The NCAR Community Climate Model (CCM3)

Aug 5 1998: CCM3.6.6 Release, fix for spectral filtering of non-T42 cases.

This message should be of interest only to those using CCM3.6 at resolutions other than T42, or to those who wish to build the code on a PC running LINUX, or who wish to use the Slab Ocean Model (SOM).

A new tar file of CCM3 code, ccm3.6.6.tar.gz, is now available on NCAR's anonymous ftp site (ftp.ucar.edu). Changes vs. CCM3.6 are as follows:

• A bug was found in the spectral filter which is applied (subroutine SPETRU) when initial dataset field PHIS was defined from a high-resolution topography file. This would be true if the initial dataset has been run through the code in ccm/tools/definesurf). Since the initial dataset (T42 truncation) provided with the distribution was not run through definesurf, users should not encounter the bug unless they are running definesurf themselves.
• The code which builds SGH, PHIS, and ORO from a high-resolution topography file (definesurf/) was inadvertently missing a subroutine (endrun.F90) in the CCM3.6 distribution. This routine has been added in the new tar file.
• Code and Makefile mods were added to enable the CCM to be run on a PC running LINUX. Though the new cpp token is named LINUX, it is implicitly assumed that the compiler is pgf77 (Portland Group f77). CCM3.6 is not yet officially supported in this configuration, but the mods are provided to those who would like to make use of them. Initial tests show that the model is producing correct results as compared with supported platforms.
• The SOM code was added back into the distribution. CCM3.6 with SOM was tested on the Sun and Cray platforms, and it was verified that SPMD gives bit-for-bit results with non-SPMD when running under SOM. However, we have NOT run a climate simulation with SOM, and are not providing data-sets for SOM use with CCM3.6. And as we have not run a SOM climate simulation ourselves, we simply do not know if the climate simulation of CCM3.6 with SOM is reasonable.
• For runs which do not involve the spectral filter mentioned above, the new code produces an identical simulation to that of CCM3.6 on all the supported IEEE platforms (SGI, SUN). Unfortunately, the Cray simulation only matches to machine roundoff.