EAMxx: add horizontal average diagnostic field #6788
Conversation
|
|
@mahf708 , thanks for re-issuing this PR on the E3SM side. Quick question, there was this discussion between Peter B. and Andrew (E3SM-Project/scream#2738 (comment)) about the algorithm for the horizontal average that takes into account mass. Any thoughts on whether this could/should be incorporated into this PR? |
This PR is only area-weighted averaging, which is the typical way of doing this as far as I am concerned. I'm happy to hear arguments for including a mass adjustment, but then would the mass adjustment apply to all quantities or specific ones? I'm not sure, but we could easily have another diagnostic such that it is "area-weighted, mass-adjusted" horiz avg |
|
I thought dp would be more critical for a vertical aggregator |
|
I typically always neglect mass when doing horizontal averages in DP. Unless @ambrad has a strong opinion about including this I'm fine with keeping it area weighted. |
|
Yeah. Additionally, I'm not sure if it is realistic or even more accurate? Also, taking into account vertically resolved mass may become nontrivial when it comes to taking averages with unrelated dimensions (say radiation band or if a field has no vertical dimension). These diag impls are supposed to be generic and applicable to any fields (not just fields similar to dp) ... |
|
@AaronDonahue not to digress the PR convo (maybe we can move to slack), but if there's the need to compute "mass" quantities (e.g., dp*some_tracer), we could add a diagnostic |
|
I consider mass weighting as probably a bit better in the case of a quantity like a tracer. But I don't think it's terribly wrong to neglect it, except in the case that the diagnostic calculation needs to be fully mass accurate (which probably is never needed). |
mahf708
force-pushed
the
mahf708/eamxx/horiz-avg-diag
branch
from
c6fa3aa to
3aaf335
Compare
mahf708
force-pushed
the
mahf708/eamxx/horiz-avg-diag
branch
from
3aaf335 to
6ec6f3f
Compare
|
@bartgol there are still issues to do with composability with the complex logic of field_at_xyz. I can handle those in a later PR, otherwise the scope of this PR will creep again... Ready for re-review! |
| @@ -30,10 +30,12 @@ initialize_impl (const RunType /*run_type*/) | |||
| using namespace ShortFieldTagsNames; | |||
| const auto& fid = f.get_header().get_identifier(); | |||
| const auto& layout = fid.get_layout(); | |||
| EKAT_REQUIRE_MSG (layout.rank()>1 && layout.rank()<=6, | |||
| EKAT_REQUIRE_MSG (layout.rank()>=2 && layout.rank()<=6, | |||
There was a problem hiding this comment.
Sorry to be annoying here, but I wanted all three checks to be as close as possible. I will edit these entirely in a future PR, when I address the composability
Adds an online diagnostic field for area-weighted horizontal average, following #6776.
Users are now able to request some_field_horiz_avg in their output files to get an area-weighted horizontal average. The following types of fields are supported: