[RegCNET] idirect=1

[Solmon Fabien]

Hello Yang Tze

Yang tze wrote:
> Hello,
> I am a new RegCM4 user, and I have a question about idirect =1.
> From the manual, it means the aerosol no coupling to dynamic and
> thermodynamic. However the clear sky surface and top of atmosphere
> aerosol radiative forcings are diagnosed. But I can't understand how
> the aerosol don't couple to the dynamic and thermodynamic. Can anybody
> give me some detail information about this? Thank you very much!
During the run and for a given time step, the radiation scheme is called
two times , one with aerosol set to zero and one with aerosol. AT each
radiative time step, the radiative forcing is calculated by the
difference of TOA and surface net flux wit and without aerosol. Then ,
If idirect equal 1 the flux and radiative temperature tendency (
calculated as the divergence of the flux) from the no aerosol call are
passed back to the model, so the model dynamic is not perturbed by
aerosol in this case. If idirect = 2 , then i the flux from the with
aerosol call are passed back and the dynamic is perturbed by aerosol.

So in fine with this method, the true radiative forcing is obtained with
idirect =1 because
1 )the aerosol and no aerosol radiative calculation are made
simultaneously with exactly the same atmospheric (T,h,p,q) profiles and
surface albedo, and
2 ) these profiles are not pertubed by aerosols during the run.


So my advice is your control run should be idirect = 1 ( with diagnostic
of aerosol radiative forcing at this stage in the CHE files ).
and the pertubed run is idirect = 2

The option idirect = 0 is to do some aerosol transport only and is just
a little bit faster to run compared to idirect = 1 since it does not
calculate the on line the radiative forcing . Practically I must admit
that this option (idirect = 0) is not very usefull, since it is nice to
have the forcing calculated anyway !

Let me also finally stress that you don't obtain a correct radiative
forcing if you were just do the difference of aerosol - control RAD
fluxes in a postprocessing stage (like many people could be tempted to
do) because in this case you include the dynamic and cloud feedbacks in
this difference.

Regards

FAbien



> Best Wishes
> Yang tze
>
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> RegCNET mailing list
> RegCNET at lists.ictp.it
> https://lists.ictp.it/mailman/listinfo/regcnet