Closed mark-petersen closed 6 years ago
@xylar or @vanroekel if either of you have a minute, this is another small annoyance out of the way. It passes nightly regression suite. I think a visual inspection from either of you is sufficient.
@mark-petersen, will this give us trouble near ice shelves where landIceMask == 0
but the grid is still distorted? I would think it would. Maybe we need to exclude things based on another mask (related to where the grid is allowed to deviate from z-level) though I'm not sure that flag makes it through to the forward run.
Actually, I guess the SSH is probably still close to zero even when the grid is distorted quite a lot from z-level. I think this should be fine
I'll run a couple of tests and then approve. I'm doing this climate hack so won't have time to test until later today or later in the week.
@xylar I think it's OK. We have 240km global case with ice shelves in the nightly test suite. Worst case is we merge and then still need to set
config_check_ssh_consistency = false
for wLI
runs, which is what we do now anyway.
Fine with me.
Previously, on runs with ice shelf cavities, we could not use this ssh check because ssh is outside of the reasonable range below ice shelves. With this PR, we check that abs(ssh)<20m, but only for the open ocean, not where landIceMask is nonzero.