[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: NetCDF-4

Glenn,

> I'm considering installing the Open ADDE server behind our gds to deal 
> with satellite data.  But-0 I wonder how the new NetCDF-4 built on HDF 
> could also help solve this problem with eh complexity of O-ADDE?  Is 
> this apples and oranges?  Thoughts? Thx, Glenn

NetCDF-4 won't be of any help with this in the near term, because:

  - NetCDF-4 is not released yet, and won't be until HDF5 1.8 (with
    features it depends on) is released.

  - Hence, no one is storing satellite data using netCDF-4 yet.

  - Even if satellite data was being stored as netCDF-4 using some
    yet-to-be-agreed-upon netCDF-4 conventions, ADDE doesn't serve
    netCDF-4 data yet.  And OpenADDE is still under development, as
    far as I know.

Ultimately if netCDF-4 succeeds (not a foregone conclusion, it may
take some years to become as widely used as netCDF-3), there may be
reason to think about using it as an API for NPOESS data, for example.
But current plans are to store NPOESS data as HDF5, and we're not
certain that it will be accessible through the netCDF-4 interface
unless it is written through the netCDF-4 interface.  

We don't envision netCDF-4 as an API thorough which you can access
arbitrary HDF5 data, just HDF5 data that conforms to the restrictions
of the Common Data Model.  An example of such restrictions is that
each Group must have a unique parent Group, unlike in HDF5 where the
Group structure can include arbitrary cyclic and disconnected graphs.

Details of the Common Data Model are still being worked out, but here
is a site Ed Hartnett just put together where we are gathering
information on the Common Data Model:

  http://my.unidata.ucar.edu/content/software/netcdf/CDM/index.html

Some of this work is supported by a one-year NASA grant to advance
netCDF support for the Common Data Model and implementation of the
highest "Scientific Data Types" layer of that data model.

--Russ