History files may be examined by various postprocessors. Examples include the "CCSM Component Model Processing Suite (CMPS)", NCAR Graphics package, FERRET, ncview, MATLAB, AVS, IDL, and Yorick. For a list of software tools for interacting with NetCDF files view the UNIDATA NetCDF web-site: http://www.unidata.ucar.edu/packages/NetCDF/software.html.
Up to six separate history files may be produced in a single run. Two history files will be produced by default. The first file contains monthly output whereas the second file contains daily output (this can be modified via the NHTFRQ namelist option). Some fields on the second tape are time-averaged and some are instantaneous. Table 2.7 lists the fields that can be output on any of the six history files as well as which fields are output by default on the first file. The list of fields output on the second file is given in Table 2.8.
The first column indicates whether the field will automatically be
included on the primary history file. Fields not on by default may be
included via the namelist variable FINCL1 (see
"Example 6 -- History File Specifications"
). Conversely, any default fields may be removed using
namelist variable FEXCL1 Some fields are active by
default only for a given model configuration (i.e. only for the CCSM
coupled model configuration) or when a given namelist setting is
chosen. The configurations outlined in the first column denote under
which conditions the given field is active. "CCSM-only" indicates that
the field is active only when the model is compiled with the CPP
token,
The second column lists the names of the output variables. Some field
names appearing in the second column contain two upper case letters
followed by two lower case x's (e.g. TRxx,
DCxx, etc.). These fields correspond to passive tracer
constituents that may be transported. In the history file, the two
x's will be replaced by the name of the given tracer (i.e. Q,
CWAT, CH4, N2O, TEST1, TEST2, TEST3, and in the case of advected or
non-advected tracers ADV or NAD followed by the tracer number).
The third column associates the history file field name with its
mathematical symbol as given in Scientific Description of the NCAR
Community Atmospheric Model (CAM2) (2002).
The fourth column provides a brief description of the field. In the
column labeled "NL" a "1" indicates a single-level field and an "N"
indicates a multilevel field (on plev vertical levels). The
default averaging flag is also given here, unless it is a time-average.
Since, most fields are time-averages by default we only describe the
ones that are not a time average. The flags available are: Instantaneous (I),
Average (A), Maximum (X), and Minimum (M). See Section 2.5.1.2
for more information on changing the default averaging flag for a given variable.
The last column in the table shows the physical units associated with
each field.
A namelist option is available to produce a
monthly-averaged history file for the first
file series. For example, by
setting NHTFRQ(1)=0, the fields on the second history file
will be output over the period beginning from the first timestep of
the current month up to and including the last timestep of that
month. Each monthly history file will contain exactly one time slice
of data, regardless of the value of MFILT(1). Note that
the option to output monthly datasets is only valid for the first
file-series. Also note that the type of averaging done is dependent
on the default average flag for the variable in question.
Fields with are new fields that have been
added since CAM2.0.
Table 2.7: Master Field List
FINCL1 = 'T:I'
This specifies that temperature is to be recorded as instantaneous
values on the first history file.
The frequency at which data is written to the history file is
specified by NHTFRQ(i). For
example, consider the namelist input shown in
Example 5. Here,
NHTFRQ(1)=72. Every 72 timesteps a time sample of data is
written to the primary history file, with each field being time
averaged over the previous 72 timesteps. If NHTFRQ(1) = 0,
then the data for the first history file will be written out on
monthly boundaries.
The number of time samples written to a single history file, primary
or auxiliary, is set via namelist variable
MFILT(i). The packing densities
of model history files is set with namelist variable variable
NDENS(i) which can have values of "1"
or "2". "1" implies output should be written out in 64-bit
NetCDF format, and "2" writes data out in 32-bit NetCDF format.
Refer to Table 2.7 in
Sec. 2.2 for a complete
description of all history file namelist variables.
The following table lists the fields that are output by default on the
secondary history files. By default secondary history files produce
daily output. This can be changed by modifying the value of the
namelist variable NHTFRQ(2). The table
lists the output averaging flag as well as the field name. The
averaging flags are: I for instantaneous, A for average, M for
minimum, and X for maximum. See
Section 2.5.1.1 for more information
on the averaging flag.
Table 2.8: Fields output by default on secondary history files (default
frequency is daily)
Upon restart, a simple ASCII text file (the "restart pointer file") is
read to obtain the full pathname of the most recently written master
restart file. Only the name of the master restart file is needed as
input for a continuation run. The other files needed for restart (such
as secondary restart files, or history files that need to be opened) are
also listed in the restart pointer file. The Master restart file itself
includes the full archive path to the files that actually need to be opened.
Master and primary restart files are always written during a model
run. A secondary restart file is written if
absorptivity/emissivity is not be calculated on the first
timestep after restart, and therefore must be saved on a restart
dataset. For a stand-alone run this occurs if the primary history file
write frequency, NHTFRQ(1), is not a multiple of the
absorptivity/emissivity calculation frequency, IRADAE (note
that for a CCSM flux coupled run, only the flux coupler determines
when the restart files are written). It is advisable to avoid this
situation if possible, since this dataset is relatively large even for
the standard T42 model.
A history buffer restart file is written in order to retain the
accumulated values in the history buffers if restart files are to be
written on a timestep when one or more history file time samples are
not written. A separate restart dataset is written for each
history file. Each history buffer restart file contains the portion of
the history buffers pertaining to that history file.
All restart files have names of the form
$CASE.cam2.r.yyyy-mm-dd-sssss, but with the ".r." changed to
the appropriate restart filename. For example, ".r." is for the master
restart filename, while , ".ra." corresponds to the
absorptivity/emissivity restart filename, and ".rh0." is the primary
history file restart filename. In the root name, $CASE
refers to the caseid, yyyy corresponds to the year (note more
than 4 digits for the year will be used if necessary), mm is
the month, dd is the day and sssss is the seconds of
the date yyyymmdd. The following is an example of the restart
files written during a simulation.
/$USERNAME/csm/$CASE/atm/hist
Restart files will be archived in the Mass Store directory
/$USERNAME/csm/$CASE/atm/rest.
And finally initial files will be archived in the Mass Store directory
/$USERNAME/csm/$CASE/atm/init.
$USERNAME is the upper-case equivalent of the user's login name,
i.e., the user's root directory on the Mass Store System, and
$CASE
is the case identifier and is set via the namelist input. It is recommended
that the user specify a non-blank write password using the namelist variable
MSS_WPASS
(see Table 2.7 ). File passwords are
the only form of security available on the Mass Storage System. If the
write password is not set, any other user can overwrite or change the files
after they have been archived..
Questions on these pages can be sent to...
erik@ucar.edu .
The table is divided into the following sections:
basic state variables,
derived variables,
water vapor and tracers,
horizontal diffusion,
precipitation,
surface variables,
planetary boundary layer fields,
convection, radiation,
clouds, and tendencies.
2.5.1.2 Changing characteristics of default output fields
There are several ways that namelist options can modify the
characteristics of the default output fields. Output fields can be
added or deleted from a file, the averaging flag can be changed and
the output frequency can be varied. In addition, the number of
time-samples on a file and the precision of the output data (double or
single NetCDF) can be changed. To add additional fields to the first
history file, the user should use the namelist variable
FINCL1.
FEXCL1 can be used to delete fields
that are on the default list of fields on the primary file. The
averaging flag may also be specified with the FINCL1 option and
determines how the data is averaged over the output frequency. Values
recorded for fields on a history file can be represented in one of
four different ways. Data may be time averaged since the last
write to the history file, instantaneous, or appear as a
point-by-point maximum or minimum over the time interval. The
representation may be specified in the namelist by including a colon
followed by the single character flag for each averaging type after
the field name. The characters are as follows: 'A' means
averaged over the interval, 'I' for instantaneous, 'M' for
point-by-point minimum, and 'X' for point-by-point maximum. An
example of this specification would be:
2.5.1.3 History Files 2 through 6
Up to six different types of history files may be written out by the
model during a model run. The capability to write additional history
files provides the user with the flexibility to vary the frequency at
which various history data are written. Additional files may contain
the same or different fields as compared with the first history
file. These fields may be written on different timesteps, and
have different averaging periods. Furthermore, each file may contain a
different number of time samples.
Field name Average flag (Average, Instantaneous, Maximum or Minimum)
FLNT Average
PRECT Average
PSL Instantaneous
TREFHT Average
TREFHTMN Minimum
TREFHTMX Maximum
TSMN Minimum
TSMX Maximum
OMEGA850 Average
OMEGA600 Average
QFLX Average
TS Average
TMQ Instantaneous
T850 Instantaneous
T300 Instantaneous
U850 Instantaneous
U200 Instantaneous
V850 Instantaneous
V200 Instantaneous
Z700 Instantaneous
Z500 Instantaneous
Z300 Instantaneous
2.5.1.4 Model generated initial condition dataset files
During a model simulation, initial condition datasets are generated
periodically by default. These datasets are simply auxiliary history
files containing instantaneous values for only those fields that are
required to begin an initial run. The naming convention for these
files (which is different for the other auxiliary files) is
$CASE.cam2.i.yyyy-mm-dd-sssss.nc, where $CASE is the
caseid, yyyy is the year (note, more than 4 digits will be
used if needed), mm is the month, dd is the day and
sssss is the seconds. The output frequency of the files is
controlled by namelist variable
INITHIST and is independent of the
output frequency of other auxiliary files.
2.5.1.5 Naming the History Files
History volumes will be named according to the history filename
specifier. The history filename
name specifier is documented in Table
2.7. By default if the first history file series contains monthly
output, history filenames will be of the form
caseid.cam2.h0.yyyy-mm.nc
where caseid, yyyy, and mm correspond to the case-name, current year,
and current month respectively. For example, if caseid="cambld", and
current date is September, 1989 the filename becomes
cambld.cam2.h0.1989-09.nc
Non-monthly file-series are named with a full date expression as
follows:
caseid.cam2.h#.yyyy-mm-dd-sssss.nc
Here, # is the file series number minus one, dd is the current day,
and sssss is the number of seconds into the current day. For example,
for the second file-series and a current date of September, 1, 1989,
0Z the filename becomes:
cambld.cam2.h1.1989-09-01-00000.nc
2.5.2 Restart Datasets
There are four types of restart datasets generated by the model:
master, primary, secondary, and history buffer restart files. Each
dataset is in binary format and contain grid-point data and other
information necessary to continue or branch a model run.
2.5.3. Mass Store Archiving
If history files and restart datasets are to be archived on the NCAR Mass
Storage System, they will be transferred asynchronously to the MSS as they
are completed. If namelist variable MSS_IRT is zero, history and restart files
will not be archived.
Mass Store pathnames for these transfers are
generated using the ARCHIVE_DIR namelist setting. By default, ARCHIVE_DIR
is set to /$USERNAME/csm/$CASE/atm/hist. As a result history files will be archived in the Mass Store
directory
Sub Sections
$Name: $ $Revision: 1.1 $ $Date: 2004/06/08 02:57:22 $ $Author: jmccaa $