Re: time bounds

To: Andrew Wittenberg <Andrew.Wittenberg@xxxxxxxx>
Subject: Re: time bounds
From: Steve Hankin <Steven.C.Hankin@xxxxxxxx>
Date: Tue, 19 Oct 2004 10:37:10 -0700
Cc: Mark.Collier@xxxxxxxx, ferret_users@xxxxxxxx
References: <Pine.LNX.4.44.0410191042480.18802-100000@atw>
Sender: owner-ferret_users@xxxxxxxxxxxxx

Andrew Wittenberg wrote:

Hi Mark,
I've had similar problems. As far as I know, descriptors cannot handle
irregular within-NetCDF-file time axes (as of Ferret v5.70). This
capability would be a most welcome addition in the future!

Hi Andrew, et. al.,

Just a little discussion about the aggregation of irregular time step files.

The current MC descriptors support irregular start times for each file, but assume that the time steps within individual files are regular. The reason for this limitation is the design requirement that only the first file in the MC-aggregation should need to be opened. Given this requirement (which is important for performance) various alternative strategies are available for obtaining the true irregular time steps. For example, we could offer a way to enumerate each time step value in the MC descriptor; or a way to point to some external resource that contains that enumeration. With a larger investment of effort we could have an initialization step (could be a separate utility) that automatically built the sequence of irregular time steps (the "external resource") for you.

Currently there are two work-arounds available -- that provide full functionality, but at some additional effort:

(as andrew suggests) Put the sequence of irregular time steps into an axis defined in a separate netCDF file and assign this axis to the (incorrect) time axis of the MC dataset using a Ferret command; or
use the Unidata aggregation server to create a virtual dataset

The existence of these work-arounds has kept the priority level (read "pressure") for a code enhancement relatively low. Perhaps an FAQ on the subject would be a sufficient solution. We invite your comments.

- steve

A workaround is to regrid your variables to the correct time axis in
Ferret. The hard part is defining the time axis; monthly data is
especially tricky due to leap years etc. The days1900 function can help
with this.
In our model data we're lucky enough to have the time axis values
available as an additional variable in the NetCDF file, so I just regrid
to those. Otherwise, you could use ncrcat to concatenate just the time
axis values from your NetCDF files into small new file, and then use those
in Ferret.
Andrew
On Mon, 18 Oct 2004 Mark.Collier@csiro.au wrote:
> Hi
>
> The following descriptor file works o.k., however, it assumes that each
> month is of equal size. Is there some way of getting ferret to use the
> time bounds information used in the individual files? These separate
> files have different time values and upper and lower limits. This would
> enable more accurate analysis/plots.
>
> Thanks,
> Mark.
>
> &FORMAT_RECORD
>    D_TYPE             = ' MC'
>    D_FORMAT           = ' 1A'
> /
>
> &BACKGROUND_RECORD
>    D_TITLE            = '20C3M Experiment'
>    D_T0TIME           = '16-DEC-1870 12:00:00'
>    D_TIME_UNIT        = 2628000.0
>    D_TIME_MODULO      = .FALSE.
>    D_CALTYPE          = 'NOLEAP'
> /
>
> &MESSAGE_RECORD
>    D_MESSAGE          = ' '
>    D_ALERT_ON_OPEN    = .FALSE.
>    D_ALERT_ON_OUTPUT = .FALSE.
> /
>
> &EXTRA_RECORD /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1871-
> 1880_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 1
>    S_END              = 120
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1881-
> 1890_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 121
>    S_END              = 240
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1891-
> 1900_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 241
>    S_END              = 360
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1901-
> 1910_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 361
>    S_END              = 480
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1911-
> 1920_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 481
>    S_END              = 600
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1921-
> 1930_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 601
>    S_END              = 720
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1931-
> 1940_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 721
>    S_END              = 840
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1941-
> 1950_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 841
>    S_END              = 960
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1951-
> 1960_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 961
>    S_END              = 1080
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1961-
> 1970_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 1081
>    S_END              = 1200
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1971-
> 1980_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 1201
>    S_END              = 1320
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1981-
> 1990_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 1321
>    S_END              = 1440
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         =
> '/cs/datastore/csdar/cec006/ocean/ipcc/data/20C3M/A1a/run1/psl_A1a_1991-
> 2000_20C3M_run1.nc'
>    S_AUX_SET_NUM      = 0
>    S_START            = 1441
>    S_END              = 1560
>    S_DELTA            = 1
>    S_NUM_OF_FILES     = 1
>    S_REGVARFLAG       = ' '
> /
>
> &STEPFILE_RECORD
>    S_FILENAME         = '**END OF STEPFILES**'
> /
>
>
+--------------------------------------------------------------+
| Dr. Andrew T. Wittenberg      | Mailing address:             |
| Physical Scientist, GFDL/NOAA | US DOC/GFDL                  |
| Andrew.Wittenberg@noaa.gov    | Attn: Andrew Wittenberg      |
| Voice: (609) 987-5064         | Forrestal Campus, US Route 1 |
|   Fax: (609) 987-5063         | Princeton, NJ 08542          |
+--------------------------------------------------------------+

Steve Hankin, NOAA/PMEL -- Steven.C.Hankin@noaa.gov
7600 Sand Point Way NE, Seattle, WA 98115-0070
ph. (206) 526-6080, FAX (206) 526-6744

References:
- Re: time bounds
  - From: Andrew Wittenberg

Previous by thread: Re: time bounds
Next by thread: 3 argument shade, sigma coordinates, layer depths

[Thread Prev][Thread Next][Index]

Dept of Commerce / NOAA / OAR / PMEL / TMAP