Differences between the GEM and GEMCLIM (GEM in climate mode)

The big differences between climate and non-climate mode are 1) the automatic re-launching of a multi-month job, 2) the large amount of pre- and post-processing that will be done: automatic testing/setting of several of the model run-time variables, creation of part of the config files and production of a comprehensive set of diagnostics, following WCRP and AMIP2 guidelines. This implies that, if diagnostics are requested (strip_phy=1 and diagnos=1), the output variable/timesteps definition file 'outcfg.out' will be created automatically. In this case the output interval is determined by the parameter 'MOYHR' in your 'gemclim_settings.nml'. But it is still possible to add extra output variables to it.

The model output is written directly into the directory ${EXECDIR}/output/current_last_step/tile instead of in ${EXECDIR}/${REPRUN}/tile to avoid unnecessary data moving, where 00-00 would denote the first tile. The 16-bit turbo compression algorithm is turned ON by default.

In climate mode the zonal-mean O3 climatology suggested by the AMIP2 experimental protocol is used. If you want to use another O3 climatology you need to call this file 'ozoclim' and place it in the directory from which you launch the model. You will also have to provide appropriate litozon2.ftn and ozopnt.cdk source files and re-generate a new model executable with them.

Basically everything that was before called 'gem' or 'gemdm' is now called 'gemclim'. See
running GEMCLIM short version.

The physic tendencies and accumulators get reset to zero each time they have been written as requested in the file 'outcfg.out'. This can cause problems when outputting time series. Therefore, if an accumulator is requested for time series output it also needs to
requested for in the file 'outcfg.out' at the same frequency.

The accumulated variables that will be reset can be found in the physics source file 'acclist.cdk'.

The pressure level under which there are no clouds considered to be is raised from 70 hPa to 100 hPa (set in 'nocld.cdk').

The minimum wind strength is changed from 1.0E-04 m2/s2 to 6.25 m2/s2 when using the Force-Restore surface scheme. This is the same value that ISBA is using (set in 'vamin.cdk').

The moving/treatment of the model output can be started during the run at every multiple of number of days specified in 'climat_cleanup' in your
'configexp.dot.cfg'.

The time series will be written in version 2.00 format. The differences with the previous format of the time series which is still written in non-climate mode are:

In case of problems you might find help in our little trouble shooting.




Author: Katja Winger
Last update: May 2008