[NARCCAP-discuss] time discrepancy

Seth McGinnis mcginnis at ucar.edu
Fri Sep 7 11:34:11 MDT 2012


Hi Shane,

The values in the time coordinate variable should be continuous in all the
published data, with only a couple exceptions (noted on the Data Status page).

This sounds like a problem related to Excel's representation of dates.  Are the
skips all after leap years?  One way to check that hypothesis would be to
download some of the data driven with HadCM3, which uses a 360-day calendar.
 If that is the problem, then you'll probably get jumps of 5 or 6 days at the
end of each year.  (Note that there's a repeated timestep between the first and
second files in the HRM3-hadcm3 driven runs; this problem is very nearly fixed,
and I expect to publish corrected data next week.)

The other possibility that springs to mind is that maybe your file downloads
were truncated, and you lost the last few steps on some files.  In that case,
try downloading one of the problem files again.

Cheers,

--Seth

On Thu, 6 Sep 2012 12:04:42 -0400
 Shane Brennan <sbren006 at odu.edu> wrote:
>Hi Seth,
>
>I notice "time" data skips a day near the end of the year in some
>cases. For example, in a few files I have looked at, the "time"
>variable goes from 1093 to 1095, as well as 2918 to 2920, etc. This
>does not happen in NCEP files which are Gregorian. Is this a
>consequence of the no leap year calendar? I had assumed that "time"
>counted up from 0 (from 1968) and that day 59 would be a leap day in
>Gregorian and March 1st in the no leap year calendar.
>
>Basically, I would like to compare NCEP data with modeled data on a
>monthly basis, so making sure I have my calendars straight is
>important. Does anyone have any suggestions on the best way to go
>about this? Currently I am creating two "days from" fields in Excel
>which have the time variable, and in one field I increment the value
>(and subsequent values) at leap years, all the way down, so that the
>final product is a field with the correct days offset depending on the
>number of preceding leap days. However, the skipped day I am running
>in to is causing me to put this method into question.
>
>Thanks, and sorry if I am missing something obvious...
>
>Shane
>
>
>On Sun, Aug 26, 2012 at 12:10 PM, Seth McGinnis <mcginnis at ucar.edu> wrote:
>> Hi Shane,
>>
>> Information about time in the NARCCAP data can be found here:
>> http://narccap.ucar.edu/about/time-periods.html
>>
>> Cheers,
>>
>> --Seth
>>
>> On Sat, 25 Aug 2012 16:09:54 -0400
>>  Shane Brennan <sbren006 at odu.edu> wrote:
>>>OK, thanks, Seth. Is the same calendar used for all models? Is there
>>>documentation on the timesteps used?
>>>
>>>Shane
>>>
>>>On Sat, Aug 25, 2012 at 11:25 AM, Seth McGinnis <mcginnis at ucar.edu> wrote:
>>>> Hi Shane,
>>>>
>>>> It's the calendar.  CCSM uses a 365-day or "noleap" calendar, in which
>every
>>>> year has 365 days and there are no leap years ever.  The date calculation
>>>> functions in Excel use the standard Gregorian calendar, which does have
>leap
>>>> years.
>>>>
>>>> If you count up the leap days between the base time (1968/01/01) and the
>>>start
>>>> of your file (1996/01/01), there are seven of them, which is the
>difference
>>>> between your expected start date and the start date you're seeing in the
>>>Excel
>>>> calculations.  (And then there's another one in 1996, so the discrepancy
>for
>>>> the end dates is 8 days.)
>>>>
>>>> Cheers,
>>>>
>>>> --Seth
>>>>
>>>> On Fri, 24 Aug 2012 11:13:44 -0400
>>>>  Shane Brennan <sbren006 at odu.edu> wrote:
>>>>>Hi,
>>>>>
>>>>>I am writing a C# program to parse and manipulate NARCCAP data. I
>>>>>wanted to verify the missing data values to make sure I was handling
>>>>>them correctly so I checked out:
>>>>>
>>>>>http://www.narccap.ucar.edu/data/missing/CRCM_ccsm-current_table3_missing.txt
>>>>>
>>>>>to find where I could look for missing values. In the URL above, it
>>>>>states that evps in the file evps_CRCM_ccsm_1996010103.nc has missing
>>>>>values for 1999-12-01 03:00 through 2000-01-01 00:00.
>>>>>
>>>>>I would expect the time series for the entire file to run from
>>>>>1996-01-01 03:00 to 2000-01-01 00:00. However, when I parse this file
>>>>>with my program and then use excel to calculate the times from
>>>>>1/1/1968 (the start date returned from ncdump for this file), I get a
>>>>>time series that runs from 12/25/1995 03:00 to 12/24/1999 00:00. In
>>>>>this file, my missing values run from 11/23/1999 03:00 to 12/24/1999
>>>>>00:00.
>>>>>
>>>>>Naturally, I became concerned that I had a bug in my program. However,
>>>>>when I use nc2text.exe using the instructions detailed in the ASCII
>>>>>HowTo, I get the same results.
>>>>>
>>>>>In Excel I calculate the offset date by entering 1/1/1968 into a cell,
>>>>>and calculating a new field as that cell value added to each time
>>>>>value (for example, if the cell for 1/1/1968 is G1, then my formula
>>>>>for the first cell is "=$G$1 + a1". I then copy this formula all the
>>>>>way down).
>>>>>
>>>>>Does anyone know what is going on? Perhaps there is some issue with
>>>>>leap years or DST?
>>>>>
>>>>>Shane
>>>>>_______________________________________________
>>>>>narccap-discuss mailing list
>>>>>narccap-discuss at mailman.ucar.edu
>>>>>http://mailman.ucar.edu/mailman/listinfo/narccap-discuss
>>>>
>>>> _______________________________________________
>>>> narccap-discuss mailing list
>>>> narccap-discuss at mailman.ucar.edu
>>>> http://mailman.ucar.edu/mailman/listinfo/narccap-discuss
>>>>
>>>>
>>>> --
>>>>
>>>_______________________________________________
>>>narccap-discuss mailing list
>>>narccap-discuss at mailman.ucar.edu
>>>http://mailman.ucar.edu/mailman/listinfo/narccap-discuss
>>
>> _______________________________________________
>> narccap-discuss mailing list
>> narccap-discuss at mailman.ucar.edu
>> http://mailman.ucar.edu/mailman/listinfo/narccap-discuss
>>
>>
>> --
>> BEGIN-ANTISPAM-VOTING-LINKS
>> ------------------------------------------------------
>>
>> Teach CanIt if this mail (ID 697113074) is spam:
>> Spam:
>
      https://www.spamtrap.odu.edu/b.php?i=697113074&m=b96d08dc8f44&t=20120826&c=s
>> Not spam:
>
  https://www.spamtrap.odu.edu/b.php?i=697113074&m=b96d08dc8f44&t=20120826&c=n
>> Forget vote:
>https://www.spamtrap.odu.edu/b.php?i=697113074&m=b96d08dc8f44&t=20120826&c=f
>> ------------------------------------------------------
>> END-ANTISPAM-VOTING-LINKS
>>
>_______________________________________________
>narccap-discuss mailing list
>narccap-discuss at mailman.ucar.edu
>http://mailman.ucar.edu/mailman/listinfo/narccap-discuss



More information about the narccap-discuss mailing list