E3SM-Project
update bin_to_cube tool
- modify ``bin_to_cube'' according to https://github.com/NCAR/Topo.git for cubed base topo generation
- read parameters from ``bin_to_cube.nl''
manually pin @mt5555
@brhillman : The important change here is the normalization of terr_cube around line 286. That's a bug that E3SM introduced when we removed land_frac from this tool chain. This bug hasn't impacted us since it would only show up if someone made a new cube3000 file (or in this case, Jishi's cube12000 file).
Another change is the introduction of a namelist for input parameters - but this will break our existing workflow. edit: actually, our existing tools probably start with the "cube3000", so maybe we dont have to worry about modifications to specifying inputs to bin_to_cube
@jsbamboo : there are still a lot of landfrac_* related variables in cube_to_bin, including writing it to the output file. Do we need this? I'm think it is no longer used by cube_to_target, so could be removed here as well.
@mt5555 I think you are right. we dont need landfrac* if ``lzero_out_ocean_point_phis'' is set to FALSE in cube_to_target, which is the default setting
please let me know if you want me to remove bin_to_cube.nl / other edits, or feel free to push commits on your end
In: https://acme-climate.atlassian.net/wiki/spaces/DOC/pages/4189520033/800m+cubed+topo+generation+from+GMTED2010+15s+DEM
@jsbamboo reported that cube_to_target required 638G to run, with a 7.5 arch-second lat/lon input data set and cube12000 output.
Storing the landrac field on this grid is ~110GB. Thus removing it I think is worthwhile.
@mt5555: to clarify, the 638G memory would be needed from 7.5 arch-second lat/lon input (~250m) to cube48000 output (~200m), which is kind of pointless since it's more than the original resolution.. maybe processing to cube40000 (250 m) or cube24000 is appropriate, which would require ~75GB or 27GB for ``landfrac_cube'' (which is also big)
the second new commit removed landfrac, i pushed it for review while it can be reverted if anyone wants it
a RRM grid test for bin_to_cube and cube_to_target has passed for the updates, and there is no landfrac anymore in the outputs
most of the latest comments were fixed by the 03/21 commits, but the last one to remove a comment has not yet. I am out of office since 04/05, I can push a commit after I get back to lab on ~04/21
@mt5555 and @brhillman I've merged this to next. If you don't want it merged to master, let me know by tomorrow.
great, thanks.
it will probably be a several more years before we next have to use this tool, but still good to resolve this PR.