Drive BRCP45WCN by BRCP45CN result

[phou@mtu_edu]

Hi all,
I want to run BRCP45WCN f19_g16 in 2050. I currently have some restart files from BRCP45CN f09_g16 in 2045. I want to regrid it, so I can get the correct restart files for a hybrid run. But I'm not sure 1) if the restart files from BRCP45CN can be used in BRCP45WCN compset; and 2) how to regrid all those restart files.
Thanks so much for your help!

Best regards,
Pei

 

[mmills]

Dear Pei,I would be happy to provide you with a restart directory from one of our WACCM CMIP5 RCP4.5 runs for you to use as a starting point. Please let me know whether you have access to NCAR's Yellowstone computer. If not, I will put them on an anonymous FTP server for you. Is 2050 the year you would like?

 

[phou@mtu_edu]

Dear Mike,Thank you so much! Yes, 2050 is the year I want, I want to make a comparison for the result of 2010 and 2050. But I figured out when I run a hybrid run, there're two or three years unbalance. So maybe I should use a restart file before 2050, such as 2046? I can't access Yellowstone computer, I used svn to download files from this site https://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/ccsm4_init/.And I'm still curious about how to build the restart files from another resolution. It seems I can use the tool interpinic to interpolate restart for lnd part, but what about other parts? Thanks again!Best regards,Pei

 

[mmills]

I can get you 2046 tomorrow. What do you mean by "unbalanced" for 2 or 3 years? There should be no such issues with a hybrid restart. You could use interpic to interpolate IC files to different vertical resolutions. However, because the top of CAM is ~50 km, while WACCM goes up to ~145 km, you would be missing the whole upper atmosphere by starting from a CAM file. Thus it is much better to start from a WACCM restart directory. In addition, there are differences in the sea ice thickness between the two models due to climate sensitivity issues.

 

[phou@mtu_edu]

Dear Mike,Thank you, now I understand that I should use interpic for regrid of all restart files, but not in this case.In my attachment (startup_hybrid.jpg), the red line is a startup run, the purple one is a hybrid one. Both of them showed the surface temperature, I knew that a startup run takes a few years to equilibrate, so I thought that the first year of the hybrid result did not seem to be correct was the same reason. Best regards,Pei 

 

[mmills]

Pei,In your plot, it looks like the red line (your startup run) is taking longer to equilibrate than the purple line (hybrid). I suspect your ocean is not equilibrated in your startup run. The deep ocean takes thousands of years to equilibrate, so when running with a full ocean you should start from a spun-up ocean and use a hybrid restart.I have put restart directories for 2046 and 2050 on anonymous ftp for you at the link below. You will want to copy all of the files in the restart directory to your run directory, and set your RUN_REFCASE to b40.rcp4_5.2deg.wcm.002, and RUN_REFDATE to either 20460101 or 20500101 depending which restart you are using. Be sure to use a hybrid restart.ftp://acd.ucar.edu/user/mmills/inputdata/ccsm4_init/b40.rcp4_5.2deg.wcm.002/ 

 

[phou@mtu_edu]

 Dear Mike,Thank you so much! I got the files in ftp and made the model run. Is b40.rcp4_5.2deg.wcm.002 the same case with b40.1955-2005.2deg.wcm.002 (https://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/ccsm4_init/b40.1955-2005.2deg.wcm.002/). Sorry that I forgot to mention I plan to compare the result of 2010 and 2050, so I was about to use this restart file for my 2010 run.I'm still a little confused about the hybrid restart, it seems that I should use 2050's restart file for a hybrid run if I want result for 2050 run. But in my plot, hybrid run seemed need a year or two for equilibrate, so does this mean that I can only get 2051's result if I use a restart file for 2050? Or did I make a mistake when I ran the hybrid run? I used exactly the same steps as you said.Best regards,Pei

 

[mmills]

b40.rcp4_5.2deg.wcm.002 and b40.1955-2005.2deg.wcm.002 were both from our ensemble of CMIP5 runs with the same forcings. As it happens, b40.rcp4_5.2deg.wcm.002 was started with the restart conditions from b40.1955-2005.2deg.wcm.003, but that should not matter. Climate variability is such that the differences between ensemble members will be independent of initial conditions within a year or two.I would not expect the kind of spinup you show in TS for a hybrid restart from spun-up conditions. Without being able to look at your setup, I can't say what is going on there. Did you validate the climate for 1850 on your machine by comparing to our CMIP5 output?

 

[phou@mtu_edu]

Dear Mike,Sorry, I'm not sure how I can compare my climate for 1850 to your CMIP5 output. All my hybrid runs worked like the plot I showed you. Such as I use "./create_newcase -case name -compset BRCP45CN -res f09_g16 -mach Superior" to create a case and only modify the STOP_OPTION from 'ndays' to 'nyears'. I wonder what kind of setting might cause this problem. I've been bothered by this problem for quite a long time. The only solution I found was to run several years more before my target year. Thank you very much for your time, patience and useful information!Best regards,Pei

 

[mmills]

Are your plots showing global average TS? If so, it is about 10K too cold for 1850, which should be in the 285-290K range. I have attached a plot of monthly global average TS for 200 years from our CMIP5 control run.Instructions for validating CESM1(WACCM) on your machine can be found on our WAWG Code Release & Tools page, which is linked from our Whole Atmosphere Working Group page. It has links to the AMWG diagnostics page, which you can use to look at the climate of your runs. This code will generate diagnostic output in web pages, such as those linked on our CESM 1.0 experiments and diagnostics page. Clicking on the "Atm" link for our CESM1 (WACCM) 2° Pre-Industrial Control case will take you to the atmosphere diagnostic output for that case.I see now that you have set up your case using the f09_g16 resolution. The only resolution supported for WACCM B-compsets is f19_g16. This is the reason why your model is not running correctly.

 

[phou@mtu_edu]

Dear Mike,It might not be the problem of resolution. Because when I use b40.1955-2005.2deg.wcm.002 for a '-compset BRCP45WCN -res f19_g16' case, I got 276.4K, 278.5K, 278.3K as result for 2005-2007 global TS. The first year and second year has a 2 degree gap. Maybe I should not use v1.2.0, I should use v1.0.5 to make sure the model work with a scientifically validated resolution and compset. If that's the case, it's such a bad news that I need to redo all my work ... ... Thank you! I really learn a lot!Best regards,Pei 

 

[mmills]

It is certainly true that if you wish to publish scientific results you should either use v1.0.5 with scientifically validated configurations, or validate that the version you are using has the same climate as our CMIP5 experiments. However, the global average temperatures you are reporting for 2005-2007 are too far off to be due to using a different released model version. What are you using for the RUN_REFCASE for that compset?Again, I think you need to step back and validate your port of the model for the B_1850_WACCM_CN compset at 1.9x2.5_gx1v6 (f19_g16) by comparing the climate you get to that we get in our CMIP5 control. If you plan to publish your results, you might want to switch to CESM1.0.5 first so you won't have to validate each configuration you use after that.

 

[phou@mtu_edu]

Dear Mike,I used 'RUN_REFCASE= b40.1955-2005.2deg.wcm.002' and 'RUN_REFDATE=2005-01-01' for the 2005-2007 case. I'll switch to CESM1.0.5 and make the same test to check the difference. Thank you! Best regards,Pei