[Ncep.list.fv3-announce] NEMSfv3gfs and FV3 git repository updates 20180914

[Jun Wang - NOAA Federal]

Dear all,

Following changes were committed to the NEMSfv3gfs and FV3 repositories:

Task #54973 <https://vlab.ncep.noaa.gov/redmine/issues/54973>: bug fix in
calling gfs radiation driver, compute Gaussian latitude using double
precision
It is found that there is a phase shift of instantaneous surface downward
shortwave in the fv3gfs output compared to operational GFS. It turns out
the gfs radiation driver is called at every time step, this causes the
cosine zenith angle changing even though the short wave and long wave
radiation are still called at correct radiation time steps. Because of
this, the cosine zenith angle does not match with the radiation fields
computed at radiation time step. This is fixed by only calling gfs
radiation driver at radiation time step.
It is also found that because fv3gfs parallel is running at 32 bit real the
Gaussian latitudes in NEMSfv3gfs output files show slight differences
compared to operational GFS which is running at 64 bit real. To make
consistent products, the Gaussian latitude is computed using hard wired 64
bit real in NEMSfv3gfs.

Bug #54894 <https://vlab.ncep.noaa.gov/redmine/issues/54894>: A bug fix in
tref update with sfccycle
In the current FV3GFS, after call gcycle, there are some unusual values in
the tref and tsfc updates over coast areas due to the mask update issue.
This is fixed by updating tref and tsfc in a different way by modifying
gcycle.f90.

Baseline results are altered due to the changes above. New baseline was
created on wcoss (phase1,2 and cray) and theia. Regression test passed on
those platforms.
Thanks.

Jun
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://www.lstsrv.ncep.noaa.gov/pipermail/ncep.list.fv3-announce/attachments/20180914/3fa70937/attachment.html