Difference between revisions of "GEOS-Chem v10-01 benchmark history"

From Geos-chem
Jump to: navigation, search
(→‎1-month benchmark with UCX chemistry)
(→‎1-month benchmark with tropospheric chemistry)
Line 164: Line 164:
 
==== 1-month benchmark with tropospheric chemistry ====
 
==== 1-month benchmark with tropospheric chemistry ====
  
−
Here is the assessment form for 1-month benchmark simulation '''v10-01c with tropospheric chemistry''' (i.e. UCX off), aka '''v10-01c_trop'''.
+
Here is the assessment form for 1-month benchmark simulation '''v10-01c with tropospheric chemistry''' (i.e. UCX off, aka '''v10-01c_trop''').
  
 
{| border=1 cellspacing=0 cellpadding=5
 
{| border=1 cellspacing=0 cellpadding=5

Revision as of 17:22, 30 May 2014

On this page we have posted complete information about all benchmark simulations (both 1-month and 1-year) for GEOS-Chem v10-01.

1-yr benchmarks

TBD

1-month benchmarks

v10-01c

Because GEOS-Chem v10-01c was the version where we introduced the UCX chemistry mechanism, we performed two 1-month benchmark simulations:

  1. v10-01c_trop: 1-month benchmark with tropopsheric chemistry
  2. v10-01c_UCX: 1-month benchmark with UCX strat-trop chemistry (72 vertical levels)

--Bob Y. 13:15, 30 May 2014 (EDT)

Final recommendation for J-values

We were not able to include this final recommendation for J_values into the v10-01c_trop and v10-01c_UCX 1-month benchmarks. We will include these in the 1-year benchmarks for v10-01c.

Jingqiu Mao wrote:

I have two more suggestions to the code and I think we then can finalize v10-01c. We can deal with unresolved J(VOC) later. Seb, please let me know if you think otherwise.
  1. For HAC, keep the QY as 0.6, but zero out the bins >335 nm.
  2. For PAN, change the reaction from 
       PAN = 0.6MCO3 + 0.6NO2 + 0.4MO2 + 0.4NO3
to 
       PAN = 0.7MCO3 + 0.7NO2 + 0.3MO2 + 0.3NO3

Daniel Jacob replied:

Thanks Jingqiu! If the 1-year benchmark run has already started just let it run - these changes will have very little effect except for HAC and we can just make a note of it. I'm glad that we resolved these J(VOC) issues thanks to Seb, Chris and Jingqiu. At this point we need to move on.

--Bob Y. 13:21, 30 May 2014 (EDT)

1-month benchmark with UCX chemistry

Here is the assessment form for 1-month benchmark simulation v10-01c with UCX chemistry (aka v10-01c_UCX).

Description
New features added into GEOS-Chem:
Developer name(s) and institution(s):
Version, resolution, met fields used: v10-01, GEOS-FP (72L), 4x5, July 2013
1-month benchmark finished on: Thu May 29 00:01:33 2014
Performance statistics:
  • Ran on 8 CPUs of bench@titan-09.as.harvard.edu (2.659 GHz x 8 CPU)
  • Wall time: 5:51
  • Scalability: 6.7089
Compared to previous benchmark: v10-01c with tropospheric chemistry
This update will impact:
(select all that apply with boldface) 		Advection, BL Mixing, Convection, Met Fields, Dry Dep, Wet Dep, Stratosphere, Anthro Emiss, Biogenic Emiss, Biomass Emiss, Photolysis, Chemistry, Other (please specify):
Unit test results may be viewed at: ftp://ftp.as.harvard.edu/gcgrid/geos-chem/1mo_benchmarks/v10-01/v10-01c/v10-01c_UCX/v10-01c_Final.results.html
Plots may be viewed at: ftp://ftp.as.harvard.edu/gcgrid/geos-chem/1mo_benchmarks/v10-01/v10-01c/v10-01c_UCX/
Metrics
Global mean OH (from log file): 13.1136815995777 x 105 molec/cm3
Methyl chloroform lifetime: 4.7459 years
Did either of these change by more than 5%? No. The mean OH differs by -3.77%, and the MCF lifetimes differ by 0.23%.
At the SURFACE, list all species that changed by 10% or more: NO, PAN, ALK4, ISOP, HNO3, H2O2, MEK, RCHO, MVK, MACR, PMN, PPN, PRPE, N2O5, HNO4, MP, DMS, SO4s, NH3, NIT, NITs, DST4, Br2, Br, BrO, HBr, BrNO2, BrNO3, MPN, ISOPN, MOBA, HAC, GLYC, MMN, RIP, IEPOX, NO2, NO3, HNO2, OH, HO2
Comments on SURFACE differences:
  • Differences in the ratio plots of ISOP and its oxidation products over the oceans probably reflect small number differences.
At 500 hPa, list all species that changed by 10% or more: NO, ALK4, ISOP, HNO3, MEK, RCHO, MVK, MACR, PMN, PRPE, C3H8, N2O5, HNO4, MP, DMS, SO2, NH3, NIT, NITs, DST4, Br2, Br, BrO, HBr, BrNO2, BrNO3, MPN, ISOPN, MOBA, HAC, GLYC, MMN, RIP, IEPOX, NO2, NO3, HNO2, OH, HO2
Comments on 500 hPa differences:
  • The "noise" pattern in the NIT and NH3 plots reflect the numerical drift caused by ISORROPIA II
In the ZONAL MEAN differences, list all species that changed by 10% or more: All species, except for MSA and OCPI, changed by 10% or more.
Comments on ZONAL MEAN differences:
  • It is understandable that we would see differences of this magnitude in these species given that we have activated stratospheric chemistry. We are also using the full 72-layer grid as opposed to the 47-layer grid in the tropopause-only simulation.
  • The following species show small absolute differences but large percent differences:
    • ALK4, ISOP, MEK, ALD2, RCHO, MVK, MACR, PMN, R4N2, PRPE, C3H8, CH2O, C2H6, SO2, SO4s, MSA, NH3, NH4, NITs, BCPO, OCPO, DST1, DST2, DST3, DST4, SALA, SALC, BrNO2, CHBr3, CH2Br2, ISOPN, HAC, GLYC, MMN, HNO2
In the EMISSION RATIO maps, list all species that changed by 10% or more: None
Comments on EMISSION RATIO differences:
  • NOx sources (anthropogenic, aircraft, lightning, soils) changed very slightly in this run. This may be an artifact of comparing the data on the full 72-layer grid to the 47-layer grid of the previous benchmark.
  • The acetone ocean sink also decreased slightly (-0.000638 Tg C). This may also be an artifact of comparing the 72-level grid to the 47-level grid.
Additional or summary comments:
  • This simulation uses the KPP chemistry solver, while all past benchmarks have used SMVGEAR. UCX simulations are extremely slow when SMVGEAR is used, so KPP is the recommended solver.
  • Many of the structural updates (i.e. to speed up GEOS-Chem, to reduce the memory footprint, to fix issues in the specialty simulations) do not impact the full-chemistry simulation. This was validated with unit tests and difference tests.
Approval
Requires further investigation: NO
Approved by: Daniel Jacob, Jingqiu Mao, Sebastian Eastham
Date of approval: 29 May 2014

--Bob Y. 13:11, 30 May 2014 (EDT)

1-month benchmark with tropospheric chemistry

Here is the assessment form for 1-month benchmark simulation v10-01c with tropospheric chemistry (i.e. UCX off, aka v10-01c_trop).

Description
New features added into GEOS-Chem:
Developer name(s) and institution(s):
Version, resolution, met fields used: v10-01, GEOS-FP (47L), 4x5, July 2013
1-month benchmark finished on: Wed May 28 17:57:45 2014
Performance statistics:
  • Ran on 8 CPUs of bench@titan-09.as.harvard.edu (2.659 GHz x 8 CPU)
  • Wall time: 2:49
  • Scalability: 6.7835
Compared to previous benchmark: v10-01b
This update will impact:
(select all that apply with boldface) 		Advection, BL Mixing, Convection, Met Fields, Dry Dep, Wet Dep, Stratosphere, Anthro Emiss, Biogenic Emiss, Biomass Emiss, Photolysis, Chemistry, Other (please specify):
Unit test results may be viewed at: ftp://ftp.as.harvard.edu/gcgrid/geos-chem/1mo_benchmarks/v10-01/v10-01c/v10-01c_trop/v10-01c_Final.results.html
Plots may be viewed at: ftp://ftp.as.harvard.edu/gcgrid/geos-chem/1mo_benchmarks/v10-01/v10-01c/v10-01c_trop
Metrics
Global mean OH (from log file): 13.6273583437531 x 105 molec/cm3
Methyl chloroform lifetime: 4.7349 years
Did either of these change by more than 5%? No. The mean OH differs by -0.77%, and the MCF lifetimes differ by 1.37%.
At the SURFACE, list all species that changed by 10% or more: NO, PAN, ALK4, ISOP, H2O2, MEK, RCHO, MVK, MACR, PMN, PPN, PRPE, CH2O, N2O5, HNO4, MP, DMS, NIT, DST4, Br2, Br, BrO, HOBr, HBr, BrNO2, BrNO3, MPN, ISOPN, MOBA, PROPNN, HAC, GLYC, MMN, RIP, IEPOX, MAP, NO3, HNO2, OH, HO2
Comments on surface differences:
  • Most of the differences in ISOP and its products are over the oceans. This could be due to small number differences, since there is not a lot of ISOP in these locations.
  • We believe that the following differences are due to photolysis (FAST-JX in v10-01c_trop vs. FAST-J in v10-01b):
    • Differences in NO at the south pole
    • Differences in bromine species, particularly at the poles
    • Differences observed in DST4
At 500 hPa, list all species that changed by 10% or more: NO, PAN, ALK4, ISOP, H2O2, ACET, MEK, MVK, MACR, PMN, PRPE, CH2O, N2O5, HNO4, MP, DMS, SO2, NH3, NIT, DST4, Br2, Br, BrO, BrNO2, BrNO3, ISOPN, MOBA, PROPNN, HAC, GLYC, MMN, RIP, IEPOX, MAP, NO2, NO3, HNO2
Comments on 500 hPa differences:
  • Most of the differences in ISOP and its products are over the oceans. This could be due to small number differences, since there is not a lot of ISOP in these locations.
  • We believe that the following differences are due to photolysis (FAST-JX in v10-01c_trop vs. FAST-J in v10-01b):
    • Differences in NO at the south pole
    • Differences in bromine species, particularly at the poles
    • Differences observed in DST4
In the ZONAL MEAN differences, list all species that changed by 10% or more: NO, PAN, ALK4, ISOP, H2O2, ACET, MEK, MVK, MACR, PMN, PRPE, C3H8, CH2O, N2O5, MP, DMS, SO4s, NH3, NIT, NITs, Br2, Br, BrO, HOBr, BrNO2, BrNO3, MPN, ISOPN, MOBA, HAC, GLYC, MMN, RIP, IEPOX, MAP, NO3, HNO2, OH, HO2
Comments on ZONAL MEAN differences:
  • The following species show large percent differences but small absolute differences:
    • NO, PAN, ALK4, ISOP, MEK, MVK, MACR, PMN, R4N2, PRPE, CH2O, C3H8 (at top of atmosphere), N2O5, DMS, SO4s (at top of atmosphere), NITs (at top of atmosphere), DST4, Br2, Br, BrO, HOBr, HBr, BrNO2, BrNO3, CHBr3, ISOPN, MOBA, NO2, NO3, HNO2,
  • Differences in ACET are probably caused by v10-01c having the correct pressure dependency for J(ACET) but v10-01b not having this fix applied.
  • Differences in NH3 and NIT are due to the numerical instability in ISORROPIA II.
  • Differences in HAC and GLYC are due to photolysis (FAST-JX vs. FAST-J).
In the EMISSION RATIO maps, list all species that changed by 10% or more: None
Comments on EMISSION RATIO differences:
  • The NO anthropogenic emissions only very slightly higher in v10-01c_trop (by 0.000343 Tg N). This is most likely due to the fix for FAST-JX wavelength bins propagating through to the PARANOX ship emissions plume model.
  • The ACET ocean sink is only slightly smaller (-0.106) in v10-01c_trop.
Additional or summary comments:
  • Many of the structural updates (i.e. to speed up GEOS-Chem, to reduce the memory footprint, to fix issues in the specialty simulations) do not impact the full-chemistry simulation. This was validated with unit tests and difference tests.
Approval
Requires further investigation: No
Approved by: TBD
Date of approval: TBD

--Bob Y. 11:45, 29 May 2014 (EDT)

v10-01b

Here is the assessment form for 1-month benchmark simulation v10-01b.

Description
New features added into GEOS-Chem:
Developer name(s) and institution(s):
  • PROPNN molecular weight: Jenny Fisher (U. Wollongong)
  • TOMS O3 regridding: Jintai Lin (Peking U.)
Version, resolution, met fields used: v10-01, GEOS-FP (47L), 4x5, July 2013
1-month benchmark finished on: Wed Mar 5 15:06:00 EST 2014
Performance statistics:
  • Ran on 8 CPUs of bench@titan-11.as.harvard.edu (2.659 GHz x 8 CPU)
  • Wall time: 2:55
  • Scalability: 6.7949
Compared to previous benchmark: v10-01a
This update will impact:
(select all that apply with boldface) 		Advection, BL Mixing, Convection, Met Fields, Dry Dep, Wet Dep, Stratosphere, Anthro Emiss, Biogenic Emiss, Biomass Emiss, Photolysis, Chemistry, Other (please specify):
Plots may be viewed at: ftp://ftp.as.harvard.edu/gcgrid/geos-chem/1mo_benchmarks/v10-01/v10-01b/
Metrics
Global mean OH (from log file): 13.7339945052522 x 105 molec/cm3
Methyl chloroform lifetime: 4.6706 years
Did either of these change by more than 5%? No. Both the mean OH and MCF lifetimes differ by < 0.01%.
At the SURFACE, list all species that changed by 10% or more: NO, ISOP, MVK, MACR, PMN, NIT, DST4, ISOPN, MOBA, RIP
Comments on SURFACE differences:
  • Differences in most species are due to small number differences over the oceans or poles, where concentrations are small.
  • Differences in NIT can be attributed to numerical noise from ISORROPIA.
At 500 hPa, list all species that changed by 10% or more: NO, ISOP, NH3, NIT, DST4, ISOPN, MOBA, RIP
Comments on 500 hPa differences:
  • Differences in most species are due to small number differences over the oceans or poles, where concentrations are small.
  • Differences in NIT can be attributed to numerical noise from ISORROPIA.
In the ZONAL MEAN differences, list all species that changed by 10% or more: ALK4, ISOP, C3H8, SO4s, NH3, NIT, GLYC,
Comments on ZONAL MEAN differences:
  • Differences in most species are due to small number differences above ~200 hPa, where concentrations are small.
  • Differences in NIT can be attributed to numerical noise from ISORROPIA.
In the EMISSION RATIO maps, list all species that changed by 10% or more: None
Comments on EMISSION RATIO differences:
Additional or summary comments:
  • Differences caused by the TOMS O3 update are not seen in the benchmark output. The benchmark simulation uses O3 columns from the GEOS-FP met fields, so TOMS O3 columns are not used.
Approval
Requires further investigation: No
Approved by: Jenny Fisher, Fabien Paulot
Date of approval: 06 Mar 2014

--Melissa Sulprizio 11:34, 6 March 2014 (EST)

v10-01a

Here is the assessment form for 1-month benchmark simulation v10-01a.

Description
New features added into GEOS-Chem:
Developer name(s) and institution(s):
  • Patrick Kim (Harvard)
Version, resolution, met fields used: v10-01, GEOS-FP (47L), 4x5, July 2013
1-month benchmark finished on: Mon Feb 3 15:05:32 EST 2014
Performance statistics:
  • Ran on 8 CPUs of bench@titan-09.as.harvard.edu (2.659 GHz x 8 CPU)
  • Wall time: 2:48
  • Scalability: 6.8703
Compared to previous benchmark: v9-02r w/ GEOS-FP meteorology
This update will impact:
(select all that apply with boldface) 		Advection, BL Mixing, Convection, Met Fields, Dry Dep, Wet Dep, Stratosphere, Anthro Emiss, Biogenic Emiss, Biomass Emiss, Photolysis, Chemistry, Other (please specify):
Plots may be viewed at: ftp://ftp.as.harvard.edu/gcgrid/geos-chem/1mo_benchmarks/v10-01/v10-01a/
Metrics
Global mean OH (from log file): 13.7338316128900 x 105 molec/cm3
Methyl chloroform lifetime: 4.6707 years
Did either of these change by more than 5%? No. The mean OH differs by -0.02%, and the MCF lifetimes differ by 0.07%.
At the SURFACE, list all species that changed by 10% or more: NO, O3, PAN, ISOP, HNO3, H2O2, MEK, ALD2, RCHO, MVK, MACR, PMN, PPN, R4N2, PRPE, N2O5, DMS, NH3, NH4, NIT, OCPI, DST4, Br2, Br, BrO, HOBr, HBr, BrNO2, BrNO3, MPN, ISOPN, MOBA, PROPNN, HAC, GLYC, MMN, RIP, IEPOX, MAP, NO3, OH
Comments on SURFACE differences:
  • Differences in isoprene related tracers can be attributed to changing biogenic emissions because we now force the MODIS LAI year to 2008.
  • Differences in NO, DST4, and the bromine species are due to small number differences where concentrations are low.
  • Differences in NH4 and NIT can be attributed to numerical noise from ISORROPIA.
At 500 hPa, list all species that changed by 10% or more: NO, ISOP, MVK, MACR, PMN, PRPE, N2O5, DMS, NH3, NIT, DST4, Br2, ISOPN, MOBA, HAC, GLYC, MMN, RIP, IEPOX, NO3
Comments on 500 hPa differences:
  • Differences in NO, DST4, and Br2 are due to small number differences where concentrations are low.
  • Differences in NH3 and NIT can be attributed to numerical noise from ISORROPIA.
In the ZONAL MEAN differences, list all species that changed by 10% or more: NO, ALK4, ISOP, MVK, MACR, PMN, PRPE, C3H8, N2O5, NH3, NIT, NITs, DST4, Br2, ISOPN, MOBA, HAC, GLYC, MMN, RIP, IEPOX
Comments on ZONAL MEAN differences:
  • Differences in isoprene related tracers can be attributed to changing biogenic emissions because we now force the MODIS LAI year to 2008.
  • Differences in NO, ALK4, C3H8, N2O5, NITs, DST4, Br2, are due to small number differences where concentrations are low.
  • Differences in NH3 and NIT can be attributed to numerical noise from ISORROPIA.
In the EMISSION RATIO maps, list all species that changed by 10% or more: ISOP, ACET, OC, PRPE, CO from monoterpenes
Comments on EMISSION RATIO differences:
  • Biogenic emissions differ because we now force the use of MODIS LAI for 2008 if the simulation year is beyond 2008. MODIS LAI for 2009 was used in the benchmark v9-02r w/ GEOS-FP met, but there are large differences in the 2009 file that still need to be investigated.
Additional or summary comments: We still need to determine what caused the changes in the MODIS LAI input files. As per our earlier discussion, we have put in an error trap in drydep_mod to make sure that the changes are only made for simulations using GEOS-FP and the Olson 2001 land map.
Approval
Requires further investigation: No
Approved by: Patrick Kim
Date of approval: 5 Feb 2014

--Melissa Sulprizio 12:16, 3 February 2014 (EST)

Retrieved from "https://wiki.seas.harvard.edu/geos-chem/index.php?title=GEOS-Chem_v10-01_benchmark_history&oldid=17380"