[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[THREDDS #GJF-200148]: TDS 4.3 GRIB names different from TDS 4.2
- Subject: [THREDDS #GJF-200148]: TDS 4.3 GRIB names different from TDS 4.2
- Date: Wed, 22 May 2013 09:20:06 -0600
Fan,
I was out of town for a few days, but I'll get my head into your issue again
today and let you know what I find.
-Lansing
> Hello Lansing,
>
> Any comments/findings about my questions?
>
> With or without featureCollection we need to serve grib data in their granule
> form, so how to make grib table work for individual granules is remaining
> question number one. So far our grib table works with aggregated
> featureCollection dataset, but not for granules under 'files' folder. I also
> tried the 'iospParam' attribute of the 'netcdf' element outside
> featureCollection, like
>
> <datasetScan name="GLDAS_CLM10SUBP_3H" ID="GLDAS_CLM10SUBP_3H"
> path="GLDAS_CLM10SUBP_3H" location="/ftp/data/s4pa_TS2/GLDAS_CLM10SUBP_3H/">
> <netcdf xmlns="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2"
> iospParam="gribParameterTable=/home/ffang/gribtab_clm.tab">
> </netcdf>
> ...
>
> and TDS didn't seem to read the grib table.
>
> The other main question is about the collection .ncx file usage. I found
> accessing that file very speedy in TDS and wonder if it should be used to
> serve our extremely long time series datasets (~30 years of hourly data),
> i.e. shall we let TDS make the file at the data location instead of under
> cache/cdm, and treat it as the timeseries dataset?
>
> -Fan
>
> ________________________________________
> From: Fang, Fan (GSFC-610.2)[ADNET SYSTEMS INC]
> Sent: Friday, May 17, 2013 2:50 PM
> To: Fang, Fan (GSFC-610.2)[ADNET SYSTEMS INC]; address@hidden
> Cc: address@hidden
> Subject: RE: [THREDDS #GJF-200148]: TDS 4.3 GRIB names different from TDS 4.2
>
> I take it back - the .tab extension apparently works. We previously made an
> error renaming the grib table.
>
> I also changed '/s' to '.s' but both came out the same. For example,
> kg/m^2/s or kg/m^2.s would show up as kgm^2s, which is wrong. I suppose we
> have to change it to kgm^-2s^-1 in the table? I hope it is consistent with
> software like 'wgrib'.
>
> It seems the variable names are essentially the long names and units in grib
> table plus something like '_surface', with space chars replaced by
> underscores. I guess we can survive with that.
>
> These show up in the collection 'Best Timeseries' (is there a way to
> re-configure this collection name?), but not for individual files under
> 'files' folder in TDS (again, is there a way to configure for the folder
> name?). Is this expected?
>
> One final question: if I drop the collection index file .ncx at the data
> location, TDS seems to be able to serve it like a dataset, and for time
> series it is surprisingly fast to serve the .ncx file instead. Is this
> intended - in other words we shall direct user to use .ncx, especially for
> long time series? I know in default it's under cache/cdm and not exposed to
> users. What's the catch here?
>
> Thanks.
>
> -Fan
> ________________________________________
> From: Fang, Fan (GSFC-610.2)[ADNET SYSTEMS INC]
> Sent: Thursday, May 16, 2013 6:13 PM
> To: address@hidden for
> Cc: address@hidden
> Subject: RE: [THREDDS #GJF-200148]: TDS 4.3 GRIB names different from TDS 4.2
>
> I can confirm changing the grib table name to have .tab as extension still
> does not work.
>
> I'll try the modifications in the units, but wonder why it fails for all
> since some of them have legit units, like 'K' for temperature.
>
> -Fan
> ________________________________________
> From: Unidata THREDDS Support [address@hidden]
> Sent: Thursday, May 16, 2013 2:37 PM
> To: Fang, Fan (GSFC-610.2)[ADNET SYSTEMS INC]
> Cc: address@hidden; Fang, Fan (GSFC-610.2)[ADNET SYSTEMS INC]; address@hidden
> Subject: [THREDDS #GJF-200148]: TDS 4.3 GRIB names different from TDS 4.2
>
> Fan,
>
> I did a little more digging, and I don't think the file name extension will
> matter - that was my misinterpretation of how the table was getting picked
> up. However, you will probably have the best result directly embedding the
> table information in your featureCollection element using the xml notation.
> It looks like you tried this already, so I am not sure why it did not work
> for you. It is possible that it broke because you did not follow udunits
> standard -- for instance, square meters is notated as m^2, rather than
> m(superscript)2. The code you are asking about is just the number that
> identifies the grib parameter.
>
> Inside your <featureCollection> element:
>
> <gribConfig>
> <parameterMap>
> <parameter code="2"> <<-- This will override the standard table values for
> parameter 2 (code < 128 are standard, > 128 are local)
> <description>Pressure reduced to MSL</description>
> <units>Pa</units>
> <name>PRMSL</name>
> </parameter>
> <parameter code="131">
> <description>Snowfall rate</description>
> <units>kg/m^2.s</units> <<--Compound units are notated with a dot, so /m^2/s
> is just /m^2.s
> <name>Snof</name>
> </parameter>
> ...
>
> </parameterMap>
>
> </gribConfig>
> -Lansing Madry
> Unidata
> Boulder, Colorado
>
> Ticket Details
> ===================
> Ticket ID: GJF-200148
> Department: Support THREDDS
> Priority: Low
> Status: Closed
>
>
-Lansing Madry
Unidata
Boulder, Colorado
Ticket Details
===================
Ticket ID: GJF-200148
Department: Support THREDDS
Priority: Low
Status: Open