MOM6-examples
MOM6-examples copied to clipboard
NOAA-GFDL
Diagnostic testing
This is a feature issue and an umbrella for several other issues.
We want:
Tests that all available diagnostics can be output. Check diagnostic checksums against baseline Check that diagnostics on different vertical grids are correct/consistent. This includes model, z, sigma and rho vertical coords. Sort out cases where diagnostics are registered but no post, or registered, posted but not calculated. See #164.
https://climate-cms.nci.org.au/jenkins/job/mom-ocean.org/job/MOM6_diagnostics/
The compile stage for this Jenkins job is failing, probably because we now need to add "-Duse_AM3_physics" to the CPP macros at the command line with the new coupler code.
All your diagnostics branches are merged. Now I'll hoping to hear we're passing all the diagnostics tests...
I was very excited to see that this morning. I'm working on adding the xy_ave fields to the tests now.
Presently the xy_ave diagnostics use a remap_cs when the horizontal averaging is calculated. This is needed to access remap_cs%h. However this calculate can be done on any coordinate including layer which doesn't have an initialised remap_cs type.
The obvious solution would be to make a remap_cs for all vertical coordinates including layer/model coords. I'll try this and see how it looks.
The tests are not passing for the above reason. There is also something going on with the remap coordinate configuration which I'm looking into. The zold and z diags appear to have layers at different depths.
On Thu, Dec 15, 2016 at 5:07 AM, Nic Hannah [email protected] wrote:
Presently the xy_ave diagnostics use a remap_cs when the horizontal averaging is calculated. This is needed to access remap_cs%h. However this calculate can be done on any coordinate including layer which doesn't have an initialised remap_cs type.
The obvious solution would be to make a remap_cs for all vertical coordinates including layer/model coords. I'll try this and see how it looks.
Or we could put a "if (.not. is_native)" around the call to "horizontally_average_diag_field()" and alternatively call an averaging for native data (like the old_z did).
There is also something going on with the remap coordinate configuration
which I'm looking into. The zold and z diags appear to have layers at different depths.
But the interface positions should be the same as before. The WOA grid is hard to interpret from a finite volume perspective; they only specify center positions and inferring the interface positions which are what defines a FV mesh is subject to some choices. It is not possible to find interfaces whose average line up exactly with the WOA centers and still looks sensible.
-- Dr Alistair Adcroft ([email protected]) Princeton University Tel: (609) 987-5073 NOAA/GFDL, 201 Forrestal Road, Princeton, NJ 08540
You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/NOAA-GFDL/MOM6-examples/issues/98#issuecomment-267289391, or mute the thread https://github.com/notifications/unsubscribe-auth/AFlo8zYq_55TJiOhjsIp24Vsg9ouYOwkks5rIRFJgaJpZM4I9e34 .