Re: 20050228: decoders and IDV

• To: laufers@xxxxxx
• Subject: Re: 20050228: decoders and IDV
• From: Don Murray <dmurray@xxxxxxxxxxxxxxxx>
• Date: Mon, 07 Mar 2005 06:24:45 -0700

Mark-

Mark J. Laufersweiler wrote:

I am confused now. The new CDL does indeed take care of the pressure
tendency variable warning in the IDV, but the IDV is still not
displaying all data from my netcdf metar files. Variable such as
temps, dewpoints, stid, etc, are not displaying while wind direction,
wind speed, station location (den and okc are where you expect to
find them on a map).

current file:
http://weather.ou.edu/~laufers/2005030412_metar.nc

I am trying to setup our DODS/ADDE/THREDD server and figured that a
good test is to set up the metars. One pqact item only is needed.
Get it working, then set up the DODS, get it working and then set up
the catalog, get it working. Then set up the ADDE. Nice plan but I
am not getting out of the batters box :)

I feel that I am missing something here and am not sure what.

There are a couple of issues here.  The first is the issue of
time.  In the IDV reader, we use the time_nominal variable in
the file as the time of the observation.  The idea was that
this would indicate a nominal time (12Z, 12:20Z, 12:40Z) that
would create a grouping of time, not the observation time which
would vary on the order of minutes.  As you know, GEMPAK and
McIDAS group observations into blocks of time (what I think
of as nominal time).  This is why we prefer to use the ADDE
interface to point data instead of netCDF.  For your file,
I end up with 45 time steps.  You may have to turn on
animation to see data at the various steps, or see the
steps below.

Second, the station model you pick may not be showing up the
observations.  Right now, there is no way to specify which
station model to use by default for the netCDF files like there
is for the ADDE interface.  So, once you load in the
file and display it as point obs, make sure you are using
something like the METAR or Temperature station model (select
from the control window).  When I use either of these, your
data displays with no problem.  You can create your own
station models to show the variables you want if neither of
these is satisfactory.

Another thing you can do to get around the time issue is to
select the "Show All Times at Once" checkbox in the Point
Data Plot control window to see all the times plotted at once.
The upside is that you can see all the reports in the file
(accounting for the decluttering setting of course).  The
downside is that you have no distinction between main hour and
intermediate (special) reports and if you had more than one
time step in the file, you'd lose the distinction between hours.

At this point, the IDV reading of netCDF point data is not
optimal and we will be working on that over the next few months.
As you see, it does not handle time well and does not handle loading
multiple files very well.  It does not handle loading data
through DODS (or HTTP I see, but that's easily correctable).
At this point, the only netCDF data sources that can be easily
cataloged and read by the IDV are grid.  There's still a lot
of middleware that needs to be developed to handle other
types of data.

We will be using the new netCDF interface if possible to improve
that.  In the mean time, I would suggest using the ADDE interface
which provides you with time selection and the ability to set
the preferred station model.   This will also allow you to test
your ADDE setup.

Hopefully, we will be able to create a similar (and improved)
interface for netCDF data.  However, issues of time selection will
still be problematic given the nature of the data.  If you have any
suggestions on how we should handle that, please let us know.

As a member of the IDV Steering Committee, you'll get the
chance to provide input on whatever interfaces we develop.

Thanks for bringing these issues to the forefront.

Don
*************************************************************
Don Murray                               UCAR Unidata Program
dmurray@xxxxxxxxxxxxxxxx                        P.O. Box 3000
(303) 497-8628                              Boulder, CO 80307
http://www.unidata.ucar.edu/staff/donm
"There's someone in my head, but it's not me"    Roger Waters
*************************************************************