Hi Don:
On 11/26/2013 8:41 AM, Don Murray (NOAA Affiliate) wrote:
Hi John-
On 11/26/13 8:24 AM, John Caron wrote:
Thanks for your ideas, I hadnt considered leaving "reference time" as
otherwise undefined. The downside would be that GRIB will then be
different from, say, netCDF-CF forecast model collections.
Isn't "run time" a construct of the CDM/FMRC, not CF? I don't see
reference to "run time" in the CF document. If that's the case, you
could do whatever you want. ;-)
"run time" is intended to reflect the semantics of forecast models.
Obviously still trying to figure out what that means ;)
CF, as usual, lags behind a bit.
Examining the NCEP IDD feed, 13 out of 400K GRIB-2 records have this
feild set to 0, the rest to 1:
significanceOfReference
0: count = 13
1: count = 406196
Im not sure if these are miscoded or not. What do you think the
difference is between "Analysis" and "Start of Forecast" ?
It would be helpful to know what those 13 are to know if they are
miscoded. I would imagine that a reanalysis grid might use analysis, or
perhaps the RTMA, but for either of those, I would expect there to be
more in your collection.
what do you think the difference would be and would it actually matter?
Do you have a script/code that can easily check this? If so, share it
and I (and others) could run it over some of our grib collections.
where else would it be but in the undocumented ToolsUI ??
ToolsUI / Iosp / GRIB2 / GRIB2-Report
enter collection spec (eg Q:/cdmUnitTest/tds/ncep/.*.grib2)
choose "idProblems"
hit RedRaw ("Make report")
the "use Index" toggle will turn on/off details
i just added significanceOfReference in v4.4.0, which im hoping to
release today. So once i do that you can give it a try.
thanks,
John
