1. GEOS-Chem model overview Daniel J. Jacob, Harvard University
GEOS-Chem Community Mission: to advance understanding of human and natural influences on the environment through a comprehensive, state-of-the-science, readily accessible global model of atmospheric composition
GEOS-Chem user groups
IGC8 meeting (Harvard, May 2017)

2. Atmospheric chemistry: many problems, many scales
Ozone
layer
Disasters
We need models to
Understand processes
Confront our understanding to observations
Interpret observations to generate new knowledge
Make projections
Link atmospheric chemistry to other Earth system processes
Visibility
Urban smog
Regional smog
Climate forcing
Plume dispersion
Biogeochemical cycles
Acid rain
LOCAL
< 100 km
REGIONAL km
GLOBAL
> 1000 km

3. GEOS-Chem atmospheric chemistry model
Input meteorological data
NASA GEOS fields, 1980-present
0.25o×0.3125o or 0.5ox0.625o, 72 vertical levels
GEOS-Chem solves 3-D chemical continuity equations
on global or nested Eulerian grid
Modules
emissions
transport
chemistry
aerosols
deposition
Applications
Model adjoint
Tropospheric and stratospheric chemistry, aerosol microphysics, carbon gases, mercury including biogeochemical coupling, POPs, isotopes…
New version releases every year: v11-2 (May 2018) to include updated SOA, isoprene chemistry, tropospheric halogen (Cl-Br-I) chemistry, methane, emissions…
Grid-independent, ESMF-compatible architecture for coupling to Earth System models

4. GEOS-Chem as on-line chemical module for Earth system models (ESMs)
any 3-D grid specified at run time O
Chemistry
(FlexChem):
dC/dt = P –L - D
Emissions
(HEMCO):
dC/dt = E
ESMF
GEOS-Chem chemical module
ESMF
ESMF
Advection
Mixing
Convection
Off-line
GEOS-Chem CTM

5. GEOS-Chem as on-line chemical module for Earth system models (ESMs)
any 3-D grid specified at run time O
Chemistry
(FlexChem):
dC/dt = P –L - D
Emissions
(HEMCO):
dC/dt = E
ESMF
GEOS-Chem chemical module
ESMF
Dynamics,
chemical transport
Earth System Model with
GEOS-Chem chemistry

6. CTM and ESM use exactly the same GEOS-Chem code
GEOS-Chem as on-line chemical module for Earth system models (ESMs)
any 3-D grid specified at run time O
Chemistry
(FlexChem):
dC/dt = P –L - D
Emissions
(HEMCO):
dC/dt = E
ESMF
GEOS-Chem chemical module
ESMF
ESMF
ESMF
Advection
Mixing
Convection
Dynamics,
chemical transport
Off-line
GEOS-Chem CTM
Earth System Model with
GEOS-Chem chemistry
CTM and ESM use exactly the same GEOS-Chem code
GEOS-Chem CTM users contribute model advances
Advances are incorporated
into GEOS-Chem
ESM GEOS-Chem module always stays referenced
to latest version
Keller et al.. [2014]; Long et al. [2015]

7. Full-year tropospheric chemistry simulation at c720 (~12 km) resolution using on-line GEOS-Chem in GEOS ESM
August 1, Z
Hu et al. [2018]

8. GEOS-Chem high performance (GCHP):
Massively parallel capability with cubed-sphere advection
uses distributed-memory MPI parallelization enabled by ESMF
1-month simulation of troposphere-stratospheric chemistry at c180 (~50 km) resolution
requires only 1 day of wall time with 540 cores
Ozone at 4 km altitude, July 2016
Number of cores
Wall time
perfect scalability
Aerosol optical depth (AOD), July 2016
Eastham et al. [2018]

9. Running GEOS-Chem on the cloud
GEOS-Chem is now fully operational on the AWS cloud; GCHP is on the way
Many advantages for access, reproducibility, sharing of model versions and data
Jiawei Zhuang, Harvard

10. GEOS-Chem development is driven by its user community
GEOS-Chem Steering Committee
Model scientist: Jacob (Harvard)
co-scientist: Martin (Dalhousie)
Adjoint scientist: Henze (U. Colorado)
Nested model scientists: Wang (Tsinghua/UH), Zhang (PKU)
Engineer: Yantosca (Harvard)
Aerosols WG:
Heald (MIT), Alexander (UW), Pierce (CSU), Yu (SUNYA)
Chemistry WG:
Henderson (EPA), Evans (York), Mao (U. Alaska), Hu (U. Montana)
Emissions and Deposition WG:
Lin (PKU), Fischer (CSU), Millet (U. Minnesota)
Chemistry-Ecosystems-Climate WG:
Liao (NIUST), Tai (CUHK), Murray (U. Rochester), Geddes (Boston U)
Carbon WG:
Jones (U. Toronto), Bowman (JPL)
Adjoint and Data Assimilation WG:
Wang (U. Iowa), Henze (U. Colorado)
Transport WG:
Liu (NIA), Molod (NASA)
Hg and POPs WG:
Holmes (FSU), Fisher (U. Wollongong)
GCHP WG: Martin (Dalhousie), Eastham (MIT)
GMAO Liaison: Keller (NASA)
At large: Kasibhatla (Duke).)
Working Groups
set priorities
contribute
developments, report bugs
support
implement priorities
Model Support Team
(Harvard, Dalhousie, U. Colorado)
benchmark updates
Model version updates
Emphasize grass-roots development,
fast innovation, traceability, user support

11. GEOS-Chem v11-02 to be released this week
V11-02a: chemical updates including PAN chemistry (US EPA, U. York, CSU)
V11-02b: GCHP capability (GEOS-Chem Support Team)
V11-02c: isoprene chemistry and SOA updates (U. Wollongong, U. Birmingham, Caltech, CSU, Harvard)
V11-02d: comprehensive halogen chemistry (U. York, U. Copenhagen, MIT, Harvard)
V11-02e: stratospheric chemistry updates, full NetCDF diagnostics (MIT, FSU)
V11-02f: many new emission inventories, new methane simulation, new Hg chemistry (Dalhousie, Harvard, UW, U. Birmingham…)

12. Building GEOS-Chem v11-3: expected release in May 2019
Search “GEOS-Chem development priorities” to get the list of potential updates
Weigh in through your WG chairs on the items you want!
Some major items:
Grid-independent natural emissions
Enabling 12-km resolution in both GCHP and GC-Classic
Rediagnosed convection to improve transport in coarse-resolution simulations
Updated tropospheric isoprene and halogen chemistry
Mesospheric chemistry
Many emission updates including MEIC
Enable nested simulations over any custom domains
CESM coupling off-line and on-line
Harmonized biosphere-atmosphere exchange, coupling to CLM

13. Chemistry challenges
How do we get to a unified mechanistic mechanism for SOA?
Simple seems to do best, but are we satisfied with that?
Full tropospheric-stratospheric Cl-Br-I chemistry
Major tropospheric ozone sink at mid-latitudes from seasalt aerosol – how do we fix?
Updating isoprene chemistry
New “reduced” Caltech mechanism has 150 species…
Aromatic chemistry and terpene chemistry are still sketchy
Speeding up the model
Mechanism reduction?
Machine learning?
Crowd sourcing through the cloud?

14. Need to facilitate update of “specialty” simulations
Progress in emissions
HEMCO has made updating emission inventories much easier
but beware of black box – need to check what comes out
Grid-independent natural emissions will be major advance
Harmonized biosphere-atmosphere exchange, coupling to CLM
Continual challenge to update emissions to latest years
Community participation is key
Need to facilitate update of “specialty” simulations
Hg chemistry moving to KPP
Need to maintain aerosol microphysics simulations

15. Push to higher resolution
AQ applications look for higher resolution,
also evidence that coarse model resolutions produce transport errors
Two approaches:
Off-line simulations matched to highest GEOS resolution (now 12 km)
Advantages:
Ease of use
Cost
Reproducibility
Inverse applications
Disadvantages:
Loss of information in temporal archiving of met data
Management of large archive
Cannot go finer than GEOS resolution
On-line simulations with hi-res met model using GEOS-Chem chemical module
Specific on-line applications: chem-climate, data assimilation
Explore different meteorological simulations, high resolution
Disadvantages: see off-line advantages
Some other considerations:
As resolution increases, on-line becomes more desirable
Many of our applications don’t need very high resolution
….so a model with both on-line and off-line capabilities makes sense
Have the exact same GEOS-Chem chemical module in all applications

16. Off-line approaches: GC-Classic and GCHP
GC-Classic is to be maintained indefinitely (or until it has outlived its usefulness)
Ease of access and use
Nested model capability can be extended to higher resolutions
GCHP: a powerful tool for the future
High-performance applications through MPI
Better transport through FV3 on cubed sphere
Challenge #1: difficulty of use
Solve with access through cloud?
Simplify the MAPL coupler?
Challenge #2: Availability of cubed-sphere native archive
Prototype archive being tested: air mass fluxes and pressure
Challenge #3: dealing with model nesting
Use stretched-grid approach?

17. On-line approach: GEOS-Chem chemical module
Coupling to GEOS-5 and to BCC is already in place
But these models are not open access…
Coupling to WRF – very exciting for AQ applications!
a much better, more transparent chemical module than current WRF-Chem
Coupling to NCAR CESM – development being led by Seb Eastham
Prototype version to be presented to NCAR next week
Strong buy-in from NCAR – they want GEOS-Chem

18. Looking to the future: balancing community needs for innovation and stability
GEOS-Chem Community Mission: to advance understanding of human and natural influences on the environment through a comprehensive, state-of-the-science, readily accessible global model of atmospheric composition
GEOS-Chem is unique among atmchem models in its ease of use; allows atm chemists with low computing resources to contribute scientific innovation
But they wouldn’t use it if it didn’t stay at the cutting edge of science – meaning that we have to continuously innovate in both the science and the underlying software engineering
But somebody’s “cutting edge of science” is someone else’s “unnecessary update” and users are stressed when the model keeps changing
This is particularly the case for specialty simulations that get broken in model updates
On the other hand, continued support for older versions would unduly stress the Support Team
Best solution is for GEOS-Chem Steering Committee to provide guidance in striking a balance between innovation and stability

19. See you at IGC9! May 6-9 2019, Harvard University
IGC8 meeting (Harvard, May 2017)