UCX chemistry mechanism

On this page, we provide information about the Universal tropospheric-stratospheric Chemistry eXtension (UCX) mechanism in GEOS-Chem. UCX combines both tropospheric and stratospheric reactions into a single chemistry mechanism. It was developed by Sebastian Eastham, Debra Weisenstein, and Steven Barrett at the Laboratory for Aviation and the Environment at the Massachusetts Institute of Technology.

NOTE: The UCX chemistry mechanism is included in the "benchmark" chemistry mechanism, which is selected by default when you compile and run GEOS-Chem "out of the box". The "benchmark" mechanism adds extra SOA species that are not present in the original UCX implementation. For more information, see our GEOS-Chem chemistry mechanisms wiki page.

Overview

The UCX adds online stratospheric chemistry to GEOS-Chem. Switching on the UCX results in the following changes to GEOS-Chem:

• Photolysis and UV transfer is calculated using the Fast-JX v7.0a model in 18 wavelength bins, covering 177-850 nm. Switching off the UCX removes 6 of the high-frequency bins for the sake of computational speed
• In-model column ozone can now be used to estimate ozone optical depth for photolysis calculations, allowing photochemistry to couple to modeled ozone depletion or production
• Long-lived species such as N₂O, CH₄, OCS, CFCs, HCFCs and halons are emitted, advected and lost as part of the chemistry mechanism, instead of ignored or set to fixed mixing ratios
• Stratospheric water is treated as an advected tracer, and can be enhanced by changes in tropospheric meteorology and upwelling CH₄
• Stratospheric aerosols (LBS, STS, NAT and ice) are formed according to thermodynamics equilibrium. These aerosols are advected, interact with radiation, can settle (based on a log-normal approximation) and take part in heterogeneous chemistry as described in Kirner et al 2011
• A mesospheric steady-state approximation is applied to NO_y and SO_x to prevent formation of an unrealistic high-altitude reservoir while conserving total nitrogen and sulfur

All other mesospheric chemistry is approximated using the linearized GMI mechanism. In total, the UCX decreases run speed by about 50% in a 4x5 simulation. For more information on the UCX please see the following documents, which describe the UCX mechanism in more detail:

1. Eastham, S.D., Weisenstein, D.K., Barrett, S.R.H., Development and evaluation of the unified tropospheric–stratospheric chemistry extension (UCX) for the global chemistry-transport model GEOS-Chem, Atmos. Env., June 2014. (Article)
2. Technical Note: The GEOS-Chem Unified Tropospheric-Stratospheric Chemistry Extension (UCX): Interim Distribution, September 2013 (Article)

Running with the UCX

Running GEOS-Chem with online stratospheric chemistry requires the following:

• The code must be compiled with the option UCX=yes, e.g. make -j4 UCX=yes GRID=4x5 MET=GEOS5
• The tracer restart file must use the native vertical resolution (72 layers in MERRA, GEOS-5 and GEOS-FP) and have the additional UCX tracers (total count of 92 in v10-01)
• The lines Use UCX strat. chem?, Online CH4 chemistry? and Active strat. H2O? should be set to T in the input.geos chemistry menu. The lines regarding LINOZ and linearized stratospheric chemistry should be left on, as they will continue to operate in the mesosphere. The additional stratospheric tracers must also be listed in input.geos

Alternatively, just download the source code with UCX options already specified, as described below.

--Sebastian D. Eastham 10:15, 11 May 2015 (EDT)

Obtaining the source code

Provisional code

Steven Barrett wrote:

As a few people have asked for the code for the strat chem version of GEOS-Chem, we've posted it as a simple compressed file here as a temporary location so they can use it right away until it gets integrated into the main code. [Until then], we'll just keep track of any fixes/issues that come up until then.

--Bob Y. 13:44, 30 September 2013 (EDT)

Standard code

UCX has been added to GEOS-Chem v10-01c, which was tested with 1-month benchmark simulations approved on 29 May 2014 and 1-year benchmark simulations approved on 26 June 2014. It is included in the GEOS-Chem v10-01 public release.

--Bob Y. 14:44, 25 April 2014 (EDT)

Spinning up the stratosphere

The default settings in the Emissions Menu of input.geos are:

=> Set initial glob MRs:---
   => strat. H2O?      : T
   => CH4?             : T
   => OCS?             : T
   => CFCs?            : T
   => Cl species?      : T
   => strat Bry (GCCM)?: T
   => Br species?      : F
   => strat. Br?       : F
   => strat. NOx/HNO3? : T
   => N2O?             : T
   => strat. SO4?      : T
=> CFC emission year   : 0

This may be problematic when splitting your GEOS–Chem simulation into several stages in order to fit within the time limits imposed by your batch queuing system.

Sebastian Eastham wrote:

[The above settings] means that the stratosphere never really spun up. Overwriting strat H2O is definitely advisable for every run – allowing stratospheric H2O to evolve is an experiment at best - and resetting the long-lived species (CH4/OCS/CFCs/N2O) is unlikely to make a difference, but Cl species, all Br, strat NOx/HNO3 and strat SO4 probably need to be allowed to evolve on their own after the initial setup. Stratospheric NOx/HNO3 in particular takes about a year or more to sort itself out after the original run – the strat in general takes up to 5 years to spin up properly due to the long transport times and slow circulation. Any of these flags being set to T results in the restart file being ignored for those species, either within the stratosphere or over the whole domain (depending on the flag), such that 6 2-month runs will not yield the same results as 1 12-month run if any of them are true.

In order to properly spin up the stratosphere, the following settings are recommended for run stages 2 and beyond:

=> Set initial glob MRs:---
   => strat. H2O?      : F
   => CH4?             : F
   => OCS?             : F
   => CFCs?            : F
   => Cl species?      : F
   => strat Bry (GCCM)?: F
   => Br species?      : F
   => strat. Br?       : F
   => strat. NOx/HNO3? : F
   => N2O?             : F
   => strat. SO4?      : F
=> CFC emission year   : 0

Sebastian Eastham wrote:

Regarding H2O, the only truly consistent approaches would be either to treat H2O identically to the other species, or to set Active strat. H2O: F throughout. For a single year benchmark the former approach would probably be fine – the stratospheric H2O does diverge from meteorology over time but the anomaly isn’t too bad over short time periods, and allowing stratospheric dehydration is a useful feature of the UCX.

--Melissa Sulprizio 11:18, 8 September 2014 (EDT)

Previous issues that are now resolved

Only allocate certain fields of State_Chm for UCX-based simulations

This update will be added into the the v11-01 provisional release.

In GEOS-Chem v10-01 and higher versions, the 3-D fields State_Chm%KHETI_SLA and State_Chm%STATE_PSC are always allocated. But these fields are only relevant to full-chemistry simulations using the UCX mechanism (e.g. Standard, UCX) and are not used otherwise. This causes GEOS-Chem to use extra, unnecessary memory for simulations that do not use the UCX mechanism.

We have therefore made the following modifications (shown in GREEN in routine Init_State_Chm (located in module Headers/state_chm_mod.F90). This will make sure that we only allocate memory to these fields if we are using the UCX mechanism:

    ! Fields for UCX mechanism
    State_Chm%STATE_PSC   => NULL()
    State_Chm%KHETI_SLA   => NULL()

    . . .

    !=====================================================================
    ! Allocate and initialize fields for UCX mechamism
    !=====================================================================
#if defined( UCX )

    ALLOCATE( State_Chm%STATE_PSC ( IM, JM, LM                     ), STAT=RC )
    IF ( RC /= GC_SUCCESS ) RETURN
    State_Chm%STATE_PSC = 0.0_f4

    ALLOCATE( State_Chm%KHETI_SLA ( IM, JM, LM, 11                 ), STAT=RC )
    IF ( RC /= GC_SUCCESS ) RETURN
    State_Chm%KHETI_SLA = 0.0_fp

#endif

--Bob Yantosca (talk) 18:32, 28 November 2016 (UTC)

Fix for mesospheric NOx chemistry

This fix will be included in v11-01j

Sebastian Eastham wrote:

The first fix is to the call to UCX_NOX itself. In the current version of GEOS-Chem, the call to UCX_NOX in the FlexChem module occurs while the species array is in molec/cm3, but UCX_NOX expects kg. Moving the call to be after the conversion fixes this issue.

The second fix is to re-establish day-time steady-state for NOx in the mesosphere. This code was already present, but was not functioning correctly, as only NO was being used to estimate total NOx. Partitioning of NOx between N, NO, NO2 and NO3 is now calculated directly in the mesosphere. This assumes that the species will be locally in steady state as long as there is daylight, and that this steady state occurs on a much faster timescale than loss by the reactions N + NO -> N2 + O or N + NO2 -> N2O + O. O(1D) and O(3P) concentrations are taken from archived 2D data, along with J-rates for NO, NO2, NO3 and N2O. The O3 concentration is taken from the species array. Loss rates for NOx and N2O (in addition to the rate of NOx production from N2O) are also estimated, based on the aforementioned loss reactions and destruction of N2O by O(1D) and photolysis.

Finally, I’ve updated the aforementioned chemistry scheme to improve accuracy. The first change is that the J-rates are no longer taken from 2D data. Instead, the J-rate calculated at the top of the stratosphere is applied vertically throughout the column. This mirrors the approach used to calculate mesospheric H2SO4 photolysis. The second change is that a heavily simplified NOx chemistry mechanism is applied during mesospheric night. Previously, it was assumed that there would be no mesospheric NOx chemistry at night. However, this led to an overestimate of N and NO, and an underestimate of NO2 and NO3 (especially at high latitudes). It is now assumed that when there is no sunlight there is negligible N, O(1D) and O(3P). This reduces the mesospheric NOx mechanism to a 3-reactant chain. All local N is converted immediately to NO, and then only the two following reactions occur: NO + O3 -> NO2 + O2 and NO2 + O3 -> NO3 + O2. The evolution of the partitioning between these species is too slow to be well-represented by steady state, but the simplicity of the mechanism allows for an analytical solution. The extra overhead should be effectively zero, as it requires the evaluation of only three exponentials in each mesospheric night-time grid cell.

--Melissa Sulprizio (talk) 16:27, 21 November 2016 (UTC)

Fix for black carbon in ucx_mod.F

This fix was validated with 1-month benchmark simulation v11-01b and 1-year benchmark simulation v11-01b-Run0. This version was approved on 19 Aug 2015.

Sebastian Eastham wrote:

After one hour of simulation with the standard initial restart file, a layer of black carbon appears at the stratopause, and the mesopause empties of black carbon entirely (although I’m not sure why there is so much up there to begin with). This layer is due to an error in the sedimentation calculation. On line 2699 of ucx_mod.F, the line

RWET(IBC) = RAA(IND999,29)

Should be changed to

RWET(IBC) = RAA(IND999,29) * 1.0e-6_fp

This was just a unit conversion error, and has probably been there since the UCX was first developed. I only noticed it because there seems to be a lot more high altitude black carbon in the model now than there was in v9-01-03, when the UCX was developed. I’m still looking into why this is the case.

--Melissa Sulprizio (talk) 16:06, 13 July 2015 (UTC)

Reactivate bromine species photolysis for tropospheric simulations

This update was validated in the 1-month benchmark simulation v10-01c and approved on 29 May 2014.

In the original Fast-JX input files, photolysis for several bromine species had been deactivated for tropospheric simulations. A full description of the error and its fix may be found in this post on our Fast-JX v7.0 photolysis mechanism wiki page.

--Melissa Sulprizio 13:43, 14 May 2014 (EDT)

Error in Fast-JX for tropospheric only simulation

This update was validated in the 1-month benchmark simulation v10-01c and approved on 29 May 2014.

An bug in the original Fast-JX code caused the cross-sections to be interpolated incorrectly for tropospheric-only chemistry simulations. A full description of this error may be found in this post on our FAST-JX v7.0 photolysis mechanism wiki page.

--Bob Y. 14:27, 20 May 2014 (EDT)

Modify treatment of N2O5 + H2O reaction in calcrate.F

This update was validated in the 1-month benchmark simulation v10-01c and approved on 29 May 2014.

NOTE: Routine calcrate.F is slated for removal from GEOS-Chem v11-01 once the new FlexChem solver is implemented.

The reaction for N2O5 + H2O was found to be problematic in long-term UCX simulations where the H2O concentration (stored in the EDUCTCONC variable in file GeosCore/calcrate.F) was low and caused the solver to choke. Sebastian Eastham suggested that we include the following fix in the standard code.

Sebastian Eastham wrote:

The division is necessary to compensate for the change from N2O5 -> 2HNO3 to N2O5 + H2O -> 2HNO3—it's still "as in base GEOS-Chem" though in the sense that it doesn't check to see if N2O5 or H2O is the limiting reactant (and possibly herein lies the problem!). It might be worth seeing if the treatment used for the other aerosol reactions (not including the check against parameter PSCMINLIFE) would help.

Following Sebastian's suggestion, we have made the following modification in this section of calcrate.F for better numerical stability:

                  ! If all these checks are passed, we set k = ki/[B].
                  ! Rxn 11 is first-order - ignore
                  IF (PSCIDX.eq.1) THEN
!------------------------------------------------------------------------------
! Prior to 5/12/14:
! Use same treatment used for other aerosol reactions (not including the
! check against PSCMINLIFE) (sde, mps, 5/12/14)
!                     ! N2O5 + H2O is an old reaction, allow to proceed
!                     ! as it did in base GEOS-Chem
!                     IF (IS_SAFE_DIV(ADJUSTEDRATE,EDUCTCONC)) THEN
!                        ADJUSTEDRATE = ADJUSTEDRATE/EDUCTCONC
!                     ELSE
!                        ADJUSTEDRATE = 0d0
!                     ENDIF
!------------------------------------------------------------------------------
                     ! Convert from 1st-order to 2nd-order
                     IF (IS_SAFE_DIV(EDUCTCONC,LIMITCONC)) THEN
                        ! Temporarily store [B]/(T*[A])
                        LIMITCONC = EDUCTCONC/(PSCMINLIFE*LIMITCONC)
                        IF (ADJUSTEDRATE.gt.LIMITCONC) THEN
                           ADJUSTEDRATE = LIMITCONC
                        ENDIF
                     ELSE
                        ADJUSTEDRATE = 0d0
                     ENDIF
                     IF ((EDUCTCONC.gt.1.d2).and.
     &                      (IS_SAFE_DIV(ADJUSTEDRATE,EDUCTCONC))) THEN
                        ADJUSTEDRATE = ADJUSTEDRATE/EDUCTCONC
                     ELSE
                        ADJUSTEDRATE = 0d0
                     ENDIF
   

--Bob Y. 14:56, 13 May 2014 (EDT)

Read 2D data for individual NOx species in ucx_mod.F

This update was validated with the 1-month benchmark simulation v10-01f and approved on Approved 13 Jan 2015.

Sebastian Eastham wrote:

A quick and easy improvement can be made now that NOx has been split out into separate species. In ucx_mod.F at line 504, the NO tracer is currently given the full NOx burden – however, the file from which they are read, DiurnalAvg.nc, contains NO, NO2, HNO2 (as HONO) and NO3 separately, as well as lumped – with HNO3 and PAN – into the single variable ‘NOx’. The attached file, a specifically updated version of ucx_mod.F based on my download of Code.tar.gz, would fully implement the update with a change of only about 15 lines (which also prevents a doubling of HNO3 at one pole upon initialization, since the old “NOx” value from the 2D model includes PAN and HNO3).

In routine SET_INITIAL_MIXRATIOS, change the following lines:

        ! Stratospheric NOy
        IF (LSETNOYSTRAT) THEN
           IF ( am_I_Root ) THEN
              WRITE(6,*) '|NOx      |HNO3     |         |'
              WRITE(6,*) '|---------|---------|---------|'
           ENDIF

           ! No Montreal scaling
           SCALEFAC = 1.d0

           TRAC_IDX = IDTHNO3
           TRAC_SET = 'HNO3'
           CALL APPLY_2DTRAC( .FALSE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )

           TRAC_IDX = IDTNO
           TRAC_SET = 'NOX'
           CALL APPLY_2DTRAC( .FALSE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )
        ENDIF

to:

        ! Stratospheric NOy
        IF (LSETNOYSTRAT) THEN
           IF ( am_I_Root ) THEN
              WRITE(6,*) '|NO       |NO2      |NO3      |'
              WRITE(6,*) '|---------|---------|---------|'
              WRITE(6,*) '|HNO2     |HNO3     |HNO4     |'
              WRITE(6,*) '|---------|---------|---------|'
              WRITE(6,*) '|N2O5     |PAN      |         |'
              WRITE(6,*) '|---------|---------|---------|'
           ENDIF

           ! No Montreal scaling
           SCALEFAC = 1.d0

           TRAC_IDX = IDTNO
           TRAC_SET = 'NO'
           CALL APPLY_2DTRAC( .FALSE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )

           TRAC_IDX = IDTNO2
           TRAC_SET = 'NO2'
           CALL APPLY_2DTRAC( .FALSE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )

           ! Unstable product of ETO2 + NO2 gets lumped into NO2 for now
           TRAC_SET = 'ETHO2NO2'
           CALL APPLY_2DTRAC( .TRUE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )

           TRAC_IDX = IDTNO3
           TRAC_SET = 'NO3'
           CALL APPLY_2DTRAC( .FALSE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )

           TRAC_IDX = IDTN2O5
           TRAC_SET = 'N2O5'
           CALL APPLY_2DTRAC( .FALSE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )

           TRAC_IDX = IDTHNO2
           TRAC_SET = 'HONO'
           CALL APPLY_2DTRAC( .FALSE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )

           TRAC_IDX = IDTHNO3
           TRAC_SET = 'HNO3'
           CALL APPLY_2DTRAC( .FALSE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )

           TRAC_IDX = IDTHNO4
           TRAC_SET = 'HO2NO2'
           CALL APPLY_2DTRAC( .FALSE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )

           TRAC_IDX = IDTPAN
           TRAC_SET = 'PAN'
           CALL APPLY_2DTRAC( .FALSE., .TRUE., .FALSE., SCALEFAC,
    &                         Input_Opt, State_Met, State_Chm )
        ENDIF

--Melissa Sulprizio 11:23, 18 June 2014 (EDT)