Particulate matter in GEOS-Chem

From Geos-chem
Jump to: navigation, search

On this page we provide information about how to compute particulate matter concentrations from GEOS-Chem output.


Randall Martin wrote, with contributions from Daven Henze, Aaron van Donkelaar, and Jeff Pierce:

The PM2.5 calculation should account for aerosol water of PM2.5 in a way that is consistent with the PM2.5 measurements. The water content of PM2.5 is operationally defined. In the United States an RH of 35% is often used and would be appropriate for the benchmark. In Europe an RH of 50% is often used.
Below are the factors that account for aerosol water in a consistent manner as used in jv_spec.dat from GC v10-01-01, as calculated by Aaron van Donkelaar and Sajeev Philip.
At 35% RH:
1.33 for SO4, NIT, and NH4
1.16 for OCPI and SOA
1.86 for SALA
At 50% RH the values are
1.51 for SO4, NIT, and NH4
1.24 for OCPI and SOA
2.42 for SALA
These growth factors are calculated using the change in radius between different RH within jv_spec.dat. Essentially, the change in radius between the dry (i.e. 0% RH) and wet (35% or 50% RH) aerosol is treated as a shell of water for the purposes of calculating the additional mass associated with the wet particle. Under this condition, it can be shown that:
WetMass2DryMassRatio = 1 + [{(radiusAtRH_wet / radiusAtRH_dry)^3 - 1} x (Density_Water / Density_DrySpecies)]
The DST2 bin includes aerosols with diameter both smaller and larger than 2.5 um. Lengthy discussion with Duncan Fairlie, Aaron van Donkelaar, Colette Heald, Jeff Pierce and Noelle Selin led to the conclusion that 38% of the DST2 bin should be included in the calculation of PM2.5.
In summary, the PM2.5 benchmark and wiki documentation should be changed to
  PM25     = 1.33 (NH4 + NIT  + SO4) + BCPI + BCPO + 2.1 (OCPO + 1.16 OCPI) + 1.16 SOA  
             + DST1 + 0.38 DST2 + 1.86 SALA
where the tracer units in ug/m3 at STP, and the value of 2.1 is the global mean OM/OC as recommended by the Aerosol WG.

--Melissa Sulprizio (talk) 13:19, 30 June 2016 (UTC)

Including SOA in PM2.5

SOA should be defined to include all SOA tracers used in the benchmark. This will evolve as the Aerosol WG recommends new SOA options. At present, the best way to represent SOA is as

  • TSOA0, TSOA1, TSOA2, TSOA3 = Lumped semivolatile aerosol products of monoterpene + sesquiterpene ox.
  • ISOA1, ISOA2, ISOA3, = Lumped semivolatile aerosol products of isoprene oxidation
  • ASOAN, ASOA1, ASOA2, ASOA3 = Lumped nonvolatile aerosol products of light aromatics and IVOCs

The molecular weights are 150 g/mol for all above SOA tracers.

GEOS-Chem v11-02c, introduced several new SOA species that should be included in the PM2.5 definition.

For the complex SOA scheme:

  • ISOAAQ = isoprene SOA from aqueous formation, includes
    • SOAGX = Aerosol-phase glyoxal (58 g/mol)
    • SOAMG = Aerosol-phase methylglyoxal (72 g/mol)
    • SOAIE = Aerosol-phase isoprene epoxide (118 g/mol)
    • SOAME = Aerosol-phase C4 epoxide from oxidation of PMN (102 g/mol)
    • INDIOL = Generic aerosol-phase organonitrate hydrolysis product (102 g/mol)
    • LVOCOA = Aerosol-phase low-volatility non-IEPOX product of ISOPOOH (RIP) oxidation (154 g/mol)
    • ISN1OA = Aerosol-phase 2nd generation hydroxynitrates formed from ISOP+NO3 reaction pathway (226 g/mol)

For the simple SOA scheme:

  • SOAS = Simple SOA (150 g/mol)

IMPORTANT: When using the complex SOA scheme, the Aerosols WG warns that the PM2.5 calculation in GEOS-Chem currently includes all the SOA formed in both the Pye et al. (2010) and Marais et al. (2016) scheme and may include some double-counting of isoprene SOA.

--Melissa Sulprizio (talk) 19:02, 8 September 2017 (UTC)

Option to include spatially and seasonally varying OM/OC

This update was included in v11-02e (approved 24 Mar 2018).

For users who seek more information on the seasonal and spatial variation of OM/OC in the lower troposphere, we provide the option to use the seasonal gridded dataset developed by Philip et al. (2014). This dataset has some uncertainty, but offers more information than a global-mean OM/OC ratio in regions where primary organic aerosols have a large fossil fuel source.

--Melissa Sulprizio (talk) 16:33, 23 May 2017 (UTC)

PM2.5 diagnostic

The following section describes how PM2.5 is computed for GEOS-Chem diagnostic output:

GEOS-Chem v11-02

This update will be added during the GEOS-Chem v11-02 provisional release period.

Jenny Fisher rightly pointed out that the PM2.5 diagnostic in GEOS-Chem v11-02 was erroneously including the ISOAAQ species in the accounting of PM2.5 when the Simple SOA option was used. After discussion with the Aerosols Working Group, we modified the PM2.5 diagnostic (in routine GeosCore/aerosol_mod.F) accordingly. The code in RED was removed and the code in GREEN added:

         ! Units: [kg/m3]
         PM25(I,J,L) = NH4(I,J,L)        * SIA_GROWTH
     &               + NIT(I,J,L)        * SIA_GROWTH
     &               + SO4(I,J,L)        * SIA_GROWTH
     &               + BCPI(I,J,L)
     &               + BCPO(I,J,L)
     &               + OCPO(I,J,L)
     &               + OCPI(I,J,L)       * ORG_GROWTH
     &               + SALA(I,J,L)       * SSA_GROWTH
     &               + SOILDUST(I,J,L,1)                 ! DST1
     &               + SOILDUST(I,J,L,2)                 ! DST1
     &               + SOILDUST(I,J,L,3)                 ! DST1
     &               + SOILDUST(I,J,L,4)                 ! DST1
     &               + SOILDUST(I,J,L,5) * 0.38          ! 38% of DST2
! Prior to 5/11/18:
! ISOAAQ should only be added to PM2.5 for Complex SOA. as per discussion
! from Aerosol Working Group in May 2018 (bmy, 5/11/18)
!     &               + ISOAAQ(I,J,L)     * ORG_GROWTH    ! Includes SOAGX+SOAMG

         ! Include either simple SOA (default) or Complex SOA in
         ! PM2.5 calculation.  In simulations where both Simple SOA and
         ! Complex SOA species are carried (i.e. "benchmark"), then
         ! only the Simple SOA will be added to PM2.5, in order to avoid
         ! double-counting. (bmy, 5/11/18)
         IF ( Is_SimpleSOA ) THEN
            PM25(I,J,L) = PM25(I,J,L)
     &                  + SOAS(I,J,L)    * ORG_GROWTH
         ELSEIF ( Is_ComplexSOA ) THEN
            PM25(I,J,L) = PM25(I,J,L)
     &                  + TSOA(I,J,L)    * ORG_GROWTH
     &                  + ISOA(I,J,L)    * ORG_GROWTH
     &                  + ASOA(I,J,L)    * ORG_GROWTH
     &                  + ISOAAQ(I,J,L)  * ORG_GROWTH    ! Includes SOAGX+SOAMG


         ! Apply STP correction factor based on ideal gas law
         PM25(I,J,L) = PM25(I,J,L) 
     &               * ( 1013.25_fp / PMID(I,J,L) ) 
     &               * ( T(I,J,L)   / 298.0_fp    )

From the Git commit history message:

To avoid double-counting of SOA, we now add
to the AOD diagnostics only when the complex SOA option is turned on.

Furthermore, for the PM2.5 diagnostic, we do the following:

  1. For simulations using Simple SOA only (Standard, Tropchem), then SOAS (but not ISOAAQ) will be added to the PM2.5 diagnostic;
  2. For simulations using Complex SOA only (complexSOA and complexSOA_SVPOA), then ISOAAQ (but not SOAS) will be added to the PM2.5 diagnostic;
  3. For the Benchmark simulation (which turns on both simple and complex SOA simultaneously) SOAS but not ISOAAQ will be added to the PM2.5 diagnostic. (This is to avoid double-counting.)


  1. This update will cause the Standard, Tropchem, and Benchmark simulations to have differences in the PM2.5 diagnostic w/r/t the prior commit (because SOAS and ISOAAQ together were added to the PM2.5 diagnostic).
  2. The complexSOA and complexSOA_SVPOA simulations will have identical PM2.5 diagnostics w/r/t the prior commit (because ISOAAQ was already being added to the PM2.5 diagnostic).

--Bob Yantosca (talk) 16:23, 15 May 2018 (UTC)

GEOS-Chem v11-01

This update was included in v11-01j and approved on 03 Dec 2016

In GEOS-Chem v11-01j, PM2.5 concentrations were added to the ND42 diagnostic for SOA concentrations. As part of this update, the code for computing SOA concentrations has been moved from diag42_mod.F to routine AEROSOL_CONC (in aerosol_mod.F). PM2.5 is calculated at the end of AEROSOL_CONC following the recommendation by Randall Martin and the Aerosol WG described above. Aerosol growth factors used in the PM2.5 calculation are computed online in the same routine using aerosol densities (stored in the species database) and aerosol radii (stored in REAA in fast_jx_mod.F).

--Melissa Sulprizio (talk) 20:11, 30 November 2016 (UTC)

Save out PM2.5 diagnostic at STP conditions

This update was included in v11-02a and approved on 12 May 2017.

Aaron van Donkelaar wrote:

As currently implemented [in GEOS-Chem v11-01], the PM2.5 diagnostic outputs PM2.5 at ambient conditions. While this is not technically an error, most PM2.5 monitors measure at STP conditions which will cause disagreement during comparison with observations and inconsistency during application of any health response curves (generally determined from the STP observations).
As a result, I’d recommend applying an STP correction factor based on ideal gas law after PM2.5 is calculated in aerosol_mod.F:
        PM25(I,J,L) = PM25(I,J,L) * (1013.25d0 / AIRPRESS(I,J,L)) * (AIRTEMP(I,J,L) / 298)

--Melissa Sulprizio (talk) 18:05, 7 February 2017 (UTC)

PM2.5 in the 1-yr benchmark plots

The following section describes how PM2.5 is plotted in the GEOS-Chem benchmark output.

GEOS-Chem v11-01 and later versions

In GEOS-Chem v11-01 the definition of PM2.5 was changed following the recommendation by Randall Martin and the Aerosol WG as described above. The definition now used to create the plots for GEOS-Chem 1-year benchmark simulations is:

  ; Convert ppbv to ug/m3
  STP_P     = 1013.25
  STP_T     = 298.
  ppb_ugm3  = 1e6 / 8.314 * 100. * STP_P/STP_T *1e-9 
  MWaer     = [18, 12, 12, 62, 96, 29, 31.4] ; NH4, EC, OC, NIT, SO4, DUST, SALA

  NH4_ugm3  = NH4( indlon, indlat, 0)*ppb_ugm3*MWaer(0)
  NIT_ugm3  = NIT( indlon, indlat, 0)*ppb_ugm3*MWaer(3)
  SO4_ugm3  = SO4( indlon, indlat, 0)*ppb_ugm3*MWaer(4)
  BCPI_ugm3 = BCi( indlon, indlat, 0)*ppb_ugm3*MWaer(1) 
  OCPI_ugm3 = OCi( indlon, indlat, 0)*ppb_ugm3*MWaer(2)
  BCPO_ugm3 = BCo( indlon, indlat, 0)*ppb_ugm3*MWaer(1)
  OCPO_ugm3 = OCo( indlon, indlat, 0)*ppb_ugm3*MWaer(2)
  DST1_ugm3 = Dst1(indlon, indlat, 0)*ppb_ugm3*MWaer(5)
  DST2_ugm3 = Dst2(indlon, indlat, 0)*ppb_ugm3*MWaer(5)
  SALA_ugm3 = SALA(indlon, indlat, 0)*ppb_ugm3*MWaer(6)

  ; Compute PM2.5
  PM25 = 1.33 * ( NH4_ugm3 + NIT_ugm3 + SO4_ugm3 ) + &
         ( BCPI_ugm3 + BCPO_ugm3 ) + 2.1 * (1.16 * OCPI_ugm3 + OCPO_ugm3) + &
         1.16*SOA_ugm3 + DST1_ugm3 + (0.38 * DST2_ugm3) + (1.86 * SALA_ugm3)

where the tracer units are ug/m3 at STP, and the value of 2.1 is the global mean OM/OC as recommended by the Aerosol WG.

--Melissa Sulprizio (talk) 13:19, 30 June 2016 (UTC)

GEOS-Chem v10-01 and earlier versions

Here is the definition of PM2.5 that we use to create the plots for GEOS-Chem 1-year benchmark simulations:

  ; Convert ppbv to ug/m3
  STP_P    = 1013.25
  STP_T    = 298.
  ppb_ugm3 = 1e6 / 8.314 * 100. * STP_P/STP_T *1e-9 

 ; Compute PM2.5
 PM25     = ( ( NH4         )       * ppb_ugm3 * 18 )
          + ( ( NIT         )       * ppb_ugm3 * 62 )
          + ( ( SO4         )       * ppb_ugm3 * 96 )
          + ( ( BCPI + BCPO )       * ppb_ugm3 * 12 )
          + ( ( OCPI + OCPO ) * 2.1 * ppb_ugm3 * 12 )
          + ( ( DST1 + DST2 )       * ppb_ugm3 * 29 )

Where NH4, NIT, SO4, BCPI, BCPO, OCPI, OCPO, NO3, DST1, and DST2 have units of ppbv.

We use the value 2.1 for OM:OC, but note from Jeff Pierce's comment in the section above that this is an open question.

Colette Heald wrote:

For completeness PM2.5 should include accumulation mode sea salt (SSa), but that's not included in the PM2.5 calculations used in the benchmarks, I think just for historical reasons since the code was written for continental US sites. It may be worth adding this comment to the wiki for folks who are using this as a guideline for calculating PM2.5.

--Bob Y. 17:11, 13 February 2015 (EST)

Anthropogenic PM2.5 dust source in GEOS-Chem

NOTE: This update will be implemented in v11-03.

Sajeev Philip and coauthors have added a new PM2.5 dust emission inventory into GEOS-Chem, termed as Anthropogenic Fugitive, Combustion and Industrial Dust (AFCID). For information on this inventory, please see our Mineral dust aerosols wiki page.

--Melissa Sulprizio (talk) 13:51, 20 March 2017 (UTC)

Past discussions

Xinyi Dong wrote:

I am trying to get PM2.5 from GEOS-Chem and have a question regarding how to sum the aerosol species to get PM2.5. In one paper Tai et al., 2012, the author said "total PM2.5 in GEOS-Chem is taken to be the sum of sulfate, nitrate, ammonium, OC and EC". While in another paper Liu et al., 2004, JGR, the author also include fine dust and sea salt (although he was examing AOT, not exactly PM2.5, but the fine dust aerosols have diameter less than 2.5um). I could not find an official equation for PM2.5 from GEOS-Chem website, so what aerosol species I need to sum in order to get PM2.5 ?
Here is the titles of the two papers:
  1. Tai et al, 2012: Meteorological modes of variability for fine particulate matter (PM2.5) air quality in the United States: implications for PM2.5 sensitivity to climate change PDF
  2. Liu et al., 2004: Mapping annual mean ground-level PM2.5 concentration using multiangle imaging spectroradiometer aerosol optical thickness over the contiguous United States.

Jeff Pierce replied:

Sulfate + Nitrate + Ammonium + OC + EC + SOA + FineSeaSalt + FineDust is probably the best bet for PM2.5 (don't forget to convert OC to OM). However, there is no perfect way in GEOS-Chem since size distributions vary and there will be some fraction of all species that are larger than 2.5 microns.

Xinyi Dong wrote:

Now I see why [the two papers] use different equations; my understanding is one paper is compared GEOS-Chem with surface observations having few impacts from sea salt and dust, while another compared with satellite AOT, so sea salt and dust were included.
I did a quick search for the conversion from OC to OM, one paper say the OM/OC ratio is 2.1, another say 1.4, so is there an official value suggested for v9-01-02?

Jeff Pierce replied:

OM:OC is an open question! 2.1 is generally more representative of aged organics, 1.4 would be fresh. I think many people use 1.8 if they choose a single value.

--Bob Y. 16:39, 7 February 2013 (EST)