Running with AMIP-ssts

Submitted by donald.dazlich on Thu, 06/04/2020 - 20:51
Forums
Medium-range Weather Application

Running the ufs medium range weather app out of the box produces runs where the SST and sea-ice cover are unchanged with time. How can I run this with AMIP-style SSTs?

Permalink

Submitted by ligia.bernardet on Wed, 06/10/2020 - 14:30

The UFS S2S App is currently being upgraded to use the CMEP Community Coupler and at some point it will have a release using the same CIME workflow used in the UFS MRW App v1.0 release. At that point, it should be possible to use data models to replace the active ocean component with forcing datasets. The timeline for this release is yet to be determined.


Regarding the current UFS MRW App v1.0 release, the SST is represented using a forcing toward climatology.  There is a related namelist parameter, fhcyc, which needs to be set to 24 (hours). Look at the "documentation" in FV3/ccpp/physics/physics/gcycle.F90 and FV3/ccpp/physics/physics/sfcsub.F - these are called from FV3/ccpp/physics/physics/GFS_phys_time_vary.fv3.F90. Perhaps you can change this climatology file to force the model toward with your desired SST. I personally do not have experience doing that.

A caveat is that If the initial conditions used are in NEMSIO format, additional modifications to the near-SST are made using the NSST physical parameterization to represent the diurnal cycle. If initial conditions used are in GRIB2 format, the SST is not modulated by the diurnal cycle. From gcycle.F90:

if ( Model%nstf_name(1) > 0 ) then

             Sfcprop(nb)%tref(ix) = TSFFCS  (len)

          else

             Sfcprop(nb)%tsfc(ix)  = TSFFCS  (len)

             Sfcprop(nb)%tsfco(ix) = TSFFCS  (len)

          endif

nstf_name(1) >  0 corresponds to running the NSST scheme; in this case the climatology is only updating the reference temperature (that gets used inside the NSST scheme), but not directly the ocean temperature fields tsfc and tfsco.

If you give this a try, let us know how it goes.

 

Permalink

Submitted by donald.dazlich on Wed, 06/10/2020 - 20:18

The setting of the nameless variables was not enough to get the SSTs to changes. But I did go into gcycle.F90 and had to add the assignment for Srcprop(nb)%tsfco(ix) myself. Now the SSTs vary. Thanks for pointing me the right direction.

Don

  • 49 views

Copyright © 2023. All rights reserved.