Wet removal process of aerosols

[weiwilliam]

Hello all,

I had some hard times understanding the wet removal processes of aerosols in physics/smoke_dust/module_wetdep_ls.F90 and physics/CONV/SAMF.
I know physics/smoke_dust/module_wetdep_ls.F90 is the same as the one used in WRF-Chem.

ccpp-physics/physics/smoke_dust/module_wetdep_ls.F90

Lines 54 to 71 in 1ab4de8

	if(rain(i,j).gt.1.e-10)then
	! convert rain back to rate
	!
	rain_clw(i,j)=rain(i,j)/dt
	! total cloud water
	!
	do k=1,kte-1
	dvar=max(0.,moist(i,k,j,p_qc)*rho(i,k,j)*vvel(i,k,j)*dz8w(i,k,j))
	var_sum_clw(i,j)=var_sum_clw(i,j)+dvar
	var_sum(i,j)=var_sum(i,j)+var(i,k,j,nv)*rho(i,k,j)
	enddo
	if(var_sum(i,j).gt.1.e-10 .and. var_sum_clw(i,j).gt.1.e-10 ) then
	! assuming that frc is onstant, it is my conversion factor
	! (just like in convec. parameterization)
	frc(i,j)=rain_clw(i,j)/var_sum_clw(i,j)
	frc(i,j)=max(1.e-6,min(frc(i,j),.005))
	endif
	endif

Can anyone help explain the physical meaning of this part of the code?
It seems that frc is not a unitless variable, and it puzzled me a lot.

For the aerosol wet removal in the SAMF scheme, does anyone have the reference for it?
And why are both samfdeepcnv_aerosols and samfshalcnv_aerosols commented out?

ccpp-physics/physics/CONV/SAMF/samfdeepcnv.f

Lines 3089 to 3093 in 1ab4de8

	! if (do_aerosols)
	! & call samfdeepcnv_aerosols(im, im, km, itc, ntc, ntr, delt,
	! & xlamde, xlamdd, cnvflg, jmin, kb, kmax, kd94, ktcon, fscav,
	! & edto, xlamd, xmb, c0t, eta, etad, zi, xlamue, xlamud, delp,
	! & qtr, qaero)

ccpp-physics/physics/CONV/SAMF/samfshalcnv.f

Lines 2084 to 2089 in 1ab4de8

	! & call samfshalcnv_aerosols(im, im, km, itc, ntc, ntr, delt,
	!! & xlamde, xlamdd, cnvflg, jmin, kb, kmax, kbcon, ktcon, fscav,
	! & cnvflg, kb, kmax, ktcon, fscav,
	!! & edto, xlamd, xmb, c0t, eta, etad, zi, xlamue, xlamud, delp,
	! & xmb, c0t, eta, zi, xlamue, xlamud, delp,
	! & qtr, qaero)

Appreciate any information that can help me improve my understanding of this aspect.
Thank you!

[zhanglikate]

Yes, the large scale wet removal scheme originated from WRF-Chem, though I'm uncertain about the original developer. My understanding based on the code structure is that this method operates as a diagnostic offline parameterization rather than being integrated with the host model's microphysical processes. It calculates a scavenging coefficient from the ratio of precipitation rate to column-integrated cloud water flux, then applies removal layer-by-layer where cloud condensate exists. While it connects wet removal to precipitation, cloud water, and vertical velocity, it does so through diagnosed fields rather than through the prognostic microphysical tendencies that actually generate clouds and precipitation.

This fundamental disconnect is why we developed the inline wet removal approach integrated directly within the Thompson microphysics scheme for UFS-Chem model in our recent development.

    Also, I think frc is a unitless variable from my understanding :

frc = rain_clw / var_sum_clw

frc = (kg m⁻² s⁻¹) / (kg m⁻² s⁻¹) = dimensionless

rain_clw = rain / dt , rain is kg m⁻² (1 mm of water depth = 1 kg m⁻² for water) , dt is s⁻¹

var_sum_clw= var_sum_clw + dvar
dvar= max(0., (moist_arr * rho * vvel * dz8w) = (kg kg⁻¹ *kg m⁻³ * m s⁻¹ * m ) = kg m⁻² s⁻¹

Thanks.

[zhanglikate]

I 've included @haiqinli here as this relates to the smoke_dust component. My understanding is that RRFSv1 ultimately uses the SAS scheme for convection rather than GF, which may be why SAMF is commented out. Haiqin would be the best person to confirm this.

In GEFS-Aerosols, UFS-Chem , GCAFS, as well as RRFSv1, they all use SAS convection scheme for convective transport and convective wet deposition

[weiwilliam]

Thank you, @zhanglikate
That is really helpful.
Is frc like how much rain formed under the integrated vertical movement of cloud liquid water?
But it actually has units of m-1.

dvar= max(0., (moist_arr * rho * vvel * dz8w) = (kg kg⁻¹ *kg m⁻³ * m s⁻¹ * m ) = kg m⁻² s⁻¹

dvar is kg m-1 s-1, because there are two "meter" from vertical velocity and thickness.

@haiqinli, please share any insights you have on SAMF development. Thank you!

[haiqinli]

@weiwilliam I am responsible for the development and maintenance of the GF (currently GFL) convection. Please feel free to email me or schedule a time to meet if you would like to discuss this further. Thanks!

[weiwilliam]

@haiqinli Thank you! I will send you an email.

[zhanglikate]

@weiwilliam You're right. I made a mistake in my quick calculation—the unit of frc is m⁻¹. My understanding is that frc represents a scavenging coefficient, and it's physically reasonable for it to have units of m⁻¹ (per unit height). However, I don't quite understand the factor/(1+factor) term that follows, since factor is not unitless. Thank you for catching this.

[weiwilliam]

@zhanglikate Thank you! That really makes sense. It is like the extinction coefficient.

ccpp-physics/physics/smoke_dust/module_wetdep_ls.F90

Lines 81 to 83 in 1ab4de8

	if(var(i,k,j,nv).gt.1.e-16 .and. moist(i,k,j,p_qc).gt.0.)then
	factor = max(0.,frc(i,j)*rho(i,k,j)*dz8w(i,k,j)*vvel(i,k,j))
	dvar=alpha*factor/(1+factor)*var(i,k,j,nv)

Can we consider this part as the scavenged aerosol flux at each level?
As frc is bounded between 0.005 and 1e-16, factor/(1+factor) is approximately factor except it is dimensionless.
Then, the next line becomes

factor/(1+factor)*var ~ factor * var => (frc * dz8w) * (rho * vvel * var)

The frc * dz8w can be considered as the total scavenging effect in a layer, and rho * vvel * var is the aerosol flux.
Then the dvar is the scavenged amount given the aerosol influx of each layer.
Do you think it makes sense?

[zhanglikate]

@weiwilliam I agree that the "factor/(1+factor)" formulation is attempting to create a dimensionless fraction by canceling units, perhaps by treating the '1' as having the same units as factor rather than being truly dimensionless. :)