MERRA2 ICs

[bbakernoaa]

@rmontuoro @zhanglikate @jianheRCM @lipan-NOAA @gjfrost @drpeppurr @perthsb

All I have created is a script that will allow us to start from a cold start using the MERRA2 inst3_3d_aero reanalysis (example file MERRA2_400.inst3_3d_aer_Nv.20160701.nc4) for the tracer ICs. This includes everything for the GOCARTv2 component except for the N component. All nitrogen species are initialized at zero.

The idea is that we can start any retro run using the MERRA2 reanalysis which would save a lot of computational effort for these prototype runs. I'm going to include this in our repository along with instructions on how to create them but we need to make sure that these are working properly before moving forward.

The script interpolates horizontally (using the ESMF bilinear interpolation) and vertically using the logP and the scipy.interpolate.interp1d function. The differences after vertical interpolation were less than 1% total mass before and after.

After just a few hours of integration, the model looks pretty stable

After 1 hour

After 6 hours

after 12 hours

The plan was to create three sets of runs to evaluate the impact of ICs from MERRA2 for the July 1st 2016 time period that coincides with the completed P7.0 runs here

The three simulations are as follows:

• Fully coupled Cold start with MERRA2 initial conditions from July 1st 2016
• Cold start with MERRA2 initial conditions from June 15th ATM only, integrate until July 1st, then merge this into the fully coupled run ICS
• Cold start ATM only with no aerosol ICs from June 1st, 2016. If possible, cycle the model on June 15th with available atmospheric ICs to minimize meteorology drift. Integrate until July 1st, then create aerosol ICs from the July 1st restart file.

From here we compare after a week of integration (July 8) to select the best way to generate aerosol ICs for the P7.1 runs.

I'm going to create a case file for us all to use to complete these runs and put it in the repository which should be completed

[zhanglikate]

@bbakernoaa Thanks very much to prepare for it. To save time, each of us (@lipan-NOAA , @bbakernoaa and me) can perform 1 of this run using the same model configurations. Here I would suggest 1 month (the whole of July ) of free run for these 3 experiments.

@bbakernoaa @rmontuoro Meanwhile, I would like to confirm again, the QFED emission will change daily automatically and we don't need to modify anything in the RC file?

@bbakernoaa If you can let me know the path July 1st 2016 ICs, I can started with the first one immediately.

@rmontuoro To get the one month run to be finished with 8 hours, do you have any suggestions for the layout and ppn?

[bbakernoaa]

@perthsb I did a fully coupled run with MERRA2 ICs could you take a look. The run completed about 28 days. We may need to do some more optimization with PETs and such but it is encouraging.

With this there shouldn't be any need for "spin up" here are assumed to be started from a real state. The output can be found at

/scratch1/NCEPDEV/rstprod/tmp/DATA/MERRA2_JULY1/2016070100/gfs/fcst.306060/gocart.inst_aod.201607?_????z.nc4

[bbakernoaa]

As mentioned above I have created a utility to create UFS-Aerosol/GEFS-Aerosol initial conditions from MERRA2 reanalysis. Note that this could easily be adapted to accept other inputs such as CAMS or something else. I have uploaded it to the repository in the feature/p7.1-gocart-dev branch in the `utils' folder.

merra2_to_fv3_cubesphere.py

This utility takes the MERRA2 3d aerosol reanalysis, e.g. MERRA2_400.inst3_3d_aer_Nv available at https://goldsmr5.gesdisc.eosdis.nasa.gov/data/MERRA2/M2I3NVAER.5.12.4. The utility uses the ESMF bilinear regridding for the horiztonal and the scipy.interpolate.interp1d linear regridding with logP to insure it is positive definate. The utility uses the grid_spec, the fv3 initial condition core_file, tracer file and target resoltion.

Requirements

• python3
• xarray
• fv3grid
• xesmf
• scipy
• netCDF4-python
• argparse

Use

./merra2_to_fv3_cubesphere.py usage: merra2_to_fv3_cubesphere.py [-h] -m MERRA_FILE -c CORE_FILE -t TRACER_FILE -r RESOLUTION [-a AEROSOL] merra2_to_fv3_cubesphere.py: error: the following arguments are required: -m/--merra_file, -c/--core_file, -t/--tracer_file, -r/--resolution

Example

This is a short bash for loop to do this for each tile for the FV3 global model

for i in {1..6}; do  

    ./merra2_to_fv3_cubesphere.py -m MERRA2_400.inst3_3d_aer_Nv.20160701.nc4 -c gfs_ctrl.nc -t gfs_data.tile${i}.nc -r C384

done

[perthsb]

@bbakernoaa @zhanglikate @rmontuoro @lipan-NOAA I looked into Barry's 27 day run for 201607 with MERRA2 IC and comparison for AOD is here :
https://docs.google.com/presentation/d/1btbh3p2m3hDCmtuSP-h_4insjVoPO9CM/edit?usp=sharing&ouid=100966050662930876047&rtpof=true&sd=true

[bbakernoaa]

Thanks Partha,

I think that this shows that u_* is a little stronger over a long period of time. Could you do this daily or weekly? I looked at it hourly (just a preliminary look) and it does seem like AOD isn't building up. I think that a more detailed look would be helpful here. The monthly average can mask a lot.

[zhanglikate]

Hi Barry,

      From Shan’s test before using CCPP-Chem for fully coupled run.  The dust may be large in the 30 day free run even it may be small in the 24 hours cycling run compared to observation. The surface wind may be far away from the GFS in these free run, also for sea salt over some areas.  

Kate

On Sep 14, 2021, at 12:33 PM, Barry Baker @.***> wrote:

Thanks Partha,

I think that this shows that u_* is a little stronger over a long period of time. Could you do this daily or weekly? I looked at it hourly (just a preliminary look) and it does seem like AOD isn't building up. I think that a more detailed look would be helpful here. The monthly average can mask a lot.

— 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/26#issuecomment-919413142, or unsubscribe https://github.com/notifications/unsubscribe-auth/APJPDRHSCCQ5JDOAMPDEORDUB6IQTANCNFSM5DVR2ZTA. Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

[zhanglikate]

Hi Partha,

    It would be helpful if  you  can plot the first hour/day AOD here, which can help to quantify  the impact from MERRA-2. The first couple of weeks results may come from MERRA-2 ICs. 

Kate

On Sep 14, 2021, at 12:33 PM, Barry Baker @.***> wrote:

Thanks Partha,

I think that this shows that u_* is a little stronger over a long period of time. Could you do this daily or weekly? I looked at it hourly (just a preliminary look) and it does seem like AOD isn't building up. I think that a more detailed look would be helpful here. The monthly average can mask a lot.

— 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/26#issuecomment-919413142, or unsubscribe https://github.com/notifications/unsubscribe-auth/APJPDRHSCCQ5JDOAMPDEORDUB6IQTANCNFSM5DVR2ZTA. Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

[jianheACM]

Hi all,

I have made some comparisons with GEFS-Aerosol output and MERRA2 for 20190701. Generally, BC2, OC2, and SO4 are lower in GEFS. Dust4 is higher in GEFS over Sahara whereas other dust tracers are generally lower. Seasalt1-5 are lower in GEFS by a factor of ~ 10 than MERRA2.

I will upload the results to Github later today.

Thanks,

On Tue, Sep 14, 2021 at 12:46 PM Kate.Zhang-NOAA @.***> wrote:

Hi Partha,

It would be helpful if you can plot the first hour/day AOD here, which can help to quantify the impact from MERRA-2. The first couple of weeks results may come from MERRA-2 ICs.

Kate

On Sep 14, 2021, at 12:33 PM, Barry Baker @.***> wrote:

Thanks Partha,

I think that this shows that u_* is a little stronger over a long period of time. Could you do this daily or weekly? I looked at it hourly (just a preliminary look) and it does seem like AOD isn't building up. I think that a more detailed look would be helpful here. The monthly average can mask a lot.

— 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/26#issuecomment-919413142>, or unsubscribe < https://github.com/notifications/unsubscribe-auth/APJPDRHSCCQ5JDOAMPDEORDUB6IQTANCNFSM5DVR2ZTA . Triage notifications on the go with GitHub Mobile for iOS < https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android < https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>.

— 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/26#issuecomment-919421991, or unsubscribe https://github.com/notifications/unsubscribe-auth/ASWQIMYYTRLN4VE53JN2CWLUB6J7NANCNFSM5DVR2ZTA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

-- Jian He CIRES Research Scientist II NOAA Chemical Sciences Laboratory 325 Broadway, Boulder, Colorado 80305

[bbakernoaa]

@jianheRCM is this GEFS-Aerosol or UFS-Aerosol? If it was GEFS was this before or after the bug fix?

We also need to be careful about doing the direct comparison to MERRA2 for the dust distribution as they use completely different dust schemes (MERRA2 uses Ginuoux). These can vary quite a bit from model to model even if they overall predict similar dust events.

I'm interested in the results though

[bbakernoaa]

@zhanglikate @perthsb I've made the 12z plots for each day. https://docs.google.com/presentation/d/1OK93kb3G60vbgPPVNyLHVE8Tf6fuYqiDTq39hPtBY_8/edit?usp=sharing

[zhanglikate]

@bbakernoaa I think Jian's results is from operational GEFS-Aerosols, we don't have UFS-Aerosols for that time now. Also, there is not bug fix in the operational GEFS-Aerosols.

[bbakernoaa]

@zhanglikate That is why I asked what exactly it is so that I can understand what I would be looking at

[zhanglikate]

@jianheRCM I am wondering your comparison is concentration or AOD? I suppose it is concentration. For concentration, thus we don't need to care about the UPP bug. While I think the FENGSHA dust bug is still there in the operational, right? @bbakernoaa

[zhanglikate]

@bbakernoaa Thanks for the plotting. I still think it deserves to try a run without aerosol ICs as a baseline for comparison. Otherwise, we don't have the idea of the impact from aerosol ICs in the 35 days, not matter which aerosol ICs we are using.

@zhanglikate @perthsb I've made the 12z plots for each day. https://docs.google.com/presentation/d/1OK93kb3G60vbgPPVNyLHVE8Tf6fuYqiDTq39hPtBY_8/edit?usp=sharing

[bbakernoaa]

I agree @zhanglikate

I am still seeing that while these storms may be over producing in the coupled run after a long period from initialization has passed these do still seem like real dust storms.

On the 18th from VIIRS Deep Blue

and the simulated result

[jianheACM]

@rmontuoro @zhanglikate @bbakernoaa @lipan-NOAA @gjfrost @drpeppurr @perthsb

Hi all,

Here are some comparisons of aerosol mass concentrations (surface and column) between GEFS-Aerosol and MERRA2. Compare.ICs.JH.20210914.pptx

https://docs.google.com/presentation/d/1Gt8Eb_CjXp9NmtVTE36Ga0HjdFfzfa4L/edit#slide=id.p1

[perthsb]

@bbakernoaa @zhanglikate I looked into day to day AOD ocmparison for MERRA2 IC run and from 7th July onward model shows more dust and OC https://docs.google.com/presentation/d/1E9hLB8d-MrxBMpm6LKGdhxhCdLpczBpW/edit?usp=sharing&ouid=100966050662930876047&rtpof=true&sd=true

[perthsb]

@bbakernoaa I have daily Dust AOD for this case here https://docs.google.com/presentation/d/1pxT5t1zbWyr_UM375B3RQ8_IIle_m7sP/edit?usp=sharing&ouid=100966050662930876047&rtpof=true&sd=true

[bbakernoaa]

@perthsb do you have the ability to do a 10m wind speed evaluation for these? I would be very interested if this is especially for the longer runs. Could explain a lot about what is happening with dust and sea salt.

[perthsb]

You want only U10m or total wind speed and direction ?

On Wed, Sep 15, 2021 at 11:12 AM Barry Baker @.***> wrote:

@perthsb https://github.com/perthsb do you have the ability to do a 10m wind speed evaluation for these? I would be very interested if this is especially for the longer runs. Could explain a lot about what is happening with dust and sea salt.

— 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/26#issuecomment-920110058, or unsubscribe https://github.com/notifications/unsubscribe-auth/AVARVDSNGV3PQU2IERRTIGDUCCZVZANCNFSM5DVR2ZTA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

-- Partha S. Bhattacharjee, Ph.D. I.M. Systems Group Environmental Modeling Centre NCWCP 5830 University Research Court. College Park, MD 20740

[bbakernoaa]

the total 10m wind speed is enough.

[bbakernoaa]

Mike Barlage would also be interested in this I'm sure

[bbakernoaa]

All I have also uploaded all the ICs for the P71 runs we would possibly need to do by uploading the MERRA2 3d data for the 1st and 15th of each month from 2013 to 2018. You can find more details here

[perthsb]

@bbakernoaa against GEOS5 analysis ? I can pull that quickly and take a look. As I was GEOS5 to verify AOD, it is a good idea to see met field verifying against same analysis?

[bbakernoaa]

GDAS might be better to verify against that. But whatever is simplest for you. I think that we just need an idea of how much the surface wind is gathering momentum. Two of our parameters are directly effected by it

[perthsb]

Comparison against GDAS 10m wind speed is here : https://docs.google.com/presentation/d/17IcXdTDfVCgzZZ_b7PLI6wdKoyGQR4RN/edit?usp=sharing&ouid=100966050662930876047&rtpof=true&sd=true

[lipan-NOAA]

@bbakernoaa @zhanglikate @rmontuoro we don't have 2013 emission inputs for both qfed and ceds

[bbakernoaa]

I’m pulling them over now

On Fri, Sep 17, 2021 at 2:18 PM lipan-NOAA @.***> wrote:

@bbakernoaa https://github.com/bbakernoaa @zhanglikate https://github.com/zhanglikate @rmontuoro https://github.com/rmontuoro we don't have 2013 emission inputs for both qfed and ceds

— 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/26#issuecomment-921991104, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFIUVN4E7ZMGUVWAW4W2D2TUCOA27ANCNFSM5DVR2ZTA .