[BUG/ISSUE] Bug when running with TOMAS and chemistry time step under 1 hr

[danamcguff]

Describe the bug

In AEROPHYS subroutine of tomas_mod, the overall time step is 1 hour.
Here is a permalink to the hard-coded bug:

geos-chem/GeosCore/tomas_mod.F

Lines 703 to 711 in 6048e58

	! Get info on this grid box
	ADT = 3600.
	PRES = State_Met%PMID(i,j,l)*100.0 ! in Pa
	TEMPTMS = State_Met%T(I,J,L)
	BOXMASS = State_Met%AD(I,J,L)
	RHTOMAS = State_Met%RH(I,J,L)/ 1.e2
	IF ( RHTOMAS > 0.99 ) RHTOMAS = 0.99
	BOXVOL = State_Met%AIRVOL(I,J,L) * 1.e6 !convert from m3 -> cm3

I suggest changing the definition of the time step so line 704 reads:
ADT = GET_TS_CHEM()
You will also need to add a line of code to the beginning of AEROPHYS subroutine such as:
USE Time_Mod, ONLY : GET_TS_CHEM

Expected behavior

When running GC with TOMAS on and using a time step smaller than 1 hour for chemistry, the simulated aerosol microphysics are faster than they should be.

Required information

• GEOS-Chem version : discovered in 12.3.2 but also in current release (12.7.0)

[yantosca]

Thanks @danamcguff for bringing this to our attention. This is now fixed by commit 7a7f3d7, which will go into GEOS-Chem 12.7.1.

I also made a couple more updates as seen here:

geos-chem/GeosCore/tomas_mod.F

Lines 670 to 701 in 7a7f3d7

	! Get chemistry timestep for use below [s]
	! NOTE: This doesn't have to be !$OMP+PRIVATE (bmy, 2/7/20)
	ADT = GET_TS_CHEM()
	!%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
	!%%% NOTE: THIS PARALLEL LOOP MAY BE ABLE TO BE REVERSED TO L-J-I
	!%%% WHICH IS MUCH MORE EFFICIENT (bmy, 1/28/14)
	!%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
	!$OMP PARALLEL DO
	!$OMP+DEFAULT( SHARED )
	!$OMP+PRIVATE( I, J, L )
	!$OMP+PRIVATE( PRES, TEMPTMS, BOXMASS, RHTOMAS, BOXVOL )
	!$OMP+PRIVATE( printneg, ionrate, lev, weight, GC, N, NK, JC )
	!$OMP+PRIVATE( MK, H2SO4rate_o, tot_n_1, k, tot_s_1, MPNUM )
	!$OMP+PRIVATE( Nkd, Mkd, TOT_NK, TOT_MK, TRANSFER )
	!$OMP+PRIVATE( Nkout,Mkout,Gcout,fn,fn1 )
	!$OMP+PRIVATE( num_iter,Nknuc,Mknuc,Nkcond )
	!$OMP+PRIVATE( Mkcond, ERRORSWITCH, tot_s_1a, tot_n_1a )
	!$OMP+PRIVATE( ERR_VAR, ERR_MSG, ERR_IND )
	!$OMP+PRIVATE( TRACNUM, NH3_TO_NH4, SURF_AREA )
	!$OMP+SCHEDULE( DYNAMIC )
	DO I = 1, State_Grid%NX
	DO J = 1, State_Grid%NY
	DO L = 1, State_Grid%NZ
	#ifdef BPCH_DIAG
	! Reset the AD61_INST array used for tracking instantaneous
	! certain rates. As of now, tracking nucleation (win, 10/7/08)
	IF ( Input_Opt%ND61 > 0 ) THEN
	AD61_INST(I,J,L,:) = 0e0
	ENDIF
	#endif

1. I moved ADT out of the parallel loop, since it does not need to be held PRIVATE (i.e. its value is the same for all threads). This will prevent calling the GET_TS_CHEM function on each (I,J,L) iteration.

2. I had to add an IF statement around the AD61_INST array. This will prevent a segfault if ND61 is not turned on.