Icepack - Sensitivity to Initial Conditions and External Forcings Dataset/Tests

[simondriscoll]

Hi, I was looking to test the Icepack model's sensitivity to initial conditions (external forcing held the same) and then test it's sensitivity similarly to external forcing with a perturbed range of forcings. I was advised to request here if anyone knew of data that would be suitable for Icepack for testing Icepack's sensitivity to I.C.'s and a small range of forcings? Many thanks if anyone can give pointers and/or recommendations! (or indeed if this has already been investigated and I have missed something please let me know also!) Thank you.

 

[dbailey]

Hi Simon,

We have 3 atmospheric forcing datasets for Icepack: CFS model output, ISPOL 2004 field experiment and NICE 2015 field experiment. The link to get these is here:

Icepack Input Data

Development repository for sea-ice column physics. Contribute to CICE-Consortium/Icepack development by creating an account on GitHub.

github.com

I'm not sure of other datasets out there. I can ask around for Icepack users.

Dave

 

[eclare@lanl_gov]

I suggested that Simon post the request here, thinking that the model output we provide could be generated for a different (nearby) set of grid cells, from the same model output. However, another perhaps simpler option would be to run CICE with the dynamics turned off, using CICE's forcing data. Then you'd be getting Icepack output in 'real' history files across the entire grid, for a more-or-less realistic range of forcings!

Usually perturbation experiments change internal parameter values (e.g. namelist) or else perturb an initial condition at roundoff level. You could do the latter manually in the code, just by adding a small value to one or more forcing fields. I don't know of anyone who has set up a framework for creating a bunch of perturbed runs like this, other than setting kdyn=0 in CICE.

 

[lzampieri]

Hi Simon and Elizabeth,

For a different project, I am extracting near-surface information from atmospheric reanalysis along various Arctic buoy trajectories. Specifically, I am doing this for ERA5 fields with an hourly time resolution and 31km spatial resolution, but the process could be generalized for other gridded reanalyses like JRA55 or NCEP. Although I have never tried it, I am sure that with a bit of preparation these reanalyses could be perturbed and used to force Icepack in a standalone configuration. In principle, a buoy trajectory could be also perturbed to capture information from nearby locations. I am not sure if this goes in the right direction for your study, but feel free to contact me if this could be helpful and you would like to know more.

Lorenzo

 

[cburgard]

Hi Simon and Elizabeth,
Back in 2018, I prepared some atmospheric forcing files based on ERA-Interim over 4.5 years, 6 or 12-hourly data (I don't remember perfectly). It is for the points 75°N00°W and 90°N. It was for an idealised study, that's why the weird choice of locations. Anyway, if it can help, do not hesitate to drop me an email and I can send you the files.
Best,
Clara

 

[simondriscoll]

I suggested that Simon post the request here, thinking that the model output we provide could be generated for a different (nearby) set of grid cells, from the same model output. However, another perhaps simpler option would be to run CICE with the dynamics turned off, using CICE's forcing data. Then you'd be getting Icepack output in 'real' history files across the entire grid, for a more-or-less realistic range of forcings!

Usually perturbation experiments change internal parameter values (e.g. namelist) or else perturb an initial condition at roundoff level. You could do the latter manually in the code, just by adding a small value to one or more forcing fields. I don't know of anyone who has set up a framework for creating a bunch of perturbed runs like this, other than setting kdyn=0 in CICE.

Dear Elizabeth,

to this very good suggestion, I have downloaded CICE and performed some basic runs. Can I clarify when the dynamics is turned off (kdyn=0), this means there are many Icepacks that do not (?) communicate with each other horizontally or do at each timestep? (I assume the former) Secondly, what is the latitude/longitude grid of CICE? I am having trouble finding this rather simple thing. Then I will know the geographical range/grid of these 'many Icepacks'. Thank you.

Simon

 

[eclare@lanl_gov]

Simon, the grid lat/lon values are printed out in the history files. When kdyn=0, the ice velocities remain zero (provided they were zero to begin with) and there is no advection, so no interaction among grid cells.
e