GEOS-Chem benchmarking

Objectives

Benchmarking supports the maintenance of GEOS-Chem as a robust state-of-the-science facility with a nimble grass-roots approach and strong version control. Benchmarking has four main objectives:

1. Document a consistent GEOS-Chem model configuration, and the expected characteristics of that configuration.
2. Support version control through traceability, and by confirming the expected behavior of model developments submitted by the community.
3. Track the evolution of the model over the years.
4. Promote scientific transparency of GEOS-Chem.

Procedure

The GEOS-Chem benchmarking procedure is described below.

1. Any update to the GEOS-Chem source code or run directories will change the GEOS-Chem version number (X.Y.Z).
2. All Y versions require a dedicated 1-month benchmark with the standard full-chemistry mechanism. The benchmark results will be posted on the wiki and an email will be sent to the developer(s) and the GEOS-Chem Steering Committee.
3. The developer(s) will assess the benchmark results and review a benchmark assessment form on the wiki. If the developer has any concerns about the benchmark results they will contact the GEOS-Chem Support Team, the relevant GEOS-Chem Working Group Chairs, and/or Model Scientist Daniel Jacob. GEOS-Chem Steering Committee members are also encouraged to examine and comment on the benchmark results.
4. If the update is for a specialty simulation (e.g. CO2, CH4, Hg), then a further benchmark may be conducted by the appropriate Working Group.
5. Once the developer is satisfied with the changes in the 1-month benchmark, GEOS-Chem Model Scientist Daniel Jacob will review the results and approve the new internal version.
6. 1-year benchmarks for Y versions will be conducted only if justifiably requested by the developer or by GEOS-Chem Steering Committee members.
7. Each new major version release (i.e. X version) will be subject to a 1-year benchmark to be inspected by the GEOS-Chem Steering Committee before approval.

List of GEOS-Chem benchmarks

Links to past 1-month and 1-year benchmark simulations can be found on the GEOS-Chem versions wiki page. For information about 1-year benchmark simulations for GEOS-Chem v8-02-04 and earlier, see this website.

Benchmark output archive

Output files and evaluation plots for 1-month and 1-year benchmark simulations are archived at Harvard as summarized below. GEOS-Chem users may utilize these output for comparisons against their own simulations.

Directory	Description
http://ftp.as.harvard.edu/gcgrid/geos-chem/1mo_benchmarks/	Contains the following data from the 1-month benchmarks used to evaluate GEOS-Chem: Restart files Model output Log files Input files Evaluation plots
http://ftp.as.harvard.edu/gcgrid/geos-chem/1yr_benchmarks/	Contains the following data from the 1-year benchmarks used to evaluate GEOS-Chem: Restart files Model output Log files Input files Evaluation plots

1-month benchmark

Overview

Dates of run:	July 1, 2013 – August 1, 2013 (prior to v11-02e) July 1, 2016 – August 1, 2016 (v11-02e and later versions)
Vertical Resolution:	GEOS-FP with 72 hybrid sigma-pressure levels
Horizontal Resolution:	4° latitude x 5° longitude
Advected species:	Gas-phase species: NO, O3, PAN, CO, ALK4, ISOP, HNO3, H2O2, ACET, MEK, ALD2, RCHO, MVK, MACR, PMN (isoprene and non-isoprene), PPN, R4N2, PRPE, C3H8, CH2O, C2H6, N2O5, HNO4, MP, MPN, ISOPND, ISOPNB, MOBA, PROPNN, HAC, GLYC, MVKN, MACRN, MAP, NO2, NO3, HNO2, BENZ, TOLU, XYLE, MTPA, LIMO, MTPO, TSOG0-3, ISOG1-3, ASOG1-3, EOH, MGLY, GLYX, ACTA, HPALD, DHDN, ETHLN, HCOOH, IEPOXA, IEPOXB, IEPOXD, ISN1, RIPA, RIPB, RIPD, IMAE, LVOC, ISN1OG, MONITS, MONITU, HONIT, HC187 Aerosol species: DMS, SO2, SO4, MSA, NH3, NH4, NIT, BCPO, OCPI, BCPI, OCPI, DST1-4, SALA, SALC, SO4s, NITs, TSOA0-3, ISOA1-3, ASOAN, ASOA1-3, SOAIE, SOAME, SOAGX, SOAMG, LVOCOA, ISN1OA, IONITA, MONITA, INDIOL Bromine species: Br2, Br, BrO, HOBr, HBr, BrNO2 BrNO3, CHBr3, CH2Br2, CH3Br Stratospherically-important species: N2O, OCS, CH4, BrCl, HCl, CCl4, CH3Cl, CH3CCl3, CFCX, HCFCX, CFC11, CFC12, HCFC22, H1211, H2402, Cl, ClO, HOCl, ClNO3, ClNO2, ClOO, OClO, Cl2, Cl202, H2O
Chemical Mechanism:	FlexChem chemical solver using the standard chemistry mechanism (NOx-Ox-HC-Aer-Br + UCX + complex SOA) Chemistry is done in both the troposphere and the stratosphere.
Photolysis Mechanism:	FAST-JX v7.0 (Prather, 2012, Wild et al, 2000).
Operations:	Advection Aerosol thermodynamical equilibriium Boundary layer mixing Chemistry Cloud convection (with updraft scavenging) Dry deposition Emissions Photolysis Wet deposition
Diagnostic Output:	The following diagnostic outputs are archived as monthly means Sulfate production and loss quantities Dust aerosol sources Carbon aerosol sources Sea salt aerosol sources Acetone sources Sulfur sources Optical Depths Noontime J-values Biomass burning emission fluxes CO sources Surface Pressure NOx sources Biofuel emission fluxes Anthropogenic emission fluxes Scavenging loss from moist convection Scavenging loss from wet deposition Concentrations of chemically produced OH and HO2 Dry deposition fluxes and velocities Tracer concentrations Biogenic emission fluxes Tropopause height DAO 3-D met fields DAO 2-D met fields Air masses and grid box heights Surface area Lifetime of Methylchloroform (CH3CCl3)
Benchmark Plots and Summaries:	The following plots, budgets, and totals are created from the oxidant-aerosol simulation: Budget of Ox and CO Mean OH concentration Methyl Chloroform Lifetime (w/r/t loss by tropospheric OH) Aerosol optical depth maps (this benchmark) Aerosol optical depth differences (this benchmark - previous benchmark) Concentration maps (this benchmark) of tracers + OH + HO2 at the surface and at 500 hPa Difference maps (this benchmark - previous benchmark) of tracers + OH + HO2 at the surface and at 500 hPa Summary of emission totals (this benchmark vs. previous benchmark) Emissions maps (this benchmark) for all emitting tracers Emissions difference maps (this benchmark - previous benchmark) for all emitting tracers Emissions ratio maps (this benchmark / previous benchmark) for all emitting tracers Frequency distribution histogram of ratios (this benchmark vs previous benchmark) for tracers, OH, and optical depths J-value maps (this benchmark) J-value difference maps (this benchmark - previous benchmark) J-value ratio maps (this benchmark / previous benchmark) Ratio maps (this benchmark / previous benchmark) of tracers + OH + HO2 at the surface and at 500 hPa Zonal mean concentration maps (this benchmark) of tracers + OH + HO2 Zonal mean difference maps (this benchmark - previous benchmark) of tracers + OH + HO2

Plotting routines

The 1-month benchmark plotting routines are included with GCPy, a Python took kit available for GEOS-Chem.

1-year benchmark

Overview

Spin-up:	January 1, 2013 – January 1, 2014 (prior to GEOS-Chem v11-02e) January 1, 2015 – January 1, 2016 (since GEOS-Chem v11-02e)
Dates of Run:	January 1, 2013 – January 1, 2014 (prior to GEOS-Chem v11-02e) January 1, 2016 – January 1, 2017 (since GEOS-Chem v11-02e)
Vertical Resolution:	GEOS-FP with 72 hybrid sigma-pressure levels
Horizontal Resolution:	4° latitude x 5° longitude
Advected species:	Gas-phase species: NO, O3, PAN, CO, ALK4, ISOP, HNO3, H2O2, ACET, MEK, ALD2, RCHO, MVK, MACR, PMN (isoprene and non-isoprene), PPN, R4N2, PRPE, C3H8, CH2O, C2H6, N2O5, HNO4, MP, MPN, ISOPND, ISOPNB, MOBA, PROPNN, HAC, GLYC, MVKN, MACRN, MAP, NO2, NO3, HNO2, BENZ, TOLU, XYLE, MTPA, LIMO, MTPO, TSOG0-3, ISOG1-3, ASOG1-3, EOH, MGLY, GLYX, ACTA, HPALD, DHDN, ETHLN, HCOOH, IEPOXA, IEPOXB, IEPOXD, ISN1, RIPA, RIPB, RIPD, IMAE, LVOC, ISN1OG, MONITS, MONITU, HONIT, HC187 Aerosol species: DMS, SO2, SO4, MSA, NH3, NH4, NIT, BCPO, OCPI, BCPI, OCPI, DST1-4, SALA, SALC, SO4s, NITs, TSOA0-3, ISOA1-3, ASOAN, ASOA1-3, SOAIE, SOAME, SOAGX, SOAMG, LVOCOA, ISN1OA, IONITA, MONITA, INDIOL Bromine species: Br2, Br, BrO, HOBr, HBr, BrNO2 BrNO3, CHBr3, CH2Br2, CH3Br Stratospherically-important species: N2O, OCS, CH4, BrCl, HCl, CCl4, CH3Cl, CH3CCl3, CFCX, HCFCX, CFC11, CFC12, HCFC22, H1211, H2402, Cl, ClO, HOCl, ClNO3, ClNO2, ClOO, OClO, Cl2, Cl202, H2O
Chemical Mechanism:	FlexChem chemical solver using the standard chemistry mechanism (NOx-Ox-HC-Aer-Br + UCX + complex SOA) Chemistry is done in both the troposphere and the stratosphere.
Photolysis Mechanism:	FAST-JX v7.0 (Prather, 2012, Wild et al, 2000).
Operations:	Advection Aerosol thermodynamical equilibriium Boundary layer mixing Chemistry Cloud convection (with updraft scavenging) Dry deposition Emissions Photolysis Wet deposition
Diagnostic Output:	All diagnostics above are saved as monthly means. Sulfate production and loss quantities Dust aerosol sources Carbon aerosol sources Sea salt aerosol sources Acetone sources Sulfur sources Optical Depths Noontime J-values Transport fluxes (N/S, E/W, up/down) Biomass burning emission fluxes CO sources Surface Pressure NOx sources Biofuel emission fluxes Anthropogenic emission fluxes Scavenging loss from moist convection Scavenging loss from wet deposition Concentrations of chemically produced OH and HO2 Dry deposition fluxes and velocities Tracer concentrations Biogenic emission fluxes Tropopause height DAO 3-D met fields DAO 2-D met fields Air masses and grid box heights Surface area Lifetime of Methylchloroform (CH3CCl3) Satellite timeseries of BrO, cloud fraction, and cloud top height
Benchmark Plots:	The following plots compare advected species from 3 different benchmark simulations (versions 1-3, where version 3 is the current benchmark version). Plots are created for January, April, July, and October. Aerosol optical depth maps (version 1, version 2, and version 3) Emission maps (version 1, version 2, and version 3) Emisison difference maps (version 3 - version 1, version 3 - version 2) Emission ratio maps (version 3 / version 1, version 3 / version 2) J-value maps (version 1, version 2, and version 3) J-value difference maps (version 3 - version 1, version 3 - version 2) J-value ratio maps (version 3 / version 1, version 3 / version 2) Tracer difference maps at the surface and 500 hPa (version 3 - version 1, version 3 - version 2) Tracer ratio maps at the surface and 500 hPa (version 3 / version 1, version 3 / version 2) Tracer difference profiles along longitude slices (15S, 42N) (version 3 - version 1, version 3 - version 2) The following plots show data from 3 benchmark simulations side-by-side with observations. GEOS-Chem output is plotted in red for version 1, in green for version 2, and in blue for version 3 (current benchmark version) BrO seasonal columns vs. observations C2H6 vertical profiles vs. observations from various aircraft campaigns C3H8 vertical profiles vs. observations from various aircraft campaigns CO seasonal cycle at the surface vs. MOZAIC observations CO vertical profiles vs. MOZAIC observations CO vertical profiles vs. various sonde observations H2O2 vertical profiles vs. observations from various aircraft campaigns IMPROVE difference maps (benchmark simulation output vs. surface sites) IMPROVE scatter plots (benchmark simulation output vs. surface sites) HNO3 vertical profiles vs. observations from various aircraft campaigns NO vertical profiles vs. observations from various aircraft campaigns O3 seasonal cycle @ the surface vs. MOZAIC observations O3 seasonal cycle @ 300, 500, and 800 hPa vs. MOZAIC observations O3 seasonal cycle @ 300, 500, and 800 hPa vs. sonde observations O3 seasonal cycle @ 150, 300, 500,and 800 hPa vs. sonde observations O3 vertical profiles vs. observations from various aircraft campaigns O3 vertical profiles vs. observations from various aircraft campaigns O3 vertical profiles vs. various sonde observations O3 vertical profiles vs. MOZAIC observations PAN vertical profiles vs. observations from various aircraft campaigns PM2.5 difference maps (benchmark simulation output vs. surface sites) PM2.5 scatter plots (benchmark simulation output vs. surface sites)

Plotting routines

NOTE: The 1-month benchmark plotting routines are currently written in IDL. We are planning on moving these routines to Python for the GEOS-Chem v11-02 release as part of the GCPy package.

The 1-year benchmark plotting routines can be downloaded from Bitbucket via Git using:

 git clone https://bitbucket.org/gcst/gc_1yr_benchmark

The data files needed for the plots are included in the repository and are discussed in more detail below.

Observations

Here we provide an overview of the data used in the 1-year benchmark plots.

Directory	Data year	Species	Data source	Provided by
BrO/	2007-2008	BrO	GOME-2	Justin Parrella (Data used to reproduce Figure 5 of Parrella et al. (2012))
cmdl/	2005, 2009 New data available	CO	GMD (formerly CMDL)	Jennifer Logan and Inna Megretskaia
co.prof.for.gmi/	2001-2008 (varies by station)	CO	MOZAIC	Jennifer Logan and Inna Megretskaia
eval/aircraft/ NOTE: Other directories in eval/ aren't currently used	1995-2003 (varies by aircraft campaign)	C2H6, C3H8, CO, H2O2, HNO3, NO, O3, PAN (Not currently used: ACET, ALK4, C2H2, CH2Br2, CH2BrCl, CH3Br, CH3OH, CH4, CHBr2Cl, CHBrCl2, CO2, ETHE, PRPE)	See README	Jennifer Logan and Inna Megretskaia
netCDF/	N/A	C2H6, C3H8, CH3I, CO, H2O2, HNO3, NO, O3, OH, PAN, Rn	Various (contains information about station locations)	Jennifer Logan and Inna Megretskaia
PAN/	Varies	PAN	ARCTAS-A, ARCTAS-B, ARCPAC, INTEX-B, ITCT-2K2	Emily Fischer (Data used to reproduce Figure S1 of Fischer et al. (2014) Supplement)
pm25_data/	2005	PM2.5	IMPROVE	Colette Heald
sondes.for.gmi/	1990-2008 (varies by station)	O3		Jennifer Logan and Inna Megretskaia
strat/	2001-2010	NOy, O3	OSIRIS	Dylan Jones
surface_ozone/		O3	CMDL	Jennifer Logan and Inna Megretskaia

--Melissa Sulprizio (talk) 14:26, 16 August 2017 (UTC)

Update 2005/2009 CMDL CO data to 2013/2014 GMD CO data

This update was included in v11-02c and approved on 21 Sep 2017.

Jenny Fisher wrote:

Attached is a zip file containing routines to process the GMD (previously CMDL) CO data for use in the benchmarking scripts, as well as the 2013 and 2014 data. I added the 2014 data because there was no ships data for 2013 (or 2012), so this seemed like the best approximation.

To replace the existing benchmark data with these, you will want to do the following in the 1-year benchmark code:

1. Put the 2013data/ and 2014data/ in the data/cmdl/ directory of the 1-year benchmark code
2. Change newdata/ to 2013data/ in plot_cmdl_3_models_4_months.pro
3. Change Feb09/ to 2013data/ in plot_surface_co_geos_3_models.pro
4. Change newdata/ to 2014data/ in plot_ships_3_models_4_months.pro and plot_ships_3_models_co.pro
5. Change the line file=pre+name_sta(kk)+'.mn' to file=pre+name_sta(kk)+’.mn.2013' in plot_cmdl_3_models_4_months.pro
6. Change the line file=pre+name_sta(kk)+’.mn.2005' to file=pre+name_sta(kk)+’.mn.2013' in plot_surface_co_geos_3_models.pro
7. Change the line file=pre+name_sta(kk)+'.mn' to file=pre+name_sta(kk)+’.mn.2014' in plot_ships_3_models_4_months.pro and plot_ships_3_models_co.pro

As I said before, I think this would be a useful improvement to our benchmarks, rather than using data >10 years old!

--Melissa Sulprizio (talk) 22:10, 15 August 2017 (UTC)