Frequently Asked Questions

Grids and Model Resolutions

What grids and model resolutions are scientifically validated in CCSM3?

The CCSM3.0 release officially supports three resolutions referred to as "T85_gx1v3", "T42_gx1v3", and "T31_gx3v5". For each resolution, multi-century control runs using all active models (a "B configuration") have been scientifically validated. See the Expermements and Data section for related output data. In addition to the fully active (B) configuration, a number of other configurations are available for scientific testing and research.

What are the "gx1v3" and "gx3v5" resolutions?

The CCSM3 ocean and ice models share the identical horizontal grid, two of which are presently officially supported in CCSM: gx1v3 and gx3v5. In both of these grids, the north pole has been displaced into Greenland. The former is the finer of the two, with longitudinal resolution approximately one degree. The latitudinal resolution is variable, with finer resolution near the equator (approximately 0.3 degrees). The latter is the coarser grid, with longitudinal resolution of 3.6 degrees. The latitudinal resolution is variable, with finer resolution near the equator (approximately 0.9 degrees).

In the ocean model, there are 40 levels in the vertical associated with the gx1v3 resolution, with level thickness monotonically increasing from approximately 10 to 250 meters. In the ocean model, it is the combination of the horizontal grid, horizontal land mask, vertical grid, and bottom topography that collectively define the "gx1v3" resolution.

There are 25 ocean-model levels in the vertical associated with the gx3v5 resolution, with level thickness monotonically increasing from approximately 12 to 450 meters. The combination of the horizontal grid, horizontal land mask, vertical grid, and bottom topography is referred to collectively as the "gx3v5" resolution in the ocean model.

What are the "T31" and "T42" resolutions?

The CCSM3 atmosphere and land models share the identical horizontal grid. T85 is a 256 by 128 regular longitude/latitude global horizontal grid (approximately 1.4 degree resolution). T85 indicates the truncation level in spectral space. The T42 is a similar grid by has 128 by 64 horizontal grids cells (approximately 2.8 degree resolution) The T31 is a similar grid by has 96 by 48 horizontal grids cells (approximately 3.75 degree resolution)

How many vertical levels does the atmosphere model have?

First off, the term "vertical levels" is a paradox and makes no sense, you're probably asking about the number of levels in the vertical. To answer your question, the atmosphere model in CCSM3 has 26 levels in the vertical for both T42 and T31 horizontal resolutions.

What are the different CCSM model components? How can they be configured?

Section 1.2 of the CCSM3 User's Guide (CCSM Model Components) answers this question.

Which resolutions and configurations are generally supported?

Section 1.2 of the CCSM3 User's Guide (CCSM Model Components) answers this question.

 

CCSM Model Restarts

How do I get the various component models to all write monthly time-averaged history files?

By default, all the components write monthly time-averaged history files. The history files can be modified for each component individually, although the data model components do not have history file writing capability. History files are not coordinated between each component (unlike restart files which are coordinated through the coupler). Please see the individual component model documentation for more information about how to change the components' history output.

How can I get model restart files?

The input data that we recommend for starting model runs is available from input data tar files found the "Model Input Data" section of the CCSM3.0 home page. It is possible to set up branch or hybrid runs that use data other than that which we recommend and provide. See section 5.4.3 of the CCSM3.0 User's Guide ("Prestaging input data for hybrid and branch runs") for information about how to set up such runs. Restart data from the CCSM3.0 control experiments is archived on NCAR's Mass Storage System (MSS) and is available to those who have login access to an appropriate NCAR computer.

 

Software

I don't have "gmake" on my system. Where can I get a copy?

The CCSM build process assumes that gmake is currently installed on your system. If it isn't already available, there are a number of steps that you can take to resolve this.

First, ask your system administrator if you already have gmake (the GNU make utility). If so, you may need to change your unix PATH definition to include the path to your copy of gmake.

Second, the gmake program is often called "make" or "gnumake". Issue the command "make -v" the "make" you are using is actually gnu make. You should see something like:

make -v
GNU Make version 3.76.1, by Richard Stallman and Roland McGrath.

If your version of make doesn't recognize the "-v" open, then it is not the required gnumake utility, and you will see the following error message:

make -v
make: Not a recognized flag: v

Finally, if you do not have the gnumake utility anywhere on your system, you can build and install gmake yourself. The gmake utility can be obtained off the web from:

ftp://ftp.gnu.org/gnu/make/

The gmake utility can also be obtained via anonymous ftp using the following commands:

ftp ftp.gnu.org
anonymous
qian@sca.uqam.ca 
cd gnu/make
binary
get make-3.79.tar.gz 

 

Computer Resources

What about model performance? How do I estimate how long my job will take to run one model year?

This question has no easy answer. There are a number of considerations to take into account when setting up other configurations or running on other machines. The model performance and efficiency are dependent on the processor layout, resolution, configuration (active or data models), hardware performance, and physics options. In addition, both the active atmosphere and land components can run either MPI, OpenMP or hybrid mode. A number of timers are printed in the log files to help determine optimal configurations. If you need help determining optimal processor configurations for productions runs, please contact cesm.ucar.edu for help on your specific problem.

To give you some idea of the performace one might expect, consider:

  • The fully active T85_gx1v3 b30.009 control run got about 4.5 years per wall clock day in the default configuration on 192 processors on the NCAR IBM-SP Power 4 bluesky machine.
  • The fully active T42_gx1v3 b30.004 control run got about 4 years per wall clock day in the default configuration on 104 processors on the NCAR IBM-SP Power 3 blackforest machine.
  • The fully active T31_gx3v5 b30.031 control run got about 24 years per wall clock day in the default configuration on 64 processors on the NCAR IBM-SP Power 4 bluesky machine.

 

Other

How can I get more help?

Please refer to the CESM Bulletin Board, which is in place to facilitate communication within the CESM community. Finally, please also refer to the Release Notes entries that are provided with every release and release update. If questions or problems still exist, then please send an email to cesm-help@cgd.ucar.edu. Support questions will be answered as resources are available.