next up previous contents
Next: Tracer Namelist Up: Namelist Variables Previous: Writing Output

Model Physics

The namelist variables for the ice model physics are listed in Table 3. restart is almost always true since most run types begin by reading in a binary restart file. See section 5 for a description of the run types and about using restart files and internally generated model data as initial conditions. kcolumn is a flag that will run the model as a single column if is set to 1. This option has not been thoroughly tested and is not supported.

The calculation of the ice velocities is subcycled ndte times per timestep so that the elastic waves are damped before the next timestep. The subcycling timestep is calculated as dte = dt/ndte and must be sufficiently smaller than the damping timescale T, which needs to be sufficiently shorter than dt.

dte < T < dt (2)

This relationship is discussed in (Hunke(2001)); also see (Hunke and Lipscomb(2008)), section 4.4. The best ratio for [dte : T : dt] is [1 : 40 : 120]. Typical combinations of dt and ndte are (3600., 120), (7200., 240) (10800., 120). The default ndte is 120 as set in ice_init.F90.

kitd determines the scheme used to redistribute sea ice within the ice thickness distribution (ITD) as the ice grows and melts. The linear remapping scheme is the default and approximates the thickness distribution in each category as a linear function ((Lipscomb(2001))). The delta function method represents g(h) in each category as a delta function ((Bitz et al.(2001))). This method can leave some categories mostly empty at any given time and cause jumps in the properties of g(h).

kdyn determines the ice dynamics used in the model. The default is the elastic-viscous-plastic (EVP) dynamics (Hunke and Dukowicz(1997)). If kdyn is set to o 0, the ice dynamics is inactive. In this case, ice velocities are not computed and ice is not transported. Since the initial ice velocities are read in from the restart file, the maximum and minimum velocities written to the log file will be non-zero in this case, but they are not used in any calculations.

The value of kstrength determines which formulation is used to calculate the strength of the pack ice. The (Hibler(1979)) calculation depends on mean ice thickness and open water fraction. The calculation of (Rothrock(1975)) is based on energetics and should not be used if the ice that participates in ridging is not well resolved.

evp_damping is used to control the damping of elastic waves in the ice dynamics. It is typically set to .true. for high-resolution simulations where the elastic waves are not sufficiently damped out in a small timestep without a significant amount of subcycling. This procedure works by reducing the effective ice strength that's used by the dynamics and is not a supported option.

advection determines the horizontal transport scheme used. The default scheme is the incremental remapping method ((Lipscomb and Hunke(2004))). This method is less diffusive and is computationally efficient for large numbers of categories or tracers. The upwind scheme is also available. The upwind scheme is only first order accurate.

The base values of the snow and ice albedos for the CCSM3 shortwave option are set in the namelist. The ice albedos are those for ice thicker than ahmax, which is currently set at 0.5 m. This thickness is a parameter that can be changed in ice_shortwave.F90. The snow albedos are for cold snow.

For the new delta-Eddington shortwave radiative transfer scheme (Briegleb and Light(2007)), the base albedos are computed based on the inherent optical properties of snow, sea ice, and melt ponds. These albedos are tunable through adjustments to the snow grain radius, R_snw, temperature to transition to melting snow, and maximum snow grain radius.


Table 3: Namelist Variables for Model Physics (ice_nml)
Variable Name Type CESM-CAM4 gx3 dipole-grid default CESM-CAM4 gx1 dipole-grid default CESM-CAM5 gx1 dipole-grid default Description
ndte Integer 1 1 1 Number of sub-cycles in EVP dynamics.
kcolumn Integer 0 0 0 Column model flag.
          0 = off
          1 = column model (not tested or supported)
kitd Integer 1 1 1 Determines ITD conversion
          0 = delta scheme
          1 = linear remapping
kdyn Integer 1 1 1 Determines ice dynamics
          0 = No ice dynamics
          1 = Elastic viscous plastic dynamics
kstrength Integer 1 1 1 Determines pressure formulation
          0 = (Hibler(1979)) parameterization
          1 = (Rothrock(1975)) parameterization
evp_damping Logical .false. .false. .false. If true, use damping procedure in evp dynamics (not supported).
advection Character remap remap remap Determines horizontal advection scheme.
          'remap' = incremental remapping
          'upwind' = first order advection
shortwave Character dEdd dEdd dEdd Shortwave Radiative Transfer Scheme
          'default' = CCSM3 Shortwave
          'dEdd' = delta-Eddington Shortwave
albicev Double 0.68 0.75 0.75 Visible ice albedo (CCSM3)
albicei Double 0.30 0.45 0.45 Near-infrared ice albedo (CCSM3)
albsnowv Double 0.91 0.98 0.98 Visible snow albedo (CCSM3)
albsnowi Double 0.63 0.73 0.73 Near-infrared snow albedo (CCSM3)
R_ice Double 0.0 0.0 0.0 Base ice tuning parameter (dEdd)
R_pnd Double 0.0 0.0 0.0 Base pond tuning parameter (dEdd)
R_snw Double -2.0 1.5 1.75 Base snow grain radius tuning parameter (dEdd)
dT_mlt_in Double 2.0 1.5 1.0 Snow melt onset temperature parameter (dEdd)
rsnw_mlt_in Double 2000. 1500. 1000. Snow melt maximum radius (dEdd)


next up previous contents
Next: Tracer Namelist Up: Namelist Variables Previous: Writing Output
David Bailey
2012-10-03