Wind effects on Circumpolar Deep Water intrusions on the West Antarctic Peninsula continental shelf Mike Dinniman John Klinck Center for Coastal Physical Oceanography Old Dominion University February 25, 2010

Outline of Presentation Introduction Description of circulation model Timing of Circumpolar Deep Water (CDW) intrusions on the WAP “Downscaling” simulations: CDW transport and ice shelf basal melt Conclusions

Introduction Relatively warm, nutrient-rich CDW found at shelf break around most of Antarctica CDW exchange provides heat and nutrients that can stimulate primary production Changes in either the temperature or quantity of warm CDW entering the ice shelf cavities and increasing the drainage rate of outlet glaciers into the Amundsen Sea have been proposed as a reason for the increased volume loss of the West Antarctic Ice Sheet Modifications in the wind forcing have been proposed as a mechanism for changes in the flux of CDW onto the continental shelf and to the cavities beneath the ice shelves

Antarctic Peninsula Model ROMS: 4 km horizontal resolution, 24 levels ROMS: 4 km horizontal resolution, 24 levels Ice shelves (mechanical and thermodynamic) Ice shelves (mechanical and thermodynamic) Dynamic sea ice Dynamic sea ice Bathymetry: ETOPO2v2 + WHOI SOGLOBEC region + Padman + BEDMAP + Maslanyj Bathymetry: ETOPO2v2 + WHOI SOGLOBEC region + Padman + BEDMAP + Maslanyj Daily (or better) wind forcing from a blend of QSCAT data and NCEP reanalyses or Antarctic Mesoscale Prediction System (AMPS) winds Daily (or better) wind forcing from a blend of QSCAT data and NCEP reanalyses or Antarctic Mesoscale Prediction System (AMPS) winds Lateral boundary conditions from SODA ocean reanalysis and SSM/I ice concentrations Lateral boundary conditions from SODA ocean reanalysis and SSM/I ice concentrations Experiments w/ dye representing CDW Experiments w/ dye representing CDW

Previous view of WAP intrusion frequency: 4- 6 events/year However, this was based on broad scale hydrographic surveys Detided, low-pass filtered time series of potential temperature at two moorings within Marguerite Trough => shows much higher intrusion frequency Moffat et al, 2009 Solid line: A2 – north side of MT Dashed line: A3 – south side of MT

Model histogram of the intrusion duration (55 dye unit- Sv. threshold) for the entire model run Model: 2.1 events/month (3.0 events/month w/ 45 dye unit- Sv. threshold) Typical duration: 1-4 d Observed histogram of the intrusion duration A2: 3.8 events/month Typical duration: 1-3 d Definition of intrusion is arbitrary, but model is much closer to mooring data than previous survey based picture of 4-6 intrusions per year

Duration of intrusions suggests wind forcing (weather band frequency: 2-8 days) No significant correlation between wind perpendicular to flux “gate”, but significant lagged correlation for wind parallel to cross-section: r = 0.58 (wind leads dye flux by 2-3 days) Red: Pseudo-stress (x 1/2) parallel to CDW-dye flux cross section Blue: CDW-dye flux anomaly (reversed sign)

Wind Change Simulations Positive Southern Annular Mode leads to stronger westerlies (and more mesoscale cyclones?) in this area Simulate regional effects of this with modified winds ACC transport in a regional model must be imposed: Simulate global effects by increasing the ACC transport Simulations run for 3 years that correspond to Reference case using current conditions (QSCAT/NCEP) Winds are simply scaled with both components multiplied by a constant factor ACC transport increased by sharpening the T and S gradients in the ACC fronts on the lateral open boundaries

Stronger winds and ACC provide more CDW to open shelf, but not necessarily to the cavity beneath GVI Ice Shelf Dye concentration for open shelf near GVI Ice shelf Dye concentration beneath GVI Ice Shelf

Melt rate beneath George VI Ice Shelf Stronger winds reduce melt rate beneath GVI (stronger ACC makes little difference) Why?

Dye concentration Temperature Base windsWind X 1.2 South GVI entrance: Feb. 2001

Conclusions Intrusions of CDW on to the shelf have a shorter timescale then previously thought and the intrusion timescales seem to be related to wind events Intrusions of CDW on to the shelf have a shorter timescale then previously thought and the intrusion timescales seem to be related to wind events Increases in winds and ACC transport lead to increases in the amount of CDW advected onto the continental shelf Increases in winds and ACC transport lead to increases in the amount of CDW advected onto the continental shelf However, increases in winds and ACC transport do not necessarily lead to increased CDW flux underneath the ice shelves or increased basal melt However, increases in winds and ACC transport do not necessarily lead to increased CDW flux underneath the ice shelves or increased basal melt

Acknowledgements BEDMAP data courtesy of the BEDMAP consortium Other bathymetry courtesy of Laurie Padman and Tom Bolmer AMPS winds courtesy of John Cassano Computer facilities and support provided by the Center for Coastal Physical Oceanography Financial support from the U.S. National Science Foundation (ANT ).

Tmax below mixed layer for the West Antarctic Peninsula (WAP) shelf Moffat et al., 2009

Potential temperature and salinity from the SO4P WOCE section (includes the WAP shelf). Moffat et al, 2009

Ice Shelf Modeling Ice Shelf does not change in time in model Ice Shelf does not change in time in model Three equation viscous sub-layer model for heat and salt fluxes (Holland and Jenkins, 1999) Three equation viscous sub-layer model for heat and salt fluxes (Holland and Jenkins, 1999) – Conservation of Heat – Conservation of Salt – Freezing point of sea water (function of pressure and salinity) PGF calculation assumes the ice shelf has no flexural rigidity and pressure at the base comes from the floating ice PGF calculation assumes the ice shelf has no flexural rigidity and pressure at the base comes from the floating ice

Dye concentration at depth (model level 4): 9/15/2003 – 9/15/2005

Model average velocity (net through flow: 0.14 Sv.) Estimated velocity from observations (Jenkins and Jacobs, 2008; net through flow: Sv.) Model mean basal melt rate (GVI): 6.0 m/yr Estimated observed basal melt rate: 2.1 m/yr (Potter and Paren, 1985) 2.8 m/yr (Corr et al., 2002) m/yr (Jenkins and Jacobs,2008)

Left: Gate for measuring CDW intrusions (blue) and upstream cross-section (green-red) Right: Model dye flux across Marguerite Trough “gate” and intrusion thresholds

“Intrusion” flowNon-“Intrusion” flow No significant correlation between upstream volume flux along continental slope and dye flux, but significant lagged correlation for inner along slope volume flux and dye flux (r = 0.74, volume flux leads dye flux by 2 days) Physical Mechanism?

Dye concentration (Layer 4 - Feb. 2002) 20% decrease in wind speed 20% increase in wind speed

Summer sea ice distribution (Feb. 2002) 20% decrease in wind speed 20% increase in wind speed

Melt rate beneath Wilkins Ice Shelf Highly seasonal melt rate, but stronger winds increase melt rate beneath Wilkins (ACC makes little difference) Why?

Base winds Wind X 1.2 South GVI entrance temperature: Feb. 2001

CDW as a source of Fe? Assume: Offshelf CDW dissolved Fe concentration = 0.5 nM C:Fe algal assimilation ratio = 450,000 mol/mol Scaling from the dye flux into the top 100m of the open shelf areas, the Fe flux can support primary production on the WAP shelf of: Growing season (11/1-2/28) Entire year Base: 0.5 gC/m 2 /yr 7.6 gC/m 2 /yr Wind*0.8: 0.0 gC/m 2 /yr 1.2 gC/m 2 /yr Wind*1.2: 4.9 gC/m 2 /yr 19.7 gC/m 2 /yr Wind*1.2+: 7.6 gC/m 2 /yr 23.7 gC/m 2 /yr Wind*1.5: 20.2 gC/m 2 /yr 48.7 gC/m 2 /yr