rhop diagnostic only available if other diagnostics requested

[davestorkey]

Context

The iom_put call for potential density ('rhop') is in diaar5.F90 and is in a block of code protected by this IF test:

IF( iom_use( 'botpres' ) .OR. iom_use( 'sshthster' ) .OR. iom_use( 'sshsteric' ) ) THEN

So it is only available if one of those other diagnostics is also requested.

Analysis

...

Fix

The potential density is not required for the calculation of the other diagnostics in that IF-block, so it should be moved out of there and protected with its own "IF(iom_use('rhop'))".

[davestorkey]

In 12711:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[davestorkey]

In 12712:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[davestorkey]

The existing call to eos calculates both in-situ and potential density and then in-situ density is used in the calcuations for the steric SSH and bottom pressure. So I have modified the solution slightly to put an ELSE clause in the main IF test - see branch.

Should we add an in-situ density diagnostic since the field is available here? (I don't think this diagnostic exists already?)

[davestorkey]

Branch (rev 12712) tested in GYRE_PISCES and rhop diagnostic is now available (whereas for the standard r4.0-HEAD code the diagnostic is full of zeroes).

[mathiot]

I agree with the fix and the suggestion to add the in-situ density as diagnostics as it is computed anyway. By the mean time, probably you could change the L_ar5 to l_ar5 to be consistent with the rest of the code and align the
') .OR. &' within the 4 lines of the IF statement.

However, I have a stupid question about the computation of zrhop and zrhd:
In step.F90, you have this line:

CALL eos    ( tsn, rhd, rhop, gdept_n(:,:,:) )  ! now in situ density for hpg computation

This is exatly the same line as in diaar5.
As rhop and rhd are public, a better fix (improve performance and less lines of code) we could simply use these variables instead of recomputed it in diaar5 and output rhop and rhd strait into diawri (This will avoid the need to activate diaar5 flag only for density output), isn't it ? Did I miss something ?

[davestorkey]

In 12727:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[davestorkey]

I think Pierre is right, so I propose to remove the unnecessary second call to eos from dia_ar5 and use the already-calculated rhd to calculate botpres (instead of zrhd). Then it seems more logical to put the iom_put call for rhop in dia_wri.

Implemented this in the branch (rev 12727) and I get bit-comparable results for the rhop and botpres diagnostics in GYRE_PISCES.

This fix should be applied to r4.0-HEAD and the trunk.

[davestorkey]

Christian - are you happy with this change to diaar5.F90?

[cetlod]

Sorry for the late reply, YES YES I agree with the proposed solution

[davestorkey]

In 13077:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[davestorkey]

In 13085:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[davestorkey]

In 13086:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[davestorkey]

In 13087:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[davestorkey]

In 13089:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[davestorkey]

Fixed in r4.0-HEAD and the trunk. Closing the ticket.