Initial Conditions and spin-up

[duderstadtk@gmail_com]

Hi,

I am currently running SD-WACCM (FSDW, cesm1_2_0, fixed ocean, f19_f19 resolution, 88 layers) on Yellowstone with a focus on winter 2004-2005.

I'd like to spin-up the model from January 2004. Do I need to use CAM initial conditions specific to January 2004 or can I use a "default" initial condition GEOS files on Yellowstone. For example:
ncdata='/glade/p/cesmdata/cseg/inputdata/atm/waccm/ic/waccm_geos5_2x_88L_2005-01-01_c110419.nc'

i.e., Are initial conditions files for 2004 available elsewhere (can't find one specific to 2004 on the svn input repository)? And will it make that much of a difference anyway, considering I plan to start looking at model history files in December of 2004?

On a related note...How are initial conditions files created? Can they be created using restart files from prior simulations? (I can't locate this information in the documentation).

Thanks!

 

[mmills]

Hi Kathy,I have put IC files for January 1, years 2000, 2001, 2002, 2003, and 2004 on yellowstone in:/glade/p/cesmdata/cseg/inputdata/atm/waccm/icThe files are named:wa4_cesm1_1_b02_geos5_2x_sim153f.cam2.i.200[01234]-01-01-00000.ncOf the SD runs we have available prior to 2005, this one has chemistry that is most similar that used in the release version that you are using. However, it has a number of tropospheric species that are not in the release version. Therefore, you should pick an IC that is about 4 years prior to any years that you wish to use, as the effects of different tropospheric chemistry may linger during the spinup period.

 

[mmills]

Hi Kathy,I have put IC files for January 1, years 2000, 2001, 2002, 2003, and 2004 on yellowstone in:/glade/p/cesmdata/cseg/inputdata/atm/waccm/icThe files are named:wa4_cesm1_1_b02_geos5_2x_sim153f.cam2.i.200[01234]-01-01-00000.ncOf the SD runs we have available prior to 2005, this one has chemistry that is most similar that used in the release version that you are using. However, it has a number of tropospheric species that are not in the release version. Therefore, you should pick an IC that is about 4 years prior to any years that you wish to use, as the effects of different tropospheric chemistry may linger during the spinup period.

 

[duderstadtk@gmail_com]

Thanks, Mike. A couple follow-up questions: 1. What is the most efficient strategy (least computationally expensive) for spinning-up the model for four years?  Are there recommended history variables (and recording frequencies) we should save along the way?  Can we use a larger timestep during spin-up than planned for final runs?  After the spin-up period we will branch several runs with 30 min timesteps, modified solar input files, and modified history files.   2. Will these initial conditions work for other versions of WACCM as well?  We are planning on running SD-WACCM for this same time period (2004-2005) for several different projects: A) Can we use these IC files for the "scientifically validated" version of WACCM (cesm1.0.5)?  If so, do we need to use a chemical mechanism input file that includes the additional tropospheric species (e.g., NMHC chemistry)?  [Is cesm1.1 on its way to being scientifically validated for WACCM/SD-WACCM?]   B) Will we be able to use these IC files to spin-up cesm1.2 runs?  We are interested in making use of current and future updates to WACCM (version cesm1.2) in order to include aerosols processes involved with the depostion of cosmogenic radionuclides (comparing SD-WACCM runs with surface measurements).  Thanks again for your help.  We're looking forward to beginning these modeling experiments! Kathy  

 

[duderstadtk@gmail_com]

Thanks, Mike. A couple follow-up questions: 1. What is the most efficient strategy (least computationally expensive) for spinning-up the model for four years?  Are there recommended history variables (and recording frequencies) we should save along the way?  Can we use a larger timestep during spin-up than planned for final runs?  After the spin-up period we will branch several runs with 30 min timesteps, modified solar input files, and modified history files.   2. Will these initial conditions work for other versions of WACCM as well?  We are planning on running SD-WACCM for this same time period (2004-2005) for several different projects: A) Can we use these IC files for the "scientifically validated" version of WACCM (cesm1.0.5)?  If so, do we need to use a chemical mechanism input file that includes the additional tropospheric species (e.g., NMHC chemistry)?  [Is cesm1.1 on its way to being scientifically validated for WACCM/SD-WACCM?]   B) Will we be able to use these IC files to spin-up cesm1.2 runs?  We are interested in making use of current and future updates to WACCM (version cesm1.2) in order to include aerosols processes involved with the depostion of cosmogenic radionuclides (comparing SD-WACCM runs with surface measurements).  Thanks again for your help.  We're looking forward to beginning these modeling experiments! Kathy  

 

[mmills]

Hi Kathy, 1. You really cannot change the model time step for your spinup due to the unfortunate fact that the cloud physics are timestep-dependent. Increasing the model time step could lead to a decrease in cloudiness that would affect the climate, and thus the model would not be in balance when you returned to the default 30-minute time step. There may be other problems with running WACCM at a longer time step that I am not aware of. I became aware of the cloud issue when I decreased the model time step and later discovered that I had significantly increased the clouds as a result.You might see some efficiency increases by decreasing model output during your spinup period. You could eliminate all but the h0 history tapes for the atmosphere, and reduce the number of variables output in the h0's. To eliminate the h1, h2, h3, and h4 output, you would specify fincl2, fincl3, fincl4 and fincl5 in your user_nl_cam to be two single quote marks with a single space in between, i.e.:fincl2 = ' 'fincl3 = ' 'fincl4 = ' 'fincl5 = ' 'For fincl1 (h0 output) you could try having only 'O3' and 'T' in your fincl list. You might want to add a few more species as you see appropriate.2. The "scientifically validated" WACCM compsets are all "free-running" and use 66 vertical levels, rather than SD, which uses 88 vertical levels. The ICs I have provided are suitable for other SD-WACCM runs starting at the same date with other versions of the model (CESM1.0.5, CESM1.1, CESM1.2). However, if you want to run the free-running version of WACCM, you should let me know what years and compsets you want ICs for, and I will be happy to provide them to you from our CMIP5 runs."Scientifically validated" is a term of art we use to describe selected compsets in model versions. It requires that a peer-reviewed paper describing the model and results be published. We currently have no scientifically validated SD-WACCM compsets, and have no plans to provide this level of validation in any forthcoming released version. SD-WACCM is currently being used in the Chemistry Climate Model Intercomparison (CCMI) project, but with developmental updates to chemistry and physics that will not be provided in any release until at least mid-2015. The free-running configurations that we used in our published CMIP5 runs will remain the only scientifically validated WACCM compsets for now.

 

[mmills]

Hi Kathy, 1. You really cannot change the model time step for your spinup due to the unfortunate fact that the cloud physics are timestep-dependent. Increasing the model time step could lead to a decrease in cloudiness that would affect the climate, and thus the model would not be in balance when you returned to the default 30-minute time step. There may be other problems with running WACCM at a longer time step that I am not aware of. I became aware of the cloud issue when I decreased the model time step and later discovered that I had significantly increased the clouds as a result.You might see some efficiency increases by decreasing model output during your spinup period. You could eliminate all but the h0 history tapes for the atmosphere, and reduce the number of variables output in the h0's. To eliminate the h1, h2, h3, and h4 output, you would specify fincl2, fincl3, fincl4 and fincl5 in your user_nl_cam to be two single quote marks with a single space in between, i.e.:fincl2 = ' 'fincl3 = ' 'fincl4 = ' 'fincl5 = ' 'For fincl1 (h0 output) you could try having only 'O3' and 'T' in your fincl list. You might want to add a few more species as you see appropriate.2. The "scientifically validated" WACCM compsets are all "free-running" and use 66 vertical levels, rather than SD, which uses 88 vertical levels. The ICs I have provided are suitable for other SD-WACCM runs starting at the same date with other versions of the model (CESM1.0.5, CESM1.1, CESM1.2). However, if you want to run the free-running version of WACCM, you should let me know what years and compsets you want ICs for, and I will be happy to provide them to you from our CMIP5 runs."Scientifically validated" is a term of art we use to describe selected compsets in model versions. It requires that a peer-reviewed paper describing the model and results be published. We currently have no scientifically validated SD-WACCM compsets, and have no plans to provide this level of validation in any forthcoming released version. SD-WACCM is currently being used in the Chemistry Climate Model Intercomparison (CCMI) project, but with developmental updates to chemistry and physics that will not be provided in any release until at least mid-2015. The free-running configurations that we used in our published CMIP5 runs will remain the only scientifically validated WACCM compsets for now.

 

[duderstadtk@gmail_com]

Mike, Thanks for the detailed answers to all the questions.   Helps immensely.  Kathy   

 

[duderstadtk@gmail_com]

Mike, Thanks for the detailed answers to all the questions.   Helps immensely.  Kathy