1. Modelling sea ice salinity: 1D, 3D modelling and implications for ecosystems Martin Vancoppenolle coll.: T. Fichefet, H. Goosse, C.M. Bitz, S. Bouillon, G. Madec, M.A. Morales Maqueda, J.-L. Tison, C. Lancelot, B. Delille, F. Jardon, F. Vivier

2. Approach and questions  What aspects of sea ice desalination are relevant for large-scale sea ice mass balance and ocean circulation?  How to represent ice salinity in models ?  What are the implications for ecosystem models ?

3. Why modelling ice salinity ? Simulated change in Arctic sea ice thickness (1979-2006) Variable salinity – S=5 Albedo + 10% Variable salinity – MY profile Vancoppenolle et al., Ocean Modelling, 2009b

4. Thermal properties of sea ice Thermal properties Diffusion of heat Growth / melt rates

5. Ice-ocean salt / freshwater exchanges drainage growth / melt Growing ice Melting ice Snow ice Congelation Brine drainage Melt Brine drainage Lateral melt water flow Flushing Freshwater flux / Salt flux Virtual salt flux (upper ocean salinity tendency)

6. Outline  1D modelling  3D modelling  Perspectives for ecosystems

7. 1D MODELLING

8. Is Thermal Effect Important ? S (ppt) h (m) Vancoppenolle, Fichefet and Bitz (GRL 2005) Sea ice model with brine thermodynamic effect (Bitz and Lipscomb, 1999) Run for 50 years using climatological forcing Using with various salinity profiles

9. How we model ice salinity (1D) ?  Thermal equilibrium  All salt is locked within brine inclusions  Salt transport breaks equilibrium   = freezing point salinity- temperature ratio [0.054 °C/ppt]  e = brine volume frac. [-]  S = ice bulk salinity [ppt]  T = ice temperature [°C]   = brine salinty [ppt]  v z = percolation velocity (function of meltwater production) [m/s]  D  = salt diffusivity in brine [m 2 /s]

10. Parameterizing Brine Drainage Gravity Drainage  Convection if Ra is >5 (Notz & Worster, 2008) Flushing  Percolation if  Surface melting  min(e) > 5%  30 % of meltwater percolates (Vancoppenolle, Bitz and Fichefet, JGR 2007) Vancoppenolle, Goosse, et al. (JGR 2010)

11. 1d Simulations of Brine Drainage Multiyear desalination of Arctic sea ice (Vancoppenolle, Fichefet and Bitz, JGR 06) Desalination of Antarctic sea ice (Vancoppenolle, Goosse et al., JGR 10) Summer desalination of Arctic sea ice (Vancoppenolle, Bitz and Fichefet, JGR 07)

12. Winter desalination is sensitive to diffusivity parameterization Turbulent diff. Model (solid) Obs (dash) Molecular Diff. Winter salinity profile Data from Tison et al.

13. Winter desalination is sensitive to diffusivity parameterization Antarctic sea ice simulations, June.

14. Summer desalination is sensitive to model parameters Simulated surface salinity at Point Barrow (Alaska), June, for several sensitivity exps. Penetration of penetrating radiation Fraction of vertical percolation Brine volume fraction permeability threshold 0 0.3 0.5 0.2 0.3 0.5 10% 8% 5%

15. Sensitivity to parameters and forcing: summary  Gravity drainage  Diffusivity parameters  Critical rayleigh number  Turbulent diffusivity  Flushing  Snow depth  Superimposed ice formation  Penetration of radiation (i o ) in the ice  Fraction of meltwater that is allowed to percolate vertically  Brine volume fraction permeability threshold

16. Full-depth convection Results of a run in Antarctic sea ice with high snowfall Sep 17 (1 – thin solid); Sep 24 (2 – dot); Oct 1 (3 – solid thin); Oct 8 (dash) 1 – 2 – 3 – 4 4 – 3 – 2 – 1 1 – 2 – 4 – 3 4 – 3 – 2 - 1

17. 3D MODELLING

18. Approach Problems: – Ice thickness categories – Advection of tracers is expensive Approach: develop a simple S equation – 1 equation for vertical mean salinity – Shape of profile function of vertical mean salinity Include this simple S equation LIM – Salt content (S.h) for each ice category – Horizontal transport (Prather, 86) – Remapping in thickness space (Lipscomb, JGR01) – Ridging / Rafting Vancoppenolle et al. (OM 2009a, 2009b) Comparison at Point Barrow (AK) Red diam: OBS Solid: Simple eq. Dash: Transport eq.

19. Simulated Hemispheric Mean Salinity with a 3D Ice-Ocean Model Forced by Reanalyses Hemispheric mean ice salinity simulated by NEMO-LIM3: Arctic (black) and Antarctic (grey). (différences hémisphériques : percolation, glace blanche, âge de la glace + cycle saisonnier, variablité interannuelle)

20. Comparison to Obs Ice salinity vs thickness in the model and from Cox and Weeks (1974) regressions computed using ice core data S (‰) h i (m) Mar Jun Sep Dec Comparison to observations in various regions (compilation from > 1000 cores)

21. Geographical Distribution  Arctic  Winter maximum  In winter, salinity reflects thickness / age of sea ice  Summer low / constant values  Antarctic  Weakear contrasts  Fall maximum  Local maxima due to polynyas and maximum snow fall Mar Sep Sep.Mar Ice salinity (1979-2006) averaged over ice thickness categories

22. Impact (Arctic)  2 configurations  Prescribed salinity  Simulated salinity  Ice thicker for varying salinity  Differences up to 1m  Difference in volume ~ 10%  More ice growth at the ice base (due to reduced energy of melting)  More surface melt (due to reduced specific heat)  More bottom melt (enhanced ice-albedo feedback) Varying – prescribed salinity Vancoppenolle et al., 2009b 1979-2008 Annual mean thickness difference

23. Impact (Antarctic) 2 configurations  Prescribed salinity  Simulated salinity Ice thicker for varying salinity  Mean volume difference ~ 20% Importance of ice-ocean interactions Including variations of S  Induce more ice formation with less salt rejection  Reduces vertical mixing in the upper ocean  Reduces the oceanic heat flux  Increases sea ice formation Varying – prescribed salinity Mean 1979-2009 Ice thickness difference Vancoppenolle et al., 2009b

24. Impact on Upper Ocean Mean 1979-2006 difference in sea surface salinity Varying - prescribedVarying – prescribed S

25. ECOSYSTEM MODELLING

26. Why modelling salinity ? 2) Ecosystems  Biophysical couplings associated with ice salinity  Nutrient distribution  Diatom transport mode  Brine salinity inhibition of growth  Brine volume fraction Vertical structure of ice ecosystems

27. Modelling Sea Ice Ecosystems  Energy-conserving thermodynamics and salt transport  1-stream Beer law with attenuation by chlorophyll-a  Tracer transport  Ecosystem model (diatoms and silicates)

28. Brine-Ecosystem Coupling  Ecosystem variables (n=1, 2) (diatoms=DAF, silicates=DSi)  C = bulk concentration   = brine concentration  e = brine volume fraction  Evolution equation  Physical sources & sinks

29. A typical run

30. Comparison to Observations December Chlorophyll-a profile solid black: simulated (dots = STD) horizontal lines: obs at ISPOL red : double snow After calibration of µ,, w 

31. Role of diatom transport mode  Different scenarios:  (a) diatoms follow brine motion  (b) diatoms are locked in ice without following brine motion  (c) diatoms follow brine motion and stick on brine inclusion walls  Observations  Highest biomass is achieved if algae move with brine and stick on brine’s walls.  If algae are not sticky but mobile, they are rejected of the ice as salt, which inhibits the community development.  The nutrient pump increases the availability of nutrients and hence promotes community development. Total biomass (mg chl-a/m 2 ) as a function of time (months) for the different scenarios of brine-biology interactions. Note the log scale on the y axis.

32. Role of convection on biomass Biomass in silica units Dissolved SilicaTotal Silica

33. Why nutrients are not limiting in winter?  Salt is rejected from ice  due to thermodynamic constrains on brine salinity  Common sense: nutrients should be rejected from the ice  Nutrient fluxes are proportional to:  Ocean-brine gradient of nutrient concentration  Diffusivity (high for growing ice)  Hence, nutrients can be fluxed to the ice. time scales of nutrient uptake is much slower than convection

34. Conclusions (1)  Ice salinity can reasonably well be simulated  1d models: winter convection, full-depth convection, flushing  Timing, magnitude and model sensitivity are uncertain  Ice salinity is an important actor of large-scale ice- ocean dynamics  Ice salinity affect thermodynamics in the Arctic and ice- ocean interactions in the Antarctic  Ice salinity is increasing now as the amount of FY ice increases. Hence, the sensitivity of coupled models may depend on how salinity could be represented

35. Conclusions (2)  Brine-ecosystem multi-layer models are quite different from single-layer based models  Brine dynamics allow to simulate vertical structure in ecosystem in a reasonable way  There are brine dynamics-nutrient interactions  Transport mode of diatoms is important  Brine salinity limitation is the second important factor after light limiation

36. Remaining problems (1)  No thermo-haline coupled term in brine transport equations  Physical inconsistencies in the model (ice/brine density, freezing point, …)  Model results rely on uncertain parameterizations

37. Remaining problems (2) Non-destructive method-based time series  Better account of spatial variability (Hajo’s talk)  Less errors at high brine volumes (Phillip’s talk) How should vertical diffusivity be parameterized ? Permeability-porosity relation for wide range of T,S What are the pathways of seawater during surface flooding ? Full-depth convection: when, how, how often ? How to represent 3D subfloe-scale circulations ? What is the impact of ridged ice desalination? How to represent gas transfer? How to design a sound model-data comparison

38. The truth about Louvain-la-Neuve Ice Model THANKS TO: Cc Bitz, Ralph Timmermann, Steve Ackley, Thierry Fichefet, Hugues Goosse, Gurvan Madec and NEMO team, Jean-Louis Tison, Tony Worby, Hajo Eicken, Bruno Delille, Miguel Angel Morales Maqueda, Bruno Tremblay, Ioulia Nikolskaia, Oiivier Lecomte, Olivier Lietaer, Sylvain Bouillon, Anne de Montety, Christiane Lancelot and Ivan Grozny + forgotten! The principle is very basic: A terrific vibration with maximal resonance… nice model huh? I found inspiration from an African instrument, but I improved it…

39. Further reading  Vancoppenolle, M., H. Goosse, A. de Montety, T. Fichefet, B. Tremblay and J.-L. Tison (2010). Modelling brine and nutrient dynamics in Antarctic sea ice : the case of dissolved silica. Journal of Geophysical Research, 115(C2), C02005, doi:/10.1029.2009JC005369.  Vancoppenolle, M., T. Fichefet, H. Goosse, S. Bouillon, G. Madec and M.A. Morales Maqueda (2009a). Simulating the mass balance and salinity of Arctic and Antarctic sea ice. 1. Model description and validation, Ocean Modelling, 27, 33-53, doi:10.1016/j.ocemod.2008.10.005.  Vancoppenolle, M., T. Fichefet, and H. Goosse (2009b). Simulating the mass balance and salinity of Arctic and Antarctic sea ice. 2. Importance of sea ice salinity variations, Ocean Modelling, 27, 54-69.  Vancoppenolle, M. (2008b). Modelling the mass balance and salinity of Arctic and Antarctic sea ice, Phd Thesis, Université Catholique de Louvain, ISBN 978-2-87463-113-9.  Vancoppenolle, M., C. M. Bitz, and T. Fichefet (2007), Summer landfast sea ice desalination at Point Barrow, Alaska: Modeling and observations, Journal of Geophysical Research, 112, C04022, doi:10.1029/2006JC003493.  Vancoppenolle, M., T. Fichefet and C.-M. Bitz (2006) : Modeling the salinity profile of undeformed Arctic sea ice, Geophysical Research Letters, L21501, doi://2006GL028342.  Vancoppenolle, M., T. Fichefet, and C.M. Bitz (2005) : On the sensitivity of undeformed Arctic sea ice to its vertical salinity profile, Geophysical Research Letters, L16502, doi://2005GL023427.