Global sum BFB code in the surface salinity restoring routine reduces precision. #6446
Assignees
Labels
Comments
jonbob
added a commit
that referenced
this issue
Jul 9, 2024
Replace global sum with full-precision reproducible sum The global sum in the surface salinity restoring routine was forced to be BFB using a truncation calculation that resulted in reduced precision and some unexplained behavior. This PR replaces this calculation with an mpas standard reproducible sum with full precision. Testing was done using a G-case with the IcoswISC30E3r5 grid for one year, then calculating the global average SSS restoring tendency term, which should be effectively zero. Also confirmed that the new code resulted in BFB solution when processor layout was changed from 10 nodes on chrysalis to 15 nodes. Fixes #6446 [non-BFB] for C- and G-cases with salinity restoring
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The global sum in subroutine ocn_get_surfaceSalinityData (required to ensure no net salt input into the ocean due to the restoring term) was originally implemented in the simplest way, which was not guaranteed to be BFB on different processor layouts. a modification was made to make it BFB, but apparently this method reduces the precision of the restoring term. while it seems to be a minor problem, we should be using the mpas reproducible sum routine instead.
this plot shows a 40 year time series of the globally averaged monthly SSS restoring tendency (log(abs(tend)) for a G-case using the IcosWISC30E3r5 mesh (in black). The red curve is a restart from year 40 using double precision output, which doesn't help. The green curve has the BFB code turned off showing much higher precision and no spikes.
The text was updated successfully, but these errors were encountered: