search

NorESMhub / noresm2cmor

A command line tool for cmorizing NorESM output
http://noresmhub.github.io/noresm2cmor/
5 stars 16 forks source link

Inconsistent sea ice and ocean grids #279

Closed YanchunHe closed 3 years ago

YanchunHe commented 3 years ago

Quote S.Wang (FIO, CN).

I found the sea ice products of NorESM2-LM are on a 360x384 grid, while the ocean products are on a 360x385 grid. When I calculate the areal summation of the ocean (e.g., total shortwave radiation), I resort to the file areacello_Ofx_NorESM2_LM_historical_r1i1p1d1_gn.nc. What file shall I use if I want to do the areal summation of the sea ice (e.g., sea ice volume in the Northern Hemisphere)?

I think sea ice model should share the same mesh with the ocean model (and most of CMIP6 models do). So I feel confused why the sea ice component and the ocean component use different meshes in NorESM2-LM.

YanchunHe commented 3 years ago

Quote M.Bentsen:

I understand these issues are a bit confusing. The ocean/sea-ice grid of NorESM2-LM is a tripolar grid with 360 and 384 unique grid cells in i- and j-direction, respectively. Due to the way variables are staggered in the ocean model, an additional j-row is required explaining the 385 grid cells in the j-direction for the ocean grid. The row with j=385 is a duplicate of the row with j=384, but with reverse i-index.

The ocean and sea-ice components of NorESM defines the grid cell area differently. In the ocean component, the grid cell area is found by computing the area of a spherical polygon with grid cell corners as vertices. The sea-ice component computes area as dx*dy where dx and dy are grid cell sizes in i- and j-direction, respectively. In order to achieve good conservation in flux exchanges, we ensure that the ocean and sea-ice components have identical grid cell areas. To obtain this with the different approaches of computing grid cell area, we nudge the sea-ice grid locations slightly.

In conclusion, it is consistent to use the areacello variable defined on the ocean grid in relation to sea-ice variables, but you have to ignore the final j-row of areacello.

YanchunHe commented 3 years ago

So to conclude, just drop the last row with j=385 of areacello when dealing with the sea ice variables.

YanchunHe commented 3 years ago

Maybe Alok can report this as an issue/warning for all areacello, thkcello and volcello variables in the ESGF doc? Not that the last column j=385 can be dropped when using these variables as masks/weights for sea-ice variables? @monsieuralok

monsieuralok commented 3 years ago

@YanchunHe do we need to mention it for all experiments and variables areacello, thkcello and volcello ? Also for NorESM2-LM, NorESM2-MM and NorESM1-F?

monsieuralok commented 3 years ago

@YanchunHe opened errata issue https://errata.es-doc.org/static/view.html?uid=fe3ef410-e0e9-24d6-a987-35285dc3c66f and it would be nice if you could check it once.

YanchunHe commented 3 years ago

@YanchunHe do we need to mention it for all experiments and variables areacello, thkcello and volcello ? Also for NorESM2-LM, NorESM2-MM and NorESM1-F?

Yes, should be all models with triangle grid.

I close this issue now.