[ESIP-AQ] Weather Observations and Forecast Datasets

[Ben Domenico]

Hi,

This is the email I promised with pointers to datasets on our prototype
real-time weather data THREDDS Data Server.

The top level THREDDS catalog is at:

 http://motherlode.ucar.edu:8080/thredds/catalog.html

In turn, it points to collections of NCEP Forecast Model Output, Radar Data,
Weather Station Observational Data, GOES Satellite Data, and a few
specialized collections.

If you drill down into the Model Output, you'll find output from many Global
Models, North American Models, Rapid Update Cycle Models, even some ensemble
outputs.

Drilling down into the half degree resolution Global Forecast System (GFS)
model, you see that there are several alternative dataset forms in which you
can access the data.  You can get at the data in the form of the individual
files that contain the output of a given run of the model.  Or you can
access a virtual dataset called the Best Time Series.

Choosing the Best Time Series Dataset finally gets you to the metadata which
includes some of the things we talked about in the teleon this morning:


   - a variety of access methods: opendap, wcs, wms, netcdfsubset, and some
   experimental ones
   - a long list of parameters for which there are forecasts,
   - spatial and temporal extent (which extends into the future)
   - documentation

If you go one step further and select netcdfsubset service, you get a page
that actually allows you to select a subset of the dataset for delivery to
your computer.

Keep in mind that each of these html pages is generated from an xml document
that can be used for programmatic access.

You can start back up at the top and drill down into Radar Data and quickly
discover the heirarchy is based on a set of arcane radar product
abbreviations (e.g., N0R for base reflectivity, N1P for 1 hour precip,
etc.).  The next level in the hierarchy gives the observing stations in
terse 3-letter "abbreviations" (FTG for Denver, GJX for Grand Junction,
etc.)

On the other hand, the weather station observations (METARS) are collected
into hourly files containing data from all reporting stations around the
globe.

One other thing to note is that not all datasets are accessible by all the
protocols.  Most are accessible as complete files via HTTP and as subsets
via OPeNDAP, and a few (mainly model output and other gridded datasets) are
accessible WCS and WMS interfaces.

I'll stop here and simply call attention to the fact that the task of
automating a metadata generation system to put the existing information into
a form useful for use in GEOSS is not trivial.  On the other hand, it is
probably tractable if we start with one of the forecast models and move from
there to the other data types.

Also I would point to the GI-cat technology from the U of Florence that
seems to be very good at harvesting catalog information from THREDDS and
other services and making it available via standard catalog interfaces.

http://zeus.pin.unifi.it/cgi-bin/twiki/view/GIcat

I hope this helps.

-- Ben
-------------- next part --------------
HTML attachment scrubbed and removed