Re: ECMWF forecast data GRIB field split

Hi Adrian,

I will speculate that NCL treats the file this way because the earlier
records have different "time range indicators" (see http://www.nco.ncep.noaa.gov/pmb/docs/on388/table5.html
  for the messy details).
The '_10' indicates that the later GRIB records are using 2 bytes to
store the forecast time (time range indicator 10) whereas the earlier
records use only 1 byte. This of course is a meaningless distinction
for an end user such as yourself, and it would be nice if it could be
overcome somehow.

Can you point me to a sample of this data or else upload a sample? See
the page http://www.ncl.ucar.edu/report_bug.shtml for instructions on
uploading data.
  -dave

On Jun 21, 2010, at 6:52 AM, Tompkins Adrian wrote:

> hello,
>
> when I retrieve GRIB data from an ECMWF forecast and try to read it in
> with "addfile", the variable seems to be split into two components,
> one
> for the first 10 days, e.g:
> CP_GDS4_SFC
>
> and then for the remainder of the forecast:
> CP_GDS4_SFC_10
>
> with each a function of a separate time variable:
> forecast_time0
> forecast_time3
>
> I do not see this split in any other software (XCONV, gribprofile,
> GRADS...) and presume it is a hardwired NCL feature? (10 days being
> the
> standard length of the operational deterministic forecast).
>
> Is there any way to switch this behaviour off, I do not see anything
> on
> the help pages?
>
> many thanks.
>
> Adrian
>
> --
> +39 040 2240579 (office) | Earth System Physics
> +39 040 2240449 (fax) | ICTP (UNESCO-IAEA)
> esp.ictp.it | Strada Costiera 11
> www.ictp.it/~tompkins | 34014 Trieste, Italy
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Mon Jun 21 13:18:55 2010