CICE6-simulated summer ice drift speed is too fast

[hspark1@gmail_com]

I am simulating CICE6.1.2 coupled to a slab ocean with both ERA5 and COREv2 atmospheric forcings.
The simulated sea ice drift speed in the summer (Aug-Sep) is about 40-50% faster than the satellite-observed ice drift speed. This is way too fast provided that the CICE6 coupled to a slab ocean does not account for the surface ocean currents that certainly enhance the ice drift speed.
I've tested this using both kstrength=0 and 1 and using both ERA5 and COREv2 forcings.
I've used default values for other dynamics parameters:
Cdn_ocn = 0.006_dbl_kind
vonkar = 0.4_dbl_kind ,&! von Karman constant
zref = 10._dbl_kind ,&! reference height for stability (m)
iceruf = 0.0005_dbl_kind ,&! ice surface roughness (m)

I suspect whether using 10m wind speed was a factor for simulating the anomalously fast summer ice drift speed. Are we supposed to use 10m wind speed? Is the air-ice drag coefficient optimized for 10m winds?

Thank you for helping,
Hyo-Seok.

 

[dbailey]

I have moved this to the CICE Consortium discussion since this appears to be a standalone question. What resolution are you running at? I will let others in the Consortium chime in on this.

 

[eclare@lanl_gov]

The speed of the ice depends on many different parameters, including internal parameters and the external (atmospheric, oceanic) forcing. Including ocean currents could help, if they oppose the flow direction or cause the ice to ridge more. In general, if your ice is overly thin, then it will move too fast. That said, the model was tuned for thicker ice regimes in decades past. The default parameters were tuned using 10 m winds, but with an older forcing data set that was less than optimal. Most groups retune the model for their standard configuration, which can sometimes result in very different parameter choices. There are several papers published with this kind of information from other modeling centers (e.g. air-ice drag coeff). We do not recommend that you use a slab ocean model for production simulations - it's best to run fully coupled.

 

[hspark1@gmail_com]

My recent paper also points out the importance of the ocean coupling:

https://tc.copernicus.org/articles/10/227/2016/tc-10-227-2016.pdf

I am not preparing any research paper... The simulated sea ice thickness is similar to that of PIOMAS. I've also tested the sensitivity of summer ice speed by changing the ice albedo. When daily mean data is used, the simulated summer sea ice drift speed is much faster than that of Polar Pathfinder daily 25 km. Please see the plot.

 

[hspark1@gmail_com]

I have moved this to the CICE Consortium discussion since this appears to be a standalone question. What resolution are you running at? I will let others in the Consortium chime in on this.

Thank you, I am running CICE-6 as a displaced pole grid with 320×384 grid points.