<div dir="ltr"><div>Dear all,<br>
<div class="gmail-wiki">
<p>Following changes were committed to the NEMSfv3gfs and FV3 repositories:</p>
<p>Task <a class="gmail-issue gmail-tracker-28 gmail-status-1 gmail-priority-4 gmail-priority-default" title="Task: Advanced physics update to fv3gfs - add MG3, fix some bugs in MG2 and vertical diffusion in shoc (New)" href="https://vlab.ncep.noaa.gov/redmine/issues/49401">#49401</a>: Advanced physics update to fv3gfs - add MG3, fix some bugs in MG2 and vertical diffusion in shoc</p>
<p>This ticket is on adding updated version of latest MG3 microphysics
from NCAR. The code has been modified to make it more efficient
including quasi-implicit treatment of sedimentation.</p>
<p>1. The updated version can be run as either MG2 or MG3 via namelist
options. The differences between MG2 and MG3 is that MG3 includes
prognostic graupel and it's number concentration.</p>
<p>2. While installing MG3 and optimizing a bug was found in the
optimized MG2 version (which is in the master now) which is now fixed.</p>
<p>3. A bug in the vertical diffusion of TKE in shoc is also fixed.
Some tunable parameters for shoc is now exposed to the namelist as well
as some parameters for MG3. The microphysics driver is optimized to
reduce memory by eliminating unnecessary arrays.</p>
<p>4. For shoc an option following Scipion et. al. is added for TKE dissipation near the lower boundary.</p>
<p>5. In this version of the code the word "INPUT/" is removed from the
file name in "ozinterp.f90" and "h2ointerp.f90". With this change to
the code the ozone and h2o forcing file needs to be in the "RUN"
directory, outside of the "INPUT" directory. Additional namelist
parameters are added (related to SHOC and MG) to the physics namelist.
To run MG3 or "MG3 as MG2" or the original MG2/MG1, proper namelist and
field_table needs to be used. Some examples can be seen in the
additional regression tests.</p>
<p>6. The IO routine "module_wrt_grid_comp.F90" has been modified to
set up file names for all time ((including every time step and forecast
hours beyond 999 hours). The code has been tested by running up to 1080
hours (45 days).</p>
<p>Regression test scripts are modified to handle the o3 and h2o input
file directory change. A new regression test, csawmg3sho127, with C96
127 vertical layers is added and currently runs on cray and theia. The
above code changes do not alter baseline results except the cs/mg/shoc
regression tests.</p>
</div>
Thanks!<br><br><br></div>Jun<br></div>