GEMCLIM model output

All model data output is written in RPN standard files.

There are 3 types of model data output:

"normal" 2-D and 3-D output
Time series (station data)
Zonal means (rarely used)
Pilot files

See the variable dictionaries at the end of this page.

The model will also save these other files:

Listings
Restart files
Jobs

Sometimes the output files are archived in an CMC-archive.

2-D and 3-D output

There are three types of output files:

dm	:	dynamics on model levels
dp	:	dynamics on pressure levels
pm	:	physics on model levels

And there are three stages of model output:

Usual direct model output
Post processing (usually not kept as final model output)
Diagnostics

Direct model output
All fields requested in 'outcfg.out' are written into the three types of output files and stored on the machine and in the directory as specified in 'xfer' in the file 'configexp.dot.cfg'.

The general fields saved in recent runs can be found here.

Post processing
If 'strip_phy=1' in 'configexp.dot.cfg' the direct model output will be processed a bit further.

A few new variables will get calculated, i.e. vertical integrals.
The 3-D physics fields on model levels will be interpolated on pressure levels.

All fields from the three types of output files will get redistributed into two types of output files:

md	:	Fields on model levels
pr	:	Fields on pressure levels

Diagnostics
If 'diagnos=1' in 'configexp.dot.cfg' the output from the post processing will get further processed.
The full diagnostics output contains:

monthly means and variances (generally) calculated from (3 hourly) model output
selected 3 hourly fields (when available)
full 12 hourly model output

You find more information about the fields contained in the full diagnostic package here.
The output files will then get archived into three archives:

gemclimdm_${exp}_md.ca	:	12 hourly output on model levels
gemclimdm_${exp}_pr.ca	:	12 hourly output on pressure levels
diag_results_${exp}.ca	:	monthly means, variances, covariances, selected 3 hourly fields

These archives are stored in 'machine:directory' as specified by 'arch_mach' and 'archdir' in the file 'configexp.dot.cfg'.

Variable dictionaries

At the beginning of each model listing you find a list of all fields from the physics which are available for output. You will find there the name of each variable as used in the code and the name as used in the FST output files, as well as a brief description and its position and length in the internal model data bus. Note that this list varies depending on the schemes used! Here is an example of the physics fields available for output when using ISBA.

Again in the model listing and just after the physics variable list, there is a list of the dynamic fields available for output, but unfortunately without description. Here is a list with the most common dynamic variables:

TT	Air temperature	C
UU	u-component of the wind (along the grid x-axis)	knots
VV	v-component of the wind (along the grid y-axis)	knots
UV	Wind modulus	knots
GZ	Geopotential height	dam
HU	Specific humidity	kg/kg
HR	Relative humidity	%
ES	Dew point depression	C
P0	Surface pressure	hPa
PN	Sea level pressure	hPa
ME	Mountain height	m
LA	Geographical latitude	deg
LO	Geographical longitude	deg

The full diagnostics web page also contains a description of the variables that will be saved when that mode is requested (diagnos=1 in your configexp.dot.cfg).

Please note that there also is a tool called r.dict that can be used to consult the official CMC-RPN variable dictionary. But be careful, the physic variables are not always correct!!! Usage:

r.dict -n variable_name

Pilot files

Pilot files are files which can be used to pilot a LAM run.
They can be saved during any model run to pilot another LAM run with them later.

Listings

At the end of each job the listings from the entry, the model and all scripts get zipped and transfered to arch_mach in archdir as specified in your 'configexp.dot.cfg'.

Restart files

At the end of each job, a restart gets written. This contains the physics and dynamics files needed to restart the model on each tile. These files contains a copy of the atmosphere and soil variables at the last time step of the job. When the next month gets started it reads the restart files of the previous month and continues from there. After the new month has started, the restart files from the previous month get archived on arch_mach in archdir as specified in your 'configexp.dot.cfg'. It is always possible to restart the model at any given month from a restart file.

Since the restart files contain a full copy of the model atmosphere and soil they are quite large. So usually we do not keep all of them but erase some by hand. The restart files get archived in an cmc-archive and gzipped afterwards. The name of a restart file look like this:

${exp}step#.ca.gz

If a restart file would get larger than 2 GByte the restarts for each tile will get saved separately:

${exp}step#_xx-yy.ca.gz

When a run is finished, the last restart file also gets archived, but a copy of it stays in the directory in which we were running the model (~gemclim/${mach}).

Jobs

During a run, several jobs get created and submitted automatically. A copy of the most important jobs gets saved, zipped and transfered to arch_mach in archdir as specified in your 'configexp.dot.cfg'.

Author: Katja Winger
Last update: June 2008