Running simulations for 2020 to 2030

[rentianyang@nus_edu_sg]

Hello, could someone educate me on how to set up my experiment? I want to examine the effects of emissions in stratosphere on ozone in the year 2020 to 2030.  How should I set the sst data? What other data do I need? How should I approach this simulation?(which part of the codes shall I modify?) Thank you so much! RegardsTianyang

 

[santos]

Hi Tianyang. I'm assuming that you're using WACCM, so I've moved this to the WACCM science forum. Response time might be a little slow due to the CESM workshop this week.I don't know about stratospheric ozone, but I can partially your question about SSTs. In a CESM case, the SST data is set in the file env_run.xml. The data file and dates are set by the variables SSTICE_DATA_FILENAME, SSTICE_YEAR_ALIGN, SSTICE_YEAR_START and SSTICE_YEAR_END.

 

[mmills]

Tianyag,Have you successfully run CESM before? If not, the place to start is the CESM 1.2.0 public release page. Once you have registered for access, you can download the code. Then you should start by trying to create a CAM case, build it , and run it before moving on to WACCM. You can get help from the User Guide, and the CESM and CAM sections of this forum. If you are porting the code to a new machine, there may be significant software engineering issues. There is much more support for CAM than WACCM, so it is easier to iron out the issues running CAM first.If you have done this already, please let us know what steps you have taken. I can help you choose the appropriate component set for your goals, and modify the input data for your experiments.Regards,Mike MillsWACCM Liaison

 

[rentianyang@nus_edu_sg]

Dear Mike,Yes, I have ported CESM1.1 to our machine, and have run a few cases with it. f45_g37 with compset A; f45_g37 with compset F_2000_WACCM_SC; f19_g16 with compset FAMIPC5 (all for 1 year of simulation) The runs were successful. Could you guide me on how to proceed further? Thank you.RegardsTianyang

 

[mmills]

Okay, great. For your information, we have a number of tutorials and a quickstart guide on our original WACCM working group web page, but I see now that that page is forwarding to a new version that doesn't have any of that information. If your browser forwards you, try the "Back" button to get back to the original site. I will have try to get the new site updated with the information from the old site. You can access the quickstart guide directly here.  The three standard resolutions we support for CESM(WACCM) are:

• 1.9°x2.5° with full ocean (B-compsets): use f19_g16
• 1.9°x2.5° with data ocean (F-compsets): use f19_f19
• 4°x5° with data ocean (F-compsets): use f45_f45

In your case, since you want to simulate the years that are later than 2005, you will need to use one of our three RCP compsets: RCP2.6, RCP4.5, or RCP8.5. These are all very similar for the years 2010-2020, as they diverge later in the 21st Century. RCP4.5 is the baseline scenario, so you would pick B_RCP4.5_WACCM_CN. Note that all of our WACCM RCP compsets have a full ocean (B-compsets). We do not have any data ocean (F) compsets set up for the RCPs. If that is something you require, it could be done with more setup.Otherwise, go ahead and create a new case using the B_RCP4.5_WACCM_CN compset, at the f19_g16 resolution. Then build the model out of the box and do a 5-day test run. After that, I can help you with modifying emissions files.What type of emissions do you want to modify? If it is a standard gas-phase species, it can likely be added somewhat easily. If you want to add an aerosol such as black carbon or sulfate, you may want to try to use CARMA microphysics. There are 2 new WACCM-CARMA compsets in CESM 1.2.0: one for black carbon in the stratosphere in an RCP4.5 scenario, and the other for stratospheric sulfates nudged by specified dynamics with a data ocean.  

 

[rentianyang@nus_edu_sg]

Dear Mike,I'm currently using CESM1.1, but the CESM website says that CESM1.1 contains no scientifically validated component sets. Will that significantly affect the simulation? Shall I switch to other version?I think B compset is fine. But just for my understanding, B compset will be computationally more expensive than F, because it involves full calculation for ocean? How may the results differ for using B and F?As for the emissions, I'll be adding in some gas phase species and also black carbon.Thank you.

 

[santos]

CESM 1.1 had some serious bugs that you probably don't want to deal with. You should switch to CESM 1.1.1, which is very similar, but has those bugs fixed.

 

[mmills]

Tianyang,Black carbon is prescribed in WACCM4. In order to study the impacts of black carbon emissions in the stratosphere, therefore, I would recommend that you use the new BNUKE_C4WBC_L40CN compset that is included in CESM1.2.0. This includes CARMA sectional (bin) microphysics set up for studies of black carbon emissions in the stratopshere. I have used this WACCM-CARMA model for studies of nuclear winter and emissions from space tourism:Mills, M. J., O. B. Toon, R. Turco, D. E. Kinnison, and R. R. Garcia, Catastrophic ozone loss following a regional nuclear conflict, Proc. Nat. Acad. Sci., 105, 5307-5312, doi:10.1073/pnas.0710058105, 2008.Ross, M., M. Mills, and D. Toohey, Potential climate impact of black carbon emitted by rockets, Geophys. Res. Lett., 37, L24810, doi:10.1029/2010GL044548, 2010.The compset is already set up for an RCP4.5 scenario, so this is good for your 2010-2020 study. The CARMA model includes only one size bin for black carbon (with radius 0.1 µm). You could change this to allow for evolution of the size distribution by coagulation, though the coagulation in the atmosphere likely forms chains and sheets in a complex fractal manner.As for your concern about using scientifically validated compsets, the lack of these for WACCM in 1.2.0 just means that we have not yet had time and resources to run long climate runs for any compsets and compare them to our CMIP5 runs to validate the climate. We have not yet validated any of our WACCM RCP compsets in 1.1.1 either, so that should not be a concern for moving to 1.2.0. In your case, you will be doing a sensitivity study, where you will be comparing to control runs that you will do with the same model, so comparison to our CMIP5 runs should be less of a concern.The inclusion of a full ocean will add some computational cost (perhaps 40-50%), but will allow you to include feedback to surface temperatures, which are dominated by the ocean.Cheers,Mike

 

[rentianyang@nus_edu_sg]

Thank you Santos and Mike for your kind replies. I got the following message while I try to build the case:ccsm_prestage ERROR: /home/students/atmod/cesm1_1/inputdata/ccsm4_init/b40.1955-2005.2deg.wcm.002/2005-01-01 is not on local disk
obtain this data from the svn input data repository:
  > mkdir -p /home/students/atmod/cesm1_1/inputdata/ccsm4_init/b40.1955-2005.2deg.wcm.002/2005-01-01
  > cd /home/students/atmod/cesm1_1/inputdata/ccsm4_init/b40.1955-2005.2deg.wcm.002/2005-01-01
  > cd ..
  > svn export --force https://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/ccsm4_init/b40.1955-2005.2deg.wcm.002/2005-01-01
or set GET_REFCASE to FALSE in env_run.xml,
   and prestage the restart data to /home/students/atmod/cesm1_1/stratosphere.001/run manuallyWhy did this happen? I thought svn will automatically download all the missing files.  What is GET_REFCASE for? What will be altered if the value for this variable is changed to FALSE? Thank you.Tianyang 

 

[mmills]

GET_REFCASE is the flag that tells the build scripts whether to move the restart data from their default location on disk ($DIN_LOC_ROOT/ccsm4_init/$case) to your run directory. You are getting this error because the restart directory does not exist on your local disk under /home/students/atmod/cesm1_1/inputdata/ccsm4_init/b40.1955-2005.2deg.wcm.002/2005-01-01You can follow the commands in your output to force the retreival of this data from the subversion repository to your local disk before rebuilding.Did you decide not to use the CARMA BC compset (BNUKE_C4WBC_L40CN), then? It has a different refcase, and a start date of 2013-01-01. It looks like you are using the regular RCP4.5 WACCM4 compset. If you do want to use the CARMA BC compset, you can move the start date back from 2013. Then you can use one of our CMIP5 RCP4.5 runs as your refcase, and I can get you the restart data.

 

[rentianyang@nus_edu_sg]

Dear Mike,I've ported CESM1.2.0 and created a case with:create_newcase -case waccmcase.001 -res f19_g16 -compset BNUKE_C4WBC_L40CN -mach catalystI have run it as a startup run for 5 days and the simulation completed successfully. How shall I proceed?By the way, while I executed .build for this case, it was actually looking for a file that is not in the svn repository. ( it was looking for b40.1955-2005.2deg.wcm.002.cam.i.2005-01-01-00000.nc,  but only b40.1955-2005.2deg.wcm.001.cam.i.2005-01-01-00000.nc existed. So I downloaded the wcm.001.cam.i data and renamed the file. Was there supposed to be a wcm.002.cam.i file somewhere?) RegardsTianyang

 

[mmills]

The IC file you are looking for is available in the restart directory in the repository here:https://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/ccsm4_init/b40.1955-2005.2deg.wcm.002/2005-01-01/Where did you find the 001 file? Do not use it; it may be from a buggy run.You can add emissions files for gas-phase species by modifying the namelist, using a user_nl_cam file, as described in the user guide and tutorials.Adding emissions for black carbon may require more coding. It is a CARMA species called SOOT01.

 

[rentianyang@nus_edu_sg]

Dear Mike,I tried to create a new case with the same command, somehow it was not looking for that particular nc this time, even though I deleted the renamed one from my local disk.I want to run a control simulation before I modify the code with my emissions, shall I run it as a startup run or hybrid/branch run? When I set up the RUN_TYPE as 'startup', it builds okay. But if i set it as 'hybrid' (which is the default I guess), it was trying to get a refcase 'b40.rcp4_5.2deg.wcm.carma.bc0tg.001', which again I couldn't find from the svn repository. 
RegardsTianyang 

 

[mmills]

I'm sorry for the delay. We will put the restart directory on the inputdata repository shortly.

 

[santos]

Sorry; this is my fault for forgetting. This data is now in the repository.

 

[santos]

Mike noticed that there is a mistake in this refcase. It should really be "b40.rcp4_5.2deg.wcm.carma.bc0tg.002", which is a different set of data. You can change this in env_run, or just run the following command in any case.xmlchange RUN_REFCASE=b40.rcp4_5.2deg.wcm.carma.bc0tg.002By default that should produce an appropriate control run for this case.

 

[rentianyang@nus_edu_sg]

Thank you Santos. Could you educate me on how I should set the run_type, run_refdate values in the env_run.xml file with this refcase? (will it be a hybrid or branch run? I couldn't really understand the explaination from the tutorial slides) Are there any other parameters I need to change?Without the refcase, I actually started the simulation with a 'start-up' run and then changed to hybrid after 5 days. The model did run and generated some results. What data was used to generate results? How valid are the results? And for such a case with compset BNUKE_C4WBC_L40CN, how long is expected for a 10-year-run?Thank you.  RegardsTianyang

 

[santos]

This compset is set up to do a hybrid run starting in January 2013 by default, and starts from the same data that Mike Mills used (except that by default this will be a control, with no injection of material to produce a nuclear winter).If you create a case and look for RUN_TYPE and RUN_REFDATE in env_run.xml, you should be able to see the quoted values, and you can change them with any text editor. Alternatively, you can use the "xmlchange" command. For instance, "xmlchange STOP_N=10" will set the variable "STOP_N" to the value "10" in env_run.xmlMike might know more about the other details than I do; this is a relatively new compset, and I've only run short tests myself.

 

[mmills]

This case should be kept as a hybrid run. That will give you the spun-up ocean and sea ice for 2013 that comes with the restart files. Running as a startup will give you a very different ocean and climate. So you should not continue the run you did as a startup. Instead, start over with a hybrid run and CONTINUE_RUN set to FALSE for your initial run.The time for your run will depend on how many processors you use, and how fast they are. You can look at the files in $casedir/timing (created after your run completes), which will tell you what the model throughput is. I don't think you are ready to run this for 10 years, as you have not added any BC emissions.

 

[rentianyang@nus_edu_sg]

Hi Mike,I'm intending to study the introduced emission effects in year 2020 to 2030. I believe that I need a control run (meaning without introduction of emissions) and then an experimental run(with emissions), and then compare the results. So shouldn't I for the first step run the simulation without any namelist modification from 2013 to 2030, and then with modification for the emissions start another run for 2013 to 2030?  Or do you mean the control data is already available somewhere so that I don't really need to run it again myself?