresh water It will melt the base of the iceshelf. This fresh water becomes supercooled as it rises (due to decrease of becomes supercooled as it rises (due to decrease of freezing temperature with depth)freezing temperature with depth) Resulting in direct basal freezing and production of frazil Resulting in direct basal freezing and production of frazil ice ice

CDW

Galton-Fenzi, 2009