Open dan-visioni opened 1 year ago
@tilmes @dan-visioni Simone, maybe Dan can look at my MAM climatology run to change the cyclical conditions?
/glade/u/home/yunqian/cesm/case/FWma2000climo_test/
Yes, I agree, @dan-visioni the emissions and other cyclical settings should be the same as in the FWma2000 climo run from Yunqian. You may also want to compare to the CARMA ISA-MIP run from Yunqian for other namelist differences to make sure they are identical, including the effgw_beres_dp etc.
I am updating the namelist to make it similar to Yunqian's ISA-MIP one. Differences worth noting - please help/correct if something is wrong.
In ext_frc_specifier there's also so4 particles - I assume this is different in CARMA as they are defined somewhere else In srf_emis_specifier same thing about so4 - plus there is the CARMA run there is a 0.3 value: should I update this in MAM as well?
&surface_emissions_opts part missing in MAM: should I add it?
All other parts made identical to CARMA
List of variables: excluding the ones specific to the aerosol bins/modes, my namelist has more variables than CARMA's. Should I trim it down or worth keeping more stuff?
@tilmes just checking: just checking: is /glade/scratch/tilmes/archive/FWmaMAM4HIST.f19_f19_mg17.carma_trop_strat09.1980_2014/rest/2000-01-01-00000 the correct restart file to use?
Also @zywshoon I'll definitely need a refresher on how to upload stuff, I did it for HerSEA but it was 2 years ago and I have no memory!! :)
@dan-visioni for emissions, you need the same species as what is in FWmaMAM4HIST.f19_f19_mg17.carma_trop_strat09.1980_2014 and yes MAM4 includes so4 particles. This is usually just 2.5% of SO2 emissions, so this should be small and I left those in. It may give a slight difference in the total sulfur emissions. &surface_emissions_opts is only for CARMA Regarding the 0.3 factor for SOAG, this is only for CARMA not MAM4. Output, I think, Yunqian has made an effort to find all the important output. It depends on what else there is in the output, it may not hurt to keep it, if it is not too much. Restart files, yes this is the correct one. Are we doing a hybrid run?
The 2000 folder is empty (the 1999 one, for instance, is not) Did you move it somewhere else? And I guess yes, we need a hybrid run
Oh, I may have just saved every other year. I think, it should not really matter if you use 1999. I however can just run the year 1999 and save the restart file for 2000 if you think that is better?
No worries I think 1999 is perfectly fine. If you want to take a look at the namelist I can try to build and submit :)
@dan-visioni Here are some changes for the namelist, I adviced wrong earlier, sorry for that, the emissions should be actually from the climo case, see point 2.
Also, please make sure you ask for ACOM computer time, if you use P19010000, however this run is not expensive is is about 2.5K per year (throughput is more than 8 years/day).
Ok, what project should I be using however? I've never had to ask for ACOM computer time so I don't know how to do it, but isn't there a specific project for this?
A comparison between @zywshoon run and one year of MAM run... (zonal mean vs time on the left, annual mean for the same year on the right). MAM is 3 to 4 times higher than CARMA everywhere: is this expected, or is something wrong in my simulations?
I would like to look at this more in depth, but am still unfamiliar with CARMA... is there a variable for total column mass of SO4 like TMso4 in MAM?
Thanks for checking! @dan-visioni can you also plot this run: /glade/scratch/tilmes/archive/FWmaMAM4HIST.f19_f19_mg17.carma_trop_strat09.1980_2014 to see if that is consistent? I thought we were doing quite well in the MAM4 run.
@dan-visioni see https://github.com/NCAR/CARMA_dev/issues/1 The plot I posted last showed very similar AODVIS for CARMA and MAM4 around year 2000 Maybe it is different because you are plotting AODVIS and not AODVISst?
Ok, great idea. Here's the comparison between the 3. ISAMIP MAM run seems = to HIST, so it's ISAMIP CARMA much lower? Just checking: I'm comparing AODVIS output from CARMA with AODVISdn MAM (because AODVIS MAM has NaNs at high latitudes) but there is almost no visual difference between AODVISdn and AODVIS MAM
@dan-visioni @tilmes This is the total AOD including the tropospheric aerosol, So I'm not surprised they are quite different. The stratospheric AOD should be close. But unfortunately, I haven't saved stratospheric AOD for CARMA or maybe there's not such an output.
Alternatively, you may compare the extinction at 50 hPa and 100 hPa between CARMA and MAM. I think those should be consistent.
Ok, I'll trust you on this! Do you know if there's an output for total mass burden I can look? Also, in term of dates, your simulations say 1980-2000 (even if the cyclical conditions are from 2000). Is this a requirement from ISAMIP, or can I just start from 2000?
@dan-visioni The reason is that they want nudged QBO. That's why Simone set up the model this way for me, ie from 1980-2000.
Well, I cannot promise CARMA and MAM are the same at 50 hPa. But if you run it correctly, they should be very close.
I see. So maybe I should start in 1980 as well. @tilmes what do you think? This was just a (useful, for me) test!
I changed a few things based on suggestions: changed usr_nl_clm to add finidata (1999) changed ncdata to 1999 changed run_type to startup changed refdate to 1999 changed startdate to 1980
All the rest should be ok? I can start after a last check from Simone
@dan-visioni yes looks good.
@tilmes when submitting, I get the error
Model cam missing file ncdata = '/glade/p/acom/acom-climate/tilmes/inputdata/init/carma/f.e21.FWmaHIST_BGC.f19_f19_mg17.CMIP6-AMIP-WACCM-MA-2deg.001.cam.i.1999-01-01-00000.nc'
I think you need to put it there (or tell me where it is!)
The IC from the transient WACCM MAM4 run would be better to use, since it has the same tuning as the run I have done, do you agree? I copied it here: /glade/p/acom/acom-climate/tilmes/inputdata/init/carma/FWmaMAM4HIST.f19_f19_mg17.carma_trop_strat09.1980_2014.cam.i.1999-01-01-00000.nc
Makes sense. I don't seem to be able to copy it, I think you need to change the permissions.
And I'll also need the FWmaMAM4HIST.f19_f19_mg17.carma_trop_strat09.1980_2014.clm2.r.1999-01-01-00000.nc for the user_nl_clm
hope it is all there now
Seems to have worked this time! Thanks
@drmikemills Mike, just a check for this intercomparison. The output so4_a1, a2 and a3 are mass of SO4 or of NH4HSO4? Thanks
They are kg/kg air for NH4HSO4. SO4 conversion would be 96.0576 g/mol divided by 115.10734 g/mol.
MA simulations look good after exploring a bit, so I'll upload them. @zywshoon can you please send me the instructions for the upload? I had them but can't find them. Thanks!
@dan-visioni did you find out why our AOD with MAM4 is much larger than with CARMA? Is this related to the tropospheric aerosol? We may wait and see what comes out of the discussion that we have regarding the issue with CARMA and likely MAM4 producing extra sulfate in the microphysics?
I haven't really looked at the CARMA runs yet, I'll try to dig in more in the next days and report back!
@dan-visioni just noticed that the run has transient SSTs, we want to run with fixed year 2000climo SSTs, so we have to change these things in env_run.xml:
I am not sure how much differences this will make. I am going to setup a run with the MG substepping turned off, and add those changes.
Ok, fixed. I'll hold on and in case rerun tomorrow/
here is the budget for 1990-1999: for reference:
I performed 2 more runs,
Here are the budgets for the second year:
Sulfates are dropping a bit from 0.4 to 0.33.
I am continuing for another year, to make sure the run is not trending.
This also reduces AOD in the stratosphere, which will make it probably smaller than observations. However, we can increase the sulfate again if we finally add the fix for the aqueous phase chemistry in. This would cause an increase in sulfate, but we have to see of how much that would be.
However, we still have to be careful, since clouds may be are changing and that seems to also change aq-chem production. For MAM4, we have to wait for a fix from Adam/Chuck/Andrew.. so we probably don't want to change this yet. This was just a test to see what it does.
Going back to having 3 substeps for MG and one 1 for macmix, is setting AQ-phase chemistry back to initial values, and the burden is also not as much impacted, see budget below. So, this seems to not change clouds too much, but would need more analysis.
Next experiment will be to add the aq-chem fix in here and see where we end up.
My last case also includes Aq.chem bug fixes, we are not including ice clouds in this case: /glade/p/cesm/chwg_dev/tilmes/cases/carma/ISAMIP_CESM2MAM4_WACCMMA2000climo.mg3_macmix1.aq_chem Numbers for second year:
@dan-visioni @zywshoon I performed two runs for WACCM-MA MAM4, as you can see above. with the changed microphyscis, and with aq.chem changes in /glade/scratch/tilmes/archive ISAMIP_CESM2MAM4_WACCMMA2000climo.mg3_macmix1.aq_chem and ISAMIP_CESM2MAM4_WACCMMA2000climo.mg3_macmix1
While the numbers are still not adding up for CARMA, I have a new run with some bug fixes, but I have not include the change in aq.chem: ISAMIP_CESM2CARMA_WACCMMA2000climo.emis_fix_mg3macmix1_wetfix
We can compare these runs, all of them have about 4 or 5 years of output, and should be in steady state staring in 1982. I included AODVISst for all the cases.
Regarding Reff calculations, please see @drmikemills code: /glade/work/mmills/scripts/ncl/MAM/calcSingleVar.ncl and grep for "Column SAD-weighted wet effective radius" This is a column SAD-weighted wet effective radius, above 300hPa. I am not sure how to best plot zonal mean weighted effective radius, I hope @drmikemills can help?
I think, the main question is, how much the changes we made impact the stratospheric aerosol burden.
@drmikemills Mike, following on my question before, do I need to apply the same conversion to get SO4 also to BURDENSO4, AQSO4_H2O2, so4_a*DDF, etc.?
@dan-visioni I do have code for calculating budgets, do you want me to share that, and Chuck has that as well. It would be interesting to look at AODVISst for now I think.
@tilmes thanks, the code would be useful. But I need an answer also fro ISA-MIP more in general :)
The answer is yes, all production and loss terms of sulfates would need the same conversion.
@cbardeen Just a thought, the WACCM MA F2000 MAM4 run seems to be balanced with regard to loss and production (see above) same as the nudged run. I am not sure why MAM4 is balanced, and CARMA is not. Maybe we are missing including something specific for CARMA in the energy fixer code, that is however there for MAM4?
I don’t know what could be in CARMA that is different than any other tracer with regard to the energy fixer. CARMA is more sensitive to errors in tracer/tracer relationships and is more dependent on having it’s pbuf fields be kept in sync with state. But fundamentally, it is just a bunch of tracers.
On Feb 6, 2023, at 7:10 PM, tilmes @.***> wrote:
@cbardeen https://github.com/cbardeen Just a thought, the WACCM MA F2000 MAM4 run seems to be balanced with regard to loss and production (see above) same as the nudged run. I am not sure why MAM4 is balanced, and CARMA is not. Maybe we are missing including something specific for CARMA in the energy fixer code, that is however there for MAM4?
— Reply to this email directly, view it on GitHub https://github.com/NCAR/CARMA_dev/issues/12#issuecomment-1420081204, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFLRNRDQMZD7ZEOWFZQT27DWWGVJDANCNFSM6AAAAAATSFVLKM. You are receiving this because you were mentioned.
Charles Bardeen National Center for Atmospheric Research P.O. Box 3000, Boulder, CO 80307-3000 Phone: (303) 497-1752, Fax: (303) 497-1400 @.***
A quick peak at the AOD for WACCM MA F2000 MAM4 in the stratosphere for the fixed version with aq.chem changes, and CARMA (without aq.chem changes), and we still have to change this, since we still have an artificial Source, Chuck is testing this... MAM seems to perform similar to what we saw before, a little low in the stratosphere, for CARMA we still have to wait and see, but so far it is not too bad.
@dan-visioni Regarding the new setup for the MAM4 run, all we need to do is add some chemistry code, make sure the SSTs are correct and we have all the output we need. You find the chemistry code change in: /glade/p/cesm/chwg_dev/tilmes/cases/carma/FWmaMAM4HIST.f19_f19_mg17.carma_trop_strat.aqchem.1980_2014.001/SourceMods/src.cam/chemistry.F90 You have to add this to your case. Also, you can use a new IC file, from my new transient run I have been doing with the code change: /glade/scratch/tilmes/archive/FWmaMAM4HIST.f19_f19_mg17.carma_trop_strat.aqchem.1980_2014.001/rest/2000-01-01-00000/FWmaMAM4HIST.f19_f19_mg17.carma_trop_strat.aqchem.1980_2014.001.cam.i.2000-01-01-00000.nc
Thanks Simone
Thank you, I'll be starting it today
New MAM simulations are ok. Looking at the output and what Christina needed (she is plotting tropospheric SAD), I took a look at SAD_AERO vs SAD_SULF: is it normal that SAD_SULF goes to 0 in the troposphere? Anyone has any thoughts or guidance that I can pass on? Because other models do not have 0 SAD from sulfate in the troposphere.
There are several different SAD fields ...
call addfld( 'SAD_STRAT', (/ 'lev' /), 'I', 'cm2/cm3', 'stratospheric aerosol SAD' )
call addfld( 'SAD_SULFC', (/ 'lev' /), 'I', 'cm2/cm3', 'chemical sulfate aerosol SAD' )
call addfld( 'SAD_SAGE', (/ 'lev' /), 'I', 'cm2/cm3', 'SAGE sulfate aerosol SAD' )
call addfld( 'SAD_LNAT', (/ 'lev' /), 'I', 'cm2/cm3', 'large-mode NAT aerosol SAD' )
call addfld( 'SAD_ICE', (/ 'lev' /), 'I', 'cm2/cm3', 'water-ice aerosol SAD' )
call addfld( 'SAD_TROP', (/ 'lev' /), 'I', 'cm2/cm3', 'tropospheric aerosol SAD' )
call addfld( 'SAD_AERO', (/ 'lev' /), 'I', 'cm2/cm3', 'aerosol surface area density' )
Have you looked at SAD_STRAT and SAD_TROP? I think that SAD_STRAT, SAD_SULFC, SAD_LNAT, and SAD_ICE are only defined in the stratosphere. They are output right after the call to sad_strat_calc().
SAD_TROP is output after the call to usrrxt() and SAD_AERO looks to be the sum of the tropospheric and stratospheric surface area.
call usrrxt( state, reaction_rates, tfld, ion_temp_fld, ele_temp_fld, invariants, h2ovmr, &
pmid, invariants(:,:,indexm), sulfate, mmr, relhum, strato_sad, &
troplevchem, dlats, ncol, sad_trop, reff, cwat, mbar, pbuf )
call outfld( 'SAD_TROP', sad_trop(:ncol,:), ncol, lchnk )
! Add trop/strat components of SAD for output
sad_trop(:ncol,:)=sad_trop(:ncol,:)+strato_sad(:ncol,:)
call outfld( 'SAD_AERO', sad_trop(:ncol,:), ncol, lchnk )
NOTE: strato_sad is the variable that is output above as SAD_STRAT. So I think you just need to look at SAD_TROP.
Created run for BG MAM4
Used same compset (except MAM instead of CARMA) ./create_newcase --compset FWmaHIST --res f19_f19_mg17 --project P19010000 --run-unsupported --case /glade/work/visioni/cases/ISAMIP_CESM2MAM4_WACCMMA2000climo --project P19010000 --driver mct
The user_nl_cam is copied from @tilmes /glade/p/cesm/chwg_dev/tilmes/cases/carma/FWmaMAM4HIST.f19_f19_mg17.carma_trop_strat09.1980_2014
Need to check @zywshoon namelist for changes to cyclical conditions (?)