|
![[NOAA Logo]](/nomads/forms/deccen/images/noaasmall.gif)
![[GFDL Logo]](/nomads/forms/deccen/images/largelogo.gif)
GFDL R30 COUPLED CLIMATE MODELS:
A Guide To Accessing R30 Model Output
Stored On GFDL's NOMADS Server
for DecCen Climate Research
GFDL R30 COUPLED MODEL OUTPUT OVERVIEW
![[NOAA bullet]](/nomads/forms/deccen/images/noaabullet.gif)
Introduction
The GFDL R30 climate model is a coupled atmosphere-ocean general circulation model
(AOGCM). Its four major components are an atmospheric spectral GCM, and ocean GCM,
and relatively simple models of sea ice and land surface processes.
The name "R30" is derived from the resolution of the atmospheric
spectral model (rhomboidal truncation at wave number 30).
This R30 model is identified as GFDL_R30_c in the Third Assessment Report of the
Intergovernmental Panel on Climate Change (IPCC).
See Chapter 9, Table 9.1 of "Climate Change 2001: The Scientific Basis" IPCC Working Group I (2001).
The model output stored on this server is taken from six experiments
conducted at GFDL using the GFDL_R30_c model. Data files produced by one
long-running control integration (one with no changes in external forcings,
e.g., constant CO2) and five climate change scenario experiments have
been made accessible to interested researchers. More information
regarding the individual experiments is found in the
Control & Transient
Forcing Experiments section of these web pages.
This numerical model was developed and the experiments conducted by members of the
Climate Dynamics & Prediction Group
at the Geophysical Fluid Dynamics Laboratory (GFDL)
located in Princeton, New Jersey.
GFDL is part of the National Oceanic and Atmospheric Administration's
(NOAA's)
Oceanic and Atmospheric Research (OAR).
File Format: NetCDF
The model output files produced by the GFDL R30 coupled model
are stored as NetCDF (network Common Data Form) files, and can be
identified by the suffix ".nc".
The files are intended to be compliant with
the COARDS
conventions for the standardization of NetCDF files.
More information about NetCDF, a machine-independent
format for representing array-oriented scientific data,
can be found at the Unidata
Program Center NetCDF website.
Directory Structure & Data Organization
Starting at
http://nomads.gfdl.noaa.gov/dods-data/DecCen/r30/coupled/
one will first see directories listed for each of the
six R30 experiments.
After selecting the experiment of interest, one then sees
that the data is arranged in "atm" and "ocn" directories. Output
from the atmospheric and land surface models are found under the "atm"
directory, while ocean and sea ice output is located under the "ocn"
directory.
Continuing down either the "atm" or "ocn" path, the user will
find the next level to be "ts", which stands for time series.
This refers to the fact that the data files stored on the server
contain multiple time levels of 2-D X-Y horizontal fields. For example,
a single file can contain a 140 year long time series of atmospheric
model temperatures for all latitude and longitude points at
sigma level 14 (for those unfamiliar with the use of sigma levels as
the vertical coordinate, a brief explanation is given in the
Atmospheric GCM
description on these web pages).
The data is arranged this time series form because we have found it to
be the one most commonly utilized for analysis purposes.
To reconstruct the full 3-D distribution of atmospheric temperatures
at a single point in time, one would need to extract a single time
level of information from 14 atmospheric time series files (one
for each model sigma level) and combine the 14 2-D fields into a
single 3-D data set.
Below the "ts" directory level, one will find "mon" and "dly" subdirectories.
Time series of monthly means are located under "mon".
Daily time series of a selected subset of variables are found
under the "dly" directory.
Progressing further, one encounters the "raw" directory, which simply
notes that the data are in their raw form (i.e., there has been
no detrending of the data). Finally, below the "raw" level, one
will find directories that identify the length of the netCDF
time series files found within.
Concerning the "daily" output: since the R30 coupled model does not simulate
the diurnal cycle the time of the day of the data dump from the model to disk is
not that important (for the curious, it occurs once a day at 0Z).
Iin the "daily" output files the time averaging of the field depends
on the variable of interest. Most of the fluxes (SW top, SW bot, sensible heat,
precipitation, etc.) are daily averages. All of the "state" variables
(temperature, wind, humidity, etc.) are instantaneous snapshots.
Calendar Used in Model Experiments
The GFDL R30 coupled model experiments utilize a 365 day calendar (i.e.,
there are no leap years in the model). The choice to ignore leap days
has been made in order to faciliate post-processing and analysis, as
a variety of procedures, from creating annual averages to spectral analyses
of model generated time series, are made more straightforward
when each year contains the same number of days.
Since the GFDL Dec-Cen climate model simulations are not designed to
replicate the specific interannual variability or weather events
of individual years, the simplification of ignoring leap years
is warranted.
The "model year" counter starts at year 1 when the atmosphere-land surface
model component is coupled to the ocean-sea ice component (both components
have been "spun up" to quasi-steady-state conditions prior to coupling).
It is this "model year" that is encoded in file names and in the netCDF
output files themselves. In the netCDF files, the time axis is given in terms of
"days since 0001-01-01 00:00:00.0" (days since 1 January of model year 1).
Since different climate change experiments were initialized from different
points in time of the control integration, the translation from model year
(time since the coupled model control run was intialized) to scenario year
(time represented in the forcing scenario hindcast or projection) can vary
from experiment to experiment. For the climate change scenario experiments
one can translate the "model year" of a particular experiment to a "scenario year"
by referring to the documentation provided in the
Control & Transient
Forcing Experiments section of these web pages.
Using "ts_catalog" Files to Find Variables
Included amongst the netCDF files of time series of monthly means
(a few levels below the "mon" directories) one will find a plain text file
named either "ats_catalog" (if under the "atm" directory) or "ats_catalog"
(if under the "ocn" directory). These "ts_catalog" files (formerly
referred to as "cheat_sheets") are used help
document the variables that are found in the netCDF files in that
particular directory.
Since there's just one line per 2-D variable in the ts_catalog
files, the descriptions are short, but they should be sufficient
to provide guidance to the user.
For example, if one is interested in the time series of monthly mean atmospheric
and land surface variables produced by the GFDL R30 experiment "IPCC_B2_01", one
should proceed to
http://nomads.gfdl.noaa.gov/dods-data/DecCen/r30/coupled/IPCC_B2_01/atm/ts/mon/raw/140yr/.
In that directory one will find a file named "ats_catalog". This is the text
file that lists the archived variables (variable numbers 3 through 267 in this case).
By the way, the "a" in ats_catalog represents "atmosphere",
and the "ts" = "time series".
The 265 netCDF files in the directory each contain 140 years worth of monthly
data (1680 time levels) of 2-dimensional horizontal fields for the full
96 by 80 (longitude by latitude) grid associated with the R30 spectral model.
In forming the monthly means, the model was sampled at least once a day.
So, if is are interested in the time series for temperatures at the
lowest sigma level (level 14) one can refer to the "ats_catalog" file,
noting the entry...
TS VAR 72 TEMP (14) deg_K temperature
... which indicates that temperature at level 14 is variable number 72.
So, one would want to work with file "ats_v72_m_y111_250.nc", which
is about 50MB in size.
Here's how to decifer the file name... let's consider the file
ats_v72_m_y111_250.nc
ats_v72_m_y111_250.nc
^ 1st character: "a" or "o" (atmosphere or ocean)
^^ the "ts" stands for "time series" format (multiple time levels).
^^^ "v" followed by a number tells us what variable number
is contained in the file. The "ats_catalog" file
can be used to translate the number into a variable name.
^ "a" or "m" (annual mean or monthly mean)
^^^^_^^^ y????_???? gives the range of years contained
in the file (in model years relative to the point
that the atmosphere and ocean components were coupled.
^^^ the ".nc" signifies a netCDF file.
Although the experiment name is not encoded in the file name, one
will be able to deduce this from either the directory in which the
file resides, (in this case IPCC_B2_01) or by using
a common netCDF utility "ncdump -h".
Click [HERE]
to see what information the command "ncdump -h" provides us for this file,
and see how it can be used to check the contents of a file.
Because not all experiments archived the same variables,
one should be careful to refer to the "ts_catalog" file
that is located in the same directory as the data files
one is using. For example, experiments with variable
tropospheric sulfate forcings have two additional atmospheric
variables archived than was the case for the control integration.
This is a documentation file for R30 coupled model output available on GFDL's NOMADS Server
http://nomads.gfdl.noaa.gov/dods-data/DecCen/r30/coupled/
|
