Configuring FAMIP CAM5 with 'none' for chempkg

[mmarcella@air-worldwide_com]

Hi,I am fairly new to CESM and its structure of compilation and trying to run a test case of FAMIP with CAM5 but to turn off the chempkg or have no aerosol model with 'none'.  Of course editing just the atm_in file gives me an error and Im unclear how to make this change in the configure or how to run such a case.  Any help would be greatly appreciated.Thanks,
Marc

 

[jedwards]

Edit cam_config_opts in env_build.xml and change the chem option to none.   

 

[mmarcella@air-worldwide_com]

Thanks jedwards! When would this edit need to take place? Before I create the new case or afterwards?

 

[jedwards]

env_build.xml is part of the case so you would need to create the case first.   Please note I have no idea of the scientific validity of running cam5 with no chemistry.   

 

[eaton]

When "-chem none" is specified with the cam5 physics the result is that the prognostic modal aerosol code (trop_mam3) is turned off and the run is configured to use prescribed modal aerosols.  The use of prescribed modal aerosols has not been widely validated, and the availability of driving datasets is limited.  If the use of prescribed aerosols is desired then using the prescribed bulk aerosols is probably a better option.  This is done by setting prescribed_aero_model='bulk' in the namelist.  Note also that running cam5 physics with no aerosols does not work since the cloud microphysics scheme requires aerosols for droplet/crystal nucleation.

 

[mmarcella@air-worldwide_com]

Hi eaton,
Thanks for the reply.  So to use this feature, I do not see the 'prescribed_aero_mode' variable in the atm_in file generated for FAMIP_CAM5, where does one add or change this?  Also what option should be used for chempkg as a result (ie with the  bulk option for prescribed_aerosol) ?  Essentially what I am trying to do is reduce the run time of CAM5, I currently see that it is nearly 3-4 times slower than CAM4.  The FAMIP of CAM4 uses no chem package so I figured that turning off the aerosol model would probably significantly reduce run time (in my experience with RCMs it has) due to the computational and I/O increases.  Are there other suggestions in the model that could significantly reduce its run time? Thanks for your help with thi.

 

[eaton]

Namelist customizations are added to the user_nl_cam file in the CASEROOT directory as explained in the user guide.To use prescribed aerosols one should set "-chem none"There will be a cost savings, but it won't get you back to the cost of cam4.  If that is a major consideration you may be better off using the FAMIP compset which uses cam4 physics.  Changing to cam5 physics without the prognostic modal aerosols will not produce the cam5 climate.  You will need to tune the model and do that evaluation yourself.

 

[mmarcella@air-worldwide_com]

Thanks eaton.  I observed a 50% increase in run time by using the prescribed bulk aerosols instead of the full chem package.  To this point, I tried also adding -rad camrt to the env_build.xml and edited the user_nl_cam to reflect a simulation with prescribed bulk aerosols and the camrt radiation package instead of rrtmg.  However, after successful compilation the model will not run, crashes nearly on initialization.  Is this configuration not supported in CAM5 as it was in CAM4? Or is there some other setting that also must be changed to accomodate?  I do see the generated atm_in does point to the prescribed camrt aerosol files and physics packages (ie none and camrt)...

 

[eaton]

The camrt radiation is only supported in the cam4 physics package.I tried a test with your proposed configuration and the problem I ran into was:(shr_sys_abort) ERROR: Attempt to find undefined name in pbuf CICEWP

This is triggered by the camrt code looking for an in-cloud ice water path variable (CICEWP) which is only being computed when the RK macro/micro physics option is on (RK is a cam4 package).  To run camrt with the MG microphysics the appropriate cloud variables required to compute the cloud optics will need to be accessed by the camrt code.  This is a task that requires expert knowledge of the microphysics and cloud optics parameterizations to be able to set up correctly.

Generally speaking the only supported physics configurations are the ones available via the cam4 or cam5 options.  These produce climate simulations which have been validated by long control runs.  There is a page on the CESM site that points to more specific information about scientifically validated configurations.