JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 116, C05008, doi:10.1029/2010JC006588, 2011 Sea ice and iceberg dynamic interaction Elizabeth C. Hunke1 and Darin Comeau2 Received 10 August 2010; revised 3 January 2011; accepted 23 February 2011; published 14 May 2011. [1] A model of iceberg motion has been implemented in the Los Alamos sea ice model (CICE). Individual bergs are tracked under the influence of winds, currents, sea surface tilt, Coriolis, and sea ice forcing. In turn, sea ice is affected by the presence of icebergs, primarily as obstacles that cause the sea ice to ridge on the upstream side or create open water on the downstream side of the bergs. Open water formed near icebergs due to sea ice ridging and blocking of sea ice advection increases level and ridged ice downstream of the bergs through increased frazil ice formation. Resulting anomalies in sea ice area and thickness (compared with a simulation without icebergs) are transported with the sea ice flow, expanding over time. Although local changes in the sea ice distribution may be important for smaller‐scale studies, these anomalies are small compared with the total volume of sea ice and their effect on climate‐scale variables appears to be insignificant. Citation: Hunke, E. C., and D. Comeau (2011), Sea ice and iceberg dynamic interaction, J. Geophys. Res., 116, C05008, doi:10.1029/2010JC006588. 1. Introduction [4] Although many of these studies include sea ice as one of the forcing influences on iceberg momentum, only [2] Icebergs populate high‐latitude seas in both hemi- Jongma et al. [2009] mention the potential effect that ice- spheres and have drawn scientific interest for several rea- bergs may have on sea ice. In their study, icebergs distri- sons. Icebergs in the western North Atlantic Ocean, which bute meltwater nonuniformly over the ocean surface, which pose a threat to shipping and resource extraction, receive a subsequently affects the area and thickness distribution of great deal of attention from monitoring agencies such as sea ice; stabilization of the water column by fresh, cold the International Ice Patrol and from modelers striving to iceberg meltwater leads to greater sea ice area, which then predict their tracks. As conduits for freshwater transport, contributes to further cooling and freshening of the surface icebergs modify oceanic water mass properties and there- ocean. They do not include any direct effects on sea ice fore have ramifications for biological communities and the through dynamical interaction with the bergs. In another physical climate system. While large numbers of small bergs modeling study using an early version of CICE, Hunke and thus influence both hemispheres, the southern hemisphere Ackley [2001] found that sea ice advection created polynas also features “giant” icebergs more than 10 nautical miles in the lee of icebergs, as they had observed previously in the (1 nautical mile = 1.852 km) in horizontal extent that are Weddell Sea. Icebergs in that study were treated as islands responsible for approximately half of the fresh watershed in the model’s land mask, however. from the Antarctic continent [Silva et al., 2006]. [5] Thus icebergs can affect sea ice behavior, not just [3] Depending on the motivating goal, two approaches through indirect effects such as ocean surface temperature have generally been taken in iceberg dynamic modeling or salinity, but also through direct contact. A deleterious efforts: treat the icebergs as a statistical distribution or example occurred during the first 5 years of this century, model and track each berg individually. For studies of when several large icebergs calved and were trapped in the meltwater distribution that must necessarily include many southern Ross Sea. The icebergs prevented the normal small bergs, the former approach is natural [e.g., Bigg et al., spring breakout of sea ice behind them [Brunt et al., 2006] 1997; Gladstone et al., 2001; Jacka and Giles, 2007; and strongly impacted nearby penguin colonies [Ainley Jongma et al., 2009]. The latter approach is used for pre- et al., 2006]. diction of berg trajectories in regions of high maritime [6] The present study explores the dynamical interaction traffic [e.g., Smith and Banke, 1983; Smith, 1993; Kubat of icebergs and sea ice and the effect of a few giant icebergs et al., 2005] and for some giant icebergs in the Southern on properties of the sea ice pack in the Weddell Sea. We Ocean [e.g., Lichey and Hellmer, 2001]. present, for the first time, a method for including icebergs in the simulation of sea ice dynamics, and we evaluate the 1Fluid Dynamics and Solid Mechanics Group, Theoretical Division, effects thereof. The icebergs are not treated as sea ice; that Los Alamos National Laboratory, Los Alamos, New Mexico, USA. 2 is, they are not assigned to a sea ice thickness category and Program in Applied Mathematics, University of Arizona, Tucson, they do not have the same thermodynamic properties of sea Arizona, USA. ice. Instead, the icebergs are treated as coherent, individual Copyright 2011 by the American Geophysical Union. ice volumes whose center of mass is tracked in a Lagrangian 0148‐0227/11/2010JC006588 manner on the grid. Icebergs and sea ice have separate C05008 1of9 C05008 HUNKE AND COMEAU: SEA ICE AND ICEBERGS C05008 a Table 1. Constants Used in the Iceberg Calculations where M is the iceberg mass, ub is its velocity, t is time, and the terms on the right‐hand side represent body or sur- Symbol Definition Value face forcing by the atmosphere, ocean, Coriolis, sea ice, and hb berg height 225 m 2 sea surface slope, respectively. The Coriolis term takes the Ah berg horizontal area 686 km −3 form rb berg density 900 kg m r −3 s snow density 300 kg m ^ ^ −3 Fc ¼2MW sin k  ub ¼Mf k  ub ri sea ice density 900 kg m −3 rw ocean density 1025 kg m −5 −1 ^ W angular velocity 7.292 × 10 rad s for latitude , where k is the vertical unit vector. (Constant ci sea ice coefficient of resistance 1 values are found in Table 1.) A geostrophic approximation ca atmosphere coefficient of resistance 0.4 gives the sea surface slope term a similar form cw ocean coefficient of resistance 0.85 −4 cda atmosphere drag coefficient 2.5 × 10 −4 ^ cdw ocean drag coefficient 5 × 10 Fss ¼ Mf k  uw; 4 −1 Ps critical sea ice strength 1 × 10 Nm Dt berg time step 2 min b where u is the ocean current. F and F dominate the Dti sea ice time step 1 h w c ss momentum balance for the icebergs simulated here. aAll sea ice parameters are set as done by Hunke and Lipscomb [2008] [10] Drag by wind and currents, Fa and Fw, take the and Hunke [2010]. quadratic form 1 momentum equations that are coupled through a bulk Fa ¼ acaAva þ acdaAha jjua À ub ðÞðua À ub 2Þ forcing term describing the horizontal momentum transfer 2 between bergs and sea ice. [7] First we describe the iceberg parameterization itself, 1 F ¼ c A þ c A jju À u ðÞu À u ; ð3Þ then the changes made in the sea ice model. Simulation w 2 w w vw a dw hw w b w b results for four giant icebergs placed in the Weddell Sea and tracked for 3 years are described in section 3, along with a where A and A represent the vertical surface area of the number of sensitivity tests. These simulations indicate that va vw iceberg in contact with air and water, respectively. Likewise, the dynamic interaction of icebergs and sea ice is not A and A represent the horizontal surface area of the important for the large‐scale sea ice simulation, although ha hw iceberg in contact with air and water. Full depth ocean local, physically intuitive changes do appear within the currents are available; currents from the surface to the ice- simulated sea ice pack. We discuss this finding and its berg depth are vector averaged vertically for u . The water consequences in section 4. w drag term (3) could be computed at each level for which we have ocean current vectors, but Kubat et al. [2005] found that this did not significantly improve their simulation and 2. Model Description we have chosen the simpler approach. [8] The iceberg parameterization is implemented in the [11] The sea ice term follows that of Lichey and Hellmer Los Alamos sea ice model, CICE version 4.0, and run on [2001], whose expression depends on the sea ice area, ai, a global, 1° mesh whose north pole is displaced into strength P, and a critical strength parameter Ps, Greenland [Hunke and Lipscomb, 2008; Hunke, 2010]. A 0; a < 15% modified version of the Common Ocean Reference Expe- F ¼ i si 1 c A jju À u ðÞu À u ; 15% < a 90% or P P : riments (CORE) [Griffies et al., 2009] atmospheric forcing 2 i i vsi i b i b i s fields for 1990–1992 is applied, along with radiation fields ð4Þ as specified by the Arctic Ocean Model Intercomparison Project [Hunke and Holland, 2007]. The 6 hourly atmo- When the sea ice is highly concentrated and strong (ai > spheric forcing is interpolated to the sea ice time step (1 h). 90% and P > Ps), the iceberg momentum equation (1) is not Ocean data, including full depth currents, are taken from used; instead the iceberg is “captured” by the sea ice: ub = the CCSM3 1990 control run (b30.009) [Collins et al., ui, which Lichey and Hellmer [2001] found necessary for 2006], averaged over 20 years into an annual climatology large, tabular bergs.