Hi,
I am using MPAS-A 8.2.2 with thompson aerosol aware scheme. I use the default input file QNWFA_QNIFA_SIGMA_MONTHLY.dat to generate .init file. nwfa in the lowest layer is shown in figure 1. The mesh is 15 km globally in the top panel, and 15 km globally with 3 km regional-refined mesh over East Asia and North Pacific in the bottom panel.
(figure 1)
After model integration, a restart file is generated. When I check the restart file, I see there are differences in nwfa2d when using different meshes. nwfa2d is much lower in the region where a refined mesh is applied (marked with red box in figure2).
(figure 2)
And it shows decreased nwfa concentration in the refined area in the lowest level after 2 days' simulation. See figure 3.
(figure 3)
To understand why nwfa2d is lower in the refined region, I explored the source code. In core_atmosphere/physics/physics_wrf/module_mp_thompson.F, nwfa is updated using nwfa2d in the following line:
Where nwfa has unit nb kg^{-1} and nwfa2d has unit nb kg^{-1} s^{-1}
When I check how nwfa2d is calculated in core_atmosphere/physics/mpas_atmphys_init_microphysics.F,
Where nwfa has unit nb kg^{-1} but nwfa2d has unit nb s^{-1} in the grid. Here nwfa2d takes areaCell into account when calculating emission over the whole grid (rather than per kg air). In the refined region, areaCell is lower than elsewhere, and I hypothesize that it is the reason nwfa2d is lower in the red box in the bottom panel (refined) than the top panel (not refined) in figure 2.
So it seems like MPAS-A uses different units for nwfa2d calculation (nb s^{-1} in the grid in mpas_atmphys_init_microphysics.F) and emission (nb kg^{-1} s^{-1} in module_mp_thompson.F).
I'd like your input to help me solve the units problem, or correct me if I misunderstand the nwfa2d calculation process.
Thanks,
Qi
I am using MPAS-A 8.2.2 with thompson aerosol aware scheme. I use the default input file QNWFA_QNIFA_SIGMA_MONTHLY.dat to generate .init file. nwfa in the lowest layer is shown in figure 1. The mesh is 15 km globally in the top panel, and 15 km globally with 3 km regional-refined mesh over East Asia and North Pacific in the bottom panel.
(figure 1)
After model integration, a restart file is generated. When I check the restart file, I see there are differences in nwfa2d when using different meshes. nwfa2d is much lower in the region where a refined mesh is applied (marked with red box in figure2).
(figure 2)
And it shows decreased nwfa concentration in the refined area in the lowest level after 2 days' simulation. See figure 3.
(figure 3)
To understand why nwfa2d is lower in the refined region, I explored the source code. In core_atmosphere/physics/physics_wrf/module_mp_thompson.F, nwfa is updated using nwfa2d in the following line:
Code:
nwfa1d(kts) = nwfa1d(kts) + nwfa2d(i,j)*dt_inWhen I check how nwfa2d is calculated in core_atmosphere/physics/mpas_atmphys_init_microphysics.F,
Code:
!... scale the lowest level aerosol data into an emissions rate. This is very far from ideal, but
!... need higher emissions where larger amount of (climo) existing and lesser emissions where there
!... exists fewer to begin as a first-order simplistic approach. Later, proper connection to emission
!... inventory would be better, but, for now, scale like this:
!... where: Nwfa=50 per cc, emit 0.875E4 aerosols per second per grid box unit
!... that was tested as ~(20kmx20kmx50m = 2.E10 m**3).
k = 1
do iCell = 1, nCellsSolve
airmass = rho_zz(k,iCell)*zz(k,iCell)
airmass = airmass*(zgrid(k+1,iCell)-zgrid(k,iCell))*areaCell(iCell) ! (in kg)
nwfa2d(iCell) = nwfa(k,iCell)*0.000196*airmass*0.5e-10
nifa2d(iCell) = 0._RKIND
! call mpas_log_write('$i $r $r $r',intArgs=(/iCell/),realArgs=(/airmass,nwfa2d(iCell),nifa2d(iCell)/))
enddo
!... deallocate local arrays:
if(allocated(hgt)) deallocate(hgt)
!call mpas_log_write('--- end subroutine init_thompson_aerosols_forMPAS.')
end subroutine init_thompson_aerosols_forMPASWhere nwfa has unit nb kg^{-1} but nwfa2d has unit nb s^{-1} in the grid. Here nwfa2d takes areaCell into account when calculating emission over the whole grid (rather than per kg air). In the refined region, areaCell is lower than elsewhere, and I hypothesize that it is the reason nwfa2d is lower in the red box in the bottom panel (refined) than the top panel (not refined) in figure 2.
So it seems like MPAS-A uses different units for nwfa2d calculation (nb s^{-1} in the grid in mpas_atmphys_init_microphysics.F) and emission (nb kg^{-1} s^{-1} in module_mp_thompson.F).
I'd like your input to help me solve the units problem, or correct me if I misunderstand the nwfa2d calculation process.
Thanks,
Qi