Open zhanglikate opened 2 years ago
This case runs from 2016-07-15
to 2016-08-23
using the default settings?
2016-07-15
2016-08-23
Dust and sea salt appears to be biased low compared to ICAP and ATOM measurements.
/scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v1_na1
Same as v1 except that AOD bug fixed in GOCART by Raffaele (GOCART: 885af772) Dust scheme is based on FENGSHA and use the *.rc file from Barry’s GITHUB (5102228b), alpha value changed from 0.4 to 0.7 Sea salt emission factor changed to 1.0 for all bins (default: emission_scale: 0.613 0.613 0.613 0.429 0.429 0.429).
AOD improved after the AOD bugs were fixed, while still low biases of AOD almost globally. Dust is still low and may need some improvement. Surprised to see the low sulfate over Eastern China. Now it is using the HTAP anthropogenic emissions. Suppose the CEDS_2019 may have much lower SO2 over that area? Also the OC is much lower in the downwind areas of the fire source region, which may be related to the absence of plume-rise. Sea salt is low over tropical, may be related to the convective wet deposition. Note: Heterogeneous chemistry is included into Nitrate simulation. Nitrate concentration will be impact by other aerosols.
/scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v2_na1
Same as v2 except
/scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v3_na1
The changes of convective wet scavenging of sea salt from 1.0 to 0.0 showed significant impact of sea salt concentration, which means that it is working well. Sulfate also showed changes when changed from 0.6 to 0.2. However, the impact on OC is very small, almost the same when changed from 0.6 to 0.2. The UPP looks working well for AOD calculation after the new implementations, which showed quite similar results as the online AOD calculated by UFS-Aerosols. Generally, the UPP AOD is a little bit higher than that of the online AOD. Meanwhile, the UPP does not include the nitrate look-up table,so the UPP total AOD is smaller than that of the online AOD.
Update to feature/p7.1
of UFS-Aerosols which includes an AOD bug fix and fengsha dust bug fix
rc files are updated to the develop branch
/scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v4_na1
Dust shows some improvements, but still low over Africa and the Middle East. Sea salt shows good improvement by using 0.6 for convective wet scavenging. However 0.2 is too low for BC, NO3, NH4, will switch back to 0.6. Compared with ICAP and GEOS-5, OC AOD is low over downwind areas of the fire source regions. Sulfate AOD slightly higher compared to GEOS-5. Nitrate can not reproduce the high concentration near the fire regions in both the Atlantic and California.Dust AOD is still low when compared to ICAP and GEOS-5, especially over the Middle EAST, part of north Africa. OC AOD is much higher over the south African and south American fire source region, however, lower over the downwind areas of atlantic.
Same as v4 except:
Sulfate AOD is overestimated over eastern China and eastern Asia.
/scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v5_na1
Same as v4 and v5 except:
/scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v6_na1
Same as v6 except:
/scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v7_na1
@zhanglikate started another run with some additional fixes/settings. Most notably a bug was found in the AOD calculation which only calculated AOD at a single wavelength. This has been hot fixed by copying code from @rmontuoro and recompiling until the pull request he has created is accepted into the authoritative repository. Also, there is a default resolution scaling that NASA uses for sea salt. This is broken on the NOAA grids as they use slightly different grid settings which do not translate to our FV3 grids. The fix is to modify the SS .rc file and set the resolution setting to a constant value across scales, i.e., change emission_scale: 0.613 0.613 0.613 0.429 0.429 0.429
to emission_scale: 1.0 1.0 1.0 1.0 1.0 1.0
2016-07-15
2016-08-23
Output: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v2_na1
Here is a comparison of Total and dust AOD from two experiment runs against ICAP. Total AOD : https://drive.google.com/file/d/17IO5Vg4u-eyTBGRcewiu9OfucNp8SUei/view?usp=sharing Dust AOD : https://drive.google.com/file/d/1N3WTVrMS0XjXX0qAQ9KjlDbltLtBQgzW/view?usp=sharing
Thanks @perthsb
All note that the dust is using preliminary dust inputs. The updates include the move from the SOILGRIDS 2017 dataset to SOILGRIDS 2019. With this, a new threshold map has been created and is still in development. We should not expect a 1 to 1 comparison with GEFS-Aerosol at the moment. We are just looking to get this on the same scale which I think that we have according to @perthsb new analysis.
@zhanglikate started another run with some additional fixes/settings. Most notably a bug was found in the AOD calculation which only calculated AOD at a single wavelength. This has been hot fixed by copying code from @rmontuoro and recompiling until the pull request he has created is accepted into the authoritative repository. Also, there is a default resolution scaling that NASA uses for sea salt. This is broken on the NOAA grids as they use slightly different grid settings which do not translate to our FV3 grids. The fix is to modify the SS .rc file and set the resolution setting to a constant value across scales, i.e., change
emission_scale: 0.613 0.613 0.613 0.429 0.429 0.429
toemission_scale: 1.0 1.0 1.0 1.0 1.0 1.0
Model Configuration Details
- Case Yaml File: ufs_aerosol_free_forecast.yaml
- Resolution: C96
- Begin Date:
2016-07-15
- End Date:
2016-08-23
- Cold Start: True
- FHOUT_GFS: 6
- FHMAX_GFS: 24 hour cycling
- na_init=1 (Note that this will have mass conservation issues)
- LEVS: 65
- CCPP suite: FV3_GFS_v16
Chemical Configuration
HTAP
- ANTHRO1 (surface)
- ANTHRO2 (elevated; energy)
- SHIP
- NH3; Agriculture, residential, energy, transport (road and ship), industry
- QFED biomass burning
- HTAP - aviation
FENGSHA dust emissions v0.6
- alpha = 0.7
- gamma = 1.0
Output: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v2_na1
One correction, the case file is using "Case Yaml File: ufs_aerosol_fire_forecast.yaml", not "Case Yaml File: ufs_aerosol_free_forecast.yaml"
ATom-1 evaluation Case directory: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v2_na1
Thanks @drpeppurr
Thanks @drpeppurr
Barry, from the underestimate of both dust AOD and dust concentration over Pacific, any suggestions for my next run about dust tuning. Feel free to let me know and I will update it into my next run. Thanks.
Honestly @zhanglikate I think that it’s on me to do some more runs to improve the inputs
I have added bias plot here based on Kate's two runs https://drive.google.com/file/d/1qq45d5kag6h0o-7BNvaXxwLMLezJheS-/view?usp=sharing
Day 1 AOD bias based on Experiment 2 https://drive.google.com/file/d/1LbH5nZlcr046HkIGX6KduHf3VCqRpomk/view?usp=sharing
Day 1 AOD bias based on Experiment 2 https://drive.google.com/file/d/1LbH5nZlcr046HkIGX6KduHf3VCqRpomk/view?usp=sharing
Thanks for the updating, that is really helpful for next tuning. Surprised to see the low sulfate over Eastern China. Now it is using the HTAP anthropogenic emissions. Surprised to see the low sulfate over Eastern China. Now it is using the HTAP anthropogenic emissions. Suppose the CEDS_2019 may have much lower SO2 over that area? Also the OC is much lower in the downwind areas of the fire source region, which may be related to the absence of plume-rise. Sea salt is low over tropical, may be related to the convective wet deposition.
@zhanglikate are you going to run a winter month with same setting as of Exp2 before tuning experiments ?
@zhanglikate are you going to run a winter month with same setting as of Exp2 before tuning experiments ?
Not yet. I would like to change emissions and do some tuning test for the EXP.2 again to get a better shape. Then may consider switch to other periods.
Could we run UPP to compute AODs from Exp2 results?
@rmontuoro
From the workflow you send to us, I don't think the UPP is working for current model output. I am wondering nobody updated that part for the UFS-Aerosols model in the workflow? I think the UPP need to be updated with current model output. Also, just roughly checked, you modified the tracer name of "SULF" to "SO4", if I modified the UPP , then it can not used for CCPP-Chem and GEFS-Aerosols. So I am wondering can we update it to be consistent for the chemical tracers? Otherwise, UPP need more extra work to design for different model which requires more extra works. Thanks.
I uploaded the side-by-side comparison for the following two cases: v1_na1: '/scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v1_na1/' v2_na1: '/scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v2_na1/'
including basin-wide average profiles for Pacific and Atlantic, as well as vertical profiles in different regions: A: North America + High Latitude Northern Pacific/Atlantic B: Subtropical Northern Pacific C: Tropical Pacific D: Subtropical Southern Pacific E: Southern Ocean F: Subtropical Southern Atlantic G: Tropical Atlantic (African dust/biomass burning) H: Subtropical Northern Atlantic (African dust/biomass burning)
@drpeppurr Thanks very much for the results, that is very helpful. @rmontuoro One confused me, there is no change for nitrate in the two experiments settings. Surprised to see they are different, not b4b identical. I only changed the parameters related to dust and sea salt. So I am wondering did you change anything in the GOCART related to nitrate and NH3 when you fixed the AOD bug?
@zhanglikate are you going to run a winter month with same setting as of Exp2 before tuning experiments ?
Not yet. I would like to change emissions and do some tuning test for the EXP.2 again to get a better shape. Then may consider switch to other periods.
@zhanglikate its because there is some interaction with SS and dust. Look here https://github.com/GEOS-ESM/GOCART/blob/main/ESMF/GOCART2G_GridComp/NI2G_GridComp/NI2G_GridCompMod.F90#L844
@zhanglikate - As @bbakernoaa pointed out, both dust and sea salt mixing ratios are used to compute contributions from heterogenous chemistry to the nitrate mixing ratios.
@rmontuoro @bbakernoaa Got it. Good to know that. Nitrate has included the heterogenous chemistry in the NASA GOCART, which would be more sensitive than other aerosols. Thanks.
I'm attaching some quick AOD evaluation using the ATom in situ AOD data product (more details of this data product can be found here: https://acp.copernicus.org/preprints/acp-2021-173/). We are using this same data product to evaluate VIIRS.
In brief, this in situ AOD is calculated using speciated, size resolved aerosol measurements from multiple instruments. A small caveat of this version of the AOD dataset is that, the stratospheric AOD is not considered. A new version should come out soon, in which the stratospheric AOD is from GLOSSAC. Overall the contribution of stratospheric AOD is small, on the order of 0.001 or so.
The AOD data is from the gridded nc4 files, e.g., /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v1_na1/gfs.yyyymmdd/00/chem/gocart.inst_aod.yyyymmdd_hh00z.nc4
The maps are "ATom-1 campaign averages", created using hourly nc4 files when there was ATom flights.
@drpeppurr Thanks for the results. I think you can only use the results from EXP.2: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v2_na1, not included EXP.1: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v1_na1 as Raffaele suggested before. Because we all know there is a bug in the AOD calculation in EXP.1. So we will not add any EXP.1 results into our evaluation figures to confused others. From EXP.2, the bug has been fixed, which can be used for our evaluation and comparison. Thanks.
@drpeppurr Thanks for the results. But I would like to check with you which experiment results did you compare? I think you should use the results from EXP.2: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v2_na1, not EXP.1: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v1_na1. Because we all know what there is a bug in the AOD calculation in EXP.1. So we will not add any EXP.1 results into our evaluation figures. From EXP.2, the bug has been fixed, which can be used for our evaluation and comparison. Thanks.
See figure legends/titles. I plotted both EXP1 (v1_na1) and EXP2 (v2_na1).
@drpeppurr Oh, I saw the figures. Some problem in my network just now, and it did not load the figures at the same time. For the ATOM-1 AOD, it looks like there is not the data over California flight track, right? I did not see that in your figure 3. Similar to Pacific and Atlantic, it may be interesting to see the comparisons over the continent of North American areas. Our previous model showed higher AOD over US. Thanks.
@zhanglikateif you check result with GEOS5 or ICAP, over California there is some underprediction of AOD over West coast and that is coming from OC AOD based on Exp2 run.
@drpeppurr Oh, I saw the figures. Some problem in my network just now, and it did not load the figures at the same time. For the ATOM-1 AOD, it looks like there is not the data over California flight track, right? I did not see that in your figure 3. Similar to Pacific and Atlantic, it may be interesting to see the comparisons over the continent of North American areas. Our previous model showed higher AOD over US. Thanks.
Welp, this dataset relies on in situ measurements, therefore AOD is calculated only if a certain flight leg extended from the surface to a certain altitude (I think it's 12 km). There weren't a lot of full profiles during the transition flights, which mostly went from point A to point B in a straight line.
I have scripts setup to evaluate AOD using L3 daily gridded VIIRS AOD, if that's useful.
Reopening this issue
All remember that there was a dust bug fix and new inputs. Please see #14 and #13
@zhanglikate @lipan-NOAA @rmontuoro @drpeppurr @perthsb @gjfrost
All, I've updated the develop branch .rc
files.
These are not ready for p7.1 as some changes need to be made to include all necessary inputs. I was delayed trying to get these finished as I was only going to release for the NASA rc1 but as there seem to be some workflow/FV3 issues I'm pushing forward with these.
I also created three different emission scenarios for future use. The default uses the HTAP climatology as was in @rmontuoro original. Next, a CEDS 2014 + HTAP daily emission scenario which is essentially what GEFS-Aerosol is running (note that this does not include the 50% reduction in SO2 that was being applied internally in the model (you can apply it in the AERO_ExtData.rc
file directly if an apple to apple comparison to GEFS is needed). Finally, a CEDS 2019 + HTAP daily emission scenario.
I also have downloaded and put the different biomass burning emission in each scenario. All that is needed to switch between datasets is to uncomment the block for the wanted emission dataset. QFED is currently the default in these. I have all the QFED and GFAS data for 2016-> 2019 and the GBBEPx period during the ATOM1-4 campaigns and for July 20th -> end of the year for 2019.
@bbakernoaa @rmontuoro @lipan-NOAA
Hi Barry, I just checkout your develop branch (commit b523dd78d122f3d8bbc61a7f35161bc14536d840 (HEAD -> develop, origin/develop)), and found the your *.rc file has not yet been updated r as the AOD fixed version. They are still the old one as AOD bug version before. I think it would better to update them in this branch.
Also, I would like to confirm with you, the following are the updated values after the dust bug fixing, right? Thanks.
alpha: 2.2 gamma: 1.0
@zhanglikate please for it and make a pull request with the fixed values. It is helpful for us all to contribute and maintain.
On Mon, Aug 16, 2021 at 7:19 PM Kate.Zhang-NOAA @.***> wrote:
@bbakernoaa https://github.com/bbakernoaa @rmontuoro https://github.com/rmontuoro @lipan-NOAA https://github.com/lipan-NOAA
Hi Barry, I just checkout your develop branch (commit b523dd7 https://github.com/noaa-oar-arl/UFS-Aerosol-Config/commit/b523dd78d122f3d8bbc61a7f35161bc14536d840 (HEAD -> develop, origin/develop)), and found the your *.rc file has not yet been updated r as the AOD fixed version. They are still the old one as AOD bug version before. I think it would better to update them in this branch.
Also, I would like to confirm with you, the following are the updated values after the dust bug fixing, right? Thanks. FENGSHA settings
alpha: 2.2 gamma: 1.0
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/noaa-oar-arl/UFS-Aerosol-Config/issues/3#issuecomment-899879924, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFIUVNY36N52N63BR25PN7TT5GMI3ANCNFSM5BJMXRGQ .
@bbakernoaa Just submit the pull request, can you review it and approve? I can not choose reviewers.
@bbakernoaa One more commit to match the model version feature/p7.1 (https://github.com/noaa-oar-arl/UFS-Aerosol-Config/issues/14), please review and approve it.
@perthsb @drpeppurr EXP.4: ATom-1 (C96, 20160715-20160823, analysis from 20160729), Aug 17, 2021 Using P7.1 mode source code (44e4c2b283) as described in: https://github.com/noaa-oar-arl/UFS-Aerosol-Config/issues/14, FENGSHA dust and AOD updates have been included. .rc files are updated as https://github.com/noaa-oar-arl/UFS-Aerosol-Config/tree/develop/rc.hera Test the convective scavenging values by turning down and turn off sea salt as followings: fscav_aero: [ ':0.2', 'so4:0.3', 'seas1:0.0', 'seas2:0.0', 'seas3:0.0', 'seas4:0.0', 'seas5:0.0' ] (default is fscav_aero: [ '*:0.6', 'seas1:1.0', 'seas2:1.0', 'seas3:1.0', 'seas4:1.0', 'seas5:1.0' ]) Continue test UPP capability for UFS-Aerosols Output: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v4_na1/
@zhanglikate Here is comparison of Inline and UPP computed AOD against ICAP and GEOS5 for all species. https://docs.google.com/presentation/d/1sxTvuVp_EDPEBFZHaehXc6UiErUucr5b/edit?usp=sharing&ouid=100966050662930876047&rtpof=true&sd=true
@perthsb Thanks very much. That is very helpful. I will have a new run soon by using CEDS_2014 emissions.
and remember @zhanglikate that this will not include the 50% blanket reduction in SO2 that was within CEDS2014 so we should expect a high bias in eastern Asia compared to other models.
On Thu, Aug 19, 2021 at 11:40 AM Kate.Zhang-NOAA @.***> wrote:
@perthsb https://github.com/perthsb Thanks very much. That is very helpful. I will have a new run soon by using CEDS_2014 emissions.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/noaa-oar-arl/UFS-Aerosol-Config/issues/3#issuecomment-902019479, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFIUVNYPZHLFILVR4TOPFVLT5UQVPANCNFSM5BJMXRGQ .
Hello all,
here are the ATom-1 overview plots.
v2_na1: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v2_na1/ v3_na1: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v3_na1/ v4_na1: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v4_na1/
Pacific and Atlantic averages. ATom measurements below LODs are replaced with 0.5*LOD. The lower bounds roughly reflect the instrument LODs, expect for nitrate which is mostly below LOD.
These are AOD comparisons. The modeled AOD fields are averaged using only days with ATom flights.
These curtains are for v4_na1 only
These are detailed vertical profiles for different regions:
A: North America + High Latitude Northern Pacific/Atlantic B: Subtropical Northern Pacific C: Tropical Pacific D: Subtropical Southern Pacific E: Southern Ocean F: Subtropical Southern Atlantic G: Tropical Atlantic (African dust/biomass burning) H: Subtropical Northern Atlantic (African dust/biomass burning)
@drpeppurr Thanks! Dust looks much improved here. Maybe even a little too much production in some areas. It does look like I'm missing some of the dust production from Patagonia and maybe Bolivia.
Asian dust is vastly improved though as well as African.
@zhanglikate could you update here what v3 and v4 are?
@bbakernoaa For dust, v4 is the version with your dust bug fixing, v3 is your old dust routine, the convective wet scavenging in v3 and v4 are the same by using 0.2. v3 and v2 are using the same dust code, but the convective wet scavenging in v3 is 0.2, however in v2 is 0.6. So I will give a try by using 0.5 for dust in my next run (v5).
@perthsb @drpeppurr @bbakernoaa Please find the EXP.5 results by using CESD_2014 emission at: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v5_na1 Other than the change in anthropogenic emission, I also modified the dust wet scavenging values from 0.2 to 0.6 compared to EXP.4.
@zhanglikate @bbakernoaa @drpeppurr I have included both inline and upp AOD comparison for Exp5 here : https://docs.google.com/presentation/d/1I8f2-pTTfQZweh8ni5tRxZ-9_DNOf-bt/edit?usp=sharing&ouid=100966050662930876047&rtpof=true&sd=true
@perthsb Can you put the online and UPP AOD in the same slide, which is more straightforward to compare? Thanks.
@zhanglikate , I put Inline and UPP aod together in each slide here : https://docs.google.com/presentation/d/1qPzes9WB1_C9xdgiUyudT6rh096hYN0f/edit?usp=sharing&ouid=100966050662930876047&rtpof=true&sd=true
@perthsb @bbakernoaa @drpeppurr Please find the EXP.6 output at: /scratch1/BMC/gsd-fv3-dev/NCEPDEV/global/Kate.Zhang/fv3gfs/comrot/NASA_C96_fire/v6_na1
In EXP.6, The differences compare to EXP.5 is the anthropogenic emission is based on CESD_2019 and the dust convective wet scavenging values have been updated as 0.7, they are 0.6 in EXP.5.
@zhanglikate Exp6 comparison are : Link : https://docs.google.com/presentation/d/1R5aXo5AxvCGmnMxhi0xlmDroMC0QS4VT/edit?usp=sharing&ouid=100966050662930876047&rtpof=true&sd=true
Please check the records in the file: https://docs.google.com/document/d/1EWZV4VOvvx5d4rpjXp5dGJlP7Pa67HuA_GaC9xuGugw/edit