Impacts of a mushy‐layer thermodynamic approach in global sea‐ice simulations using the CICE sea‐ice model

Abstract

We perform global simulations of the Los Alamos sea-ice model, CICE, with a new thermodynamics component that has a fully prognostic, variable bulk salinity vertical profile based on mushy layer physics. The processes of gravity drainage, melt-water flushing and snow-ice formation are parameterized to allow the bulk salinity to evolve with time. We analyze the seasonal and spatial variation of sea-ice bulk salinity, area, volume and thickness and compare these quantities to simulations using the previous thermodynamic component. Adjusting one of the gravity drainage parameters, we find good agreement between simulation results and fieldwork ice-core observations of sea-ice bulk salinity. As expected, bulk salinity is highest during periods of ice growth and lowest after periods of ice melt. In the northern hemisphere the new thermodynamics component produces thicker ice than the previous thermodynamics component. Of the nine major differences between the two models, differences in how salinities are calculated and how melt-pond flushing is parameterized are the principal causes of this thickness difference. Thickness differences are smaller in the southern hemisphere than in the northern hemisphere since a greater fraction of ice melts, and differences cannot accumulate year-on-year. Model differences in how ice thickness changes and snow-ice formation are calculated are the most important causes of the different thickness between the two thermodynamic components in the southern hemisphere. The melt-pond area and volume are found to be highly sensitive to a parameter choice controlling drainage through macroscopic holes in the ice, in both hemispheres.

A high-resolution ocean and sea-ice modelling system for the Arctic and North Atlantic oceans

“…The two latter advances are already available in CICE5 (Turner and Hunke, 2015). We also plan to move to NEMO version 3.6 in the upcoming future, which will offer support for coupling to CICE5.…”

neXtSIM: a new Lagrangian sea ice model

“…Virtually all modern sea ice models use either the VP or EVP formulation, combined with a thermodynamics model (e.g. Semtner, 1976;Bitz and Lipscomb, 1999;Vancoppenolle et al, 2009;Turner and Hunke, 2015) and variously detailed subgrid-scale parameterisations (for commonly used large-scale sea ice models as for instance CICE , LIM3 (Vancoppenolle et al, 2009), MITgcm (Adcroft et al, 2016) or MPI-ESM (Notz et al, 2013)). …”

Processes controlling surface, bottom and lateral melt of Arctic sea ice in a state of the art sea ice model

“…We choose to resolve five ice layers and one snow layer vertically and compare model results between the fixed salinity profile parametrization of Bitz & Lipscomb [36] and the newly available mushy parametrization, in which the salinity within the ice can evolve in time (halodynamic model of Turner et al [37]). The differences between the two models as well as the impact of both halodynamic components on the main sea ice characteristics are discussed in details in Turner & Hunke [38].…”

“…The names and changes in these sensitivity runs are as follows. In MLD_CST, we use the default fixed depth slab ocean ML described in §2b; in MLD_MIN_2M, we set the minimum allowed ML depth to h mix = 2 m; in NO_3EQTN, we revert to the default boundary condition treatment with T 0 = T f (S mix ) ( §2b); in NO_MUSHY, we replace the mushy parametrization and flushing of Turner & Hunke [38] by the fixed salinity profile scheme of Bitz & Lipscomb [36] ( §2a); DBL_ALPHA_H, DBL_ALPHA_H / NO_3EQTN and DBL_ALPHA_H / NO_MUSHY are the same as REF, NO_3EQTN and NO_MUSHY but with a doubling of α h ( §2a); in NO_POND, we artificially set the thickness of the melt ponds to zero; in FALSE_BOTTOM to simply model the impact of under ice fresh water accumulation on the bottom heat flux we double α h where melt ponds cover less than 20% of the ice surface; in NO_FORM_DRAG, we switch off the Tsamados et al [40] form drag parametrization ( §2a); in LAT_MELT, we switch on the lateral melt parametrization described in §2b; finally, in SST_TIME, we restore the sea surface temperature (SST) to the time-dependent temperature of the MYO reanalysis surface ocean temperature over the period 1993-2010 (because the ocean reanalysis is limited to this period). All the sensitivity runs are summarized in table 1.…”

“…The only simulations that have similar SIA and SIE (but lower SIV) to the SFFT14 run are NO_MUSHY and DBL_ALPHA_H / NO_MUSHY that use the same thermodynamic treatment of the ice Bitz & Lipscomb [36] and the same parametrization of the flushing of melt ponds [38] as is used in Schröder et al [57]. Two outlier runs on figure 12, NO_MUSHY (and DBL_ALPHA_H/ NO_MUSHY not shown) and SST_TIME (and to a lesser degree NO_POND) show a very low and high total volume of ice throughout the season ( figure 12g-i).…”