Changing Parameterization Schemes

[bnepal22]

How to change the current parameterization scheme in the UFS model. Currently, I am using the FV3_GFS_v15p2 version. Is there any detail process manual for doing so?

[gspetro-NOAA]

Hi @bnepal22 ,

My understanding is that you can just indicate a different suite wherever FV3_GFS_v15p2 appears in your previous configuration file(s). In general, the build/run chapter of the User's Guide has some information on how to set the CMAKE_FLAGS if you are compiling using build.sh. Specifically, you would change the -DCCPP_SUITES flag (e.g., -DCCPP_SUITES=FV3_GFS_v16). That chapter also has information on running regression tests, which you could try if you are running an existing/supported configuration and just switching out the physics. In that case, you would add the physics suite you want to use in the COMPILE statement of your configuration file if it isn't already listed (e.g., -DCCPP_SUITES=FV3_GFS_v16, and you would add or modify the CCPP_SUITE variable in the test you are using (e.g., CCPP_SUITE=FV3_GFS_v16).

Note that not all physics suites are appropriate for a given configuration, but without additional information (e.g., on machine, configuration, method for running), it would be difficult to advise you on which suites to use. In general, the CCPP forum is the best place to ask questions like that.

@jkbk2004 Do you have anything to add here?

Best,
Gillian

[bnepal22]

I am sorry for not being clear in the previous question. I was trying to switch to Thompson Cloud Microphysics from GFDL Microphysics just for some test, and when I only changed the "imp_physics = 11" to "imp_physics = 8" under "&gfs_phys_nml" in the namelist file, but it showed error. I'd like to know if there is any detailed manual.

[gspetro-NOAA]

@mkavulich Is this question best posted in the CCPP-physics discussions? Regardless, could you advise?
@bnepal22 I know that for the SRW App, there is some guidance here, but I think that if you want to change it in the WM itself, you would have to turn on (and off) certain physics-related options in your test file and probably update the CCPP suite definition file. In general, we encourage users to stick to the well-tested physics suites already in use, but we have definitely had users with reasons to change the usual combinations (e.g., for research purposes).

[bnepal22]

Yes, I am attempting such a physics change for research purposes, and basically, I am doing it for S2S scale prediction using a stretched grid.

[mkavulich]

@gspetro-NOAA I think this question is best answered here since it does have some dependence on the specific dycore.

@bnepal22 For models that use CCPP (as UFS weather model does), the physics suites (i.e. collections of schemes) that can be run are set at compile time with Suite Definition Files. So in order to use a brand new suite (which is typically the case if you are simply swapping out

In order to change just the microphysics scheme, unless there is an existing suite that uses the exact schemes you require, you will need to update the CCPP suite file (or create a new one) and re-compile. In this case, there is an existing suite suite_FV3_GFS_v15_thompson_mynn_lam3km.xml that uses Thompson, but it also has some other changes, so I would recommend creating a new file so that you can still use the old and new suites interchangeably.

From the ufs-weather-model directory:

cd FV3/ccpp/suites
cp suite_FV3_GFS_v15p2.xml suite_FV3_GFS_v15p2_thompson.xml

Then edit the new file to swap out GFDL microphysics with Thompson. The Thompson scheme actually consists of three schemes that must be run in order, so those changes will look like this:

@@ -1,6 +1,6 @@
 <?xml version="1.0" encoding="UTF-8"?>
 
-<suite name="FV3_GFS_v15p2" version="1">
+<suite name="FV3_GFS_v15p2_thompson" version="1">
   <!-- <init></init> -->
   <group name="fast_physics">
     <subcycle loop="1">
@@ -77,7 +77,9 @@
       <scheme>GFS_suite_interstitial_4</scheme>
       <scheme>cnvc90</scheme>
       <scheme>GFS_MP_generic_pre</scheme>
-      <scheme>gfdl_cloud_microphys</scheme>
+      <scheme>mp_thompson_pre</scheme>
+      <scheme>mp_thompson</scheme>
+      <scheme>mp_thompson_post</scheme>
       <scheme>GFS_MP_generic_post</scheme>
       <scheme>maximum_hourly_diagnostics</scheme>
       <scheme>GFS_physics_post</scheme>

Next, you will need to re-compile with this new suite. The easiest way to do this will be to set the CCPP_SUITES variable manually as was described above to include the new FV3_GFS_v15p2_thompson suite. This will depend on the context in which you are running the model: are you running the UFS weather model stand-alone or with one of the UFS applications?

[bnepal22]

I think it's stand-alone since I am running the Atmospheric model.