Re: [Ncwms-users] FMRC

[Zachary R. <Zac...@ge...>]

Jon and Adam,
It is funny this is coming up now.  We are currently in a meeting trying to establish some standardized uses of the WMS spec to support this very same idea across different model producers in the US DOD.

We think you can use another generic dimension called Base_Reference_Time and still be within the OGC spec.  An small issue pops up with what the get capabilities advertises.  It does not capture the concept that we don't have data for all permutations of "TIME" and "Base_Reference_Time".

We could use :time_origin = "2009-02-13 00:00:00"; from the netCDF header for this "Base_Reference_Time".  If we set the default value for "Base_Reference_Time" to "current", other clients can just get the latest forecast when that dimension is not specified.  Maybe ncWMS getmetadata could be be extended to take an optional (real) time and return the available "B_R_T's" to help the GUI.  

We are hoping to get someone to the June '09 OGC Technical Committee Meeting to run this up the official flagpole.  Does anyone know a voting member?

Zach Rouse
Diamond Data Systems,
A Geocent Company
zac...@ge...
cell:   303.502.4007
desk:  504.697.2110
office:  800.218.9009


-----Original Message-----
From: jon...@gm... [mailto:jon...@gm...] On Behalf Of Jon Blower
Sent: June 2, 2009 4:24 PM
To: Lundrigan, Adam
Cc: ncw...@li...
Subject: Re: [Ncwms-users] FMRC

Hi Adam,

Yes, this would be a nice addition to ncWMS (which currently only uses
the "best estimates" timeseries).  Perhaps the easiest way to handle
this would be to create a new dataset for each forecast time, e.g.:

My FMRC dataset
   Best estimates
      temperature
      salinity
      velocity
   6hr forecasts
      temperature
      salinity
      velocity
   12hr forecasts
      ...

Would this answer your needs?  The alternative would be to put *two*
time controls on the Godiva2 web interface - one to select the run
time and one to select the forecast time.  However, the WMS spec
doesn't support two time axes so we'd have to invent a way of passing
these parameters to the WMS.

Any thoughts?
Cheers, Jon