Re: Visualizing model output

Steve,

> I don't see a WRF in this directory, what would be the next best one for us
> to work with?  ETA?

Yes, I think ETA would be good, as would GFS or RUC.  There is a
version of ETA called Workstation ETA that people run for local
(mesoscale) forecasts, because it's freely available and can run on
desktop and departmental platforms.  I'm not an expert on the
differences among these models, but all of these are used regularly.
If you wanted some data from a recent Workstation ETA run for the
Colorado region, for example, I could probably find some and make it
available.

I probably should not have recommended WRF for the application you're
talking about, since it's still a research effort.  But it's being
used in some of the projects we're involved with, and is probably
destined to become the model that reflects the best state-of-the-art
for forecasts.  We're involved in a multi-organization project named
LEAD that runs WRF and archives model results, but I don't think they
are made available outside of the project currently.

> To make this easier for you maybe, which one would have something like
> temperature, pressure, wind direction, wind speed? Having all that in one
> dataset would keep me busy for a while.

I think they all have surface temperature and surface pressure, but
none have wind direction and wind speed explicitly.  Instead, they
have "u wind" and "v wind" which are the eastward and northward
components of wind, respectively, so you have to compute the wind
speed and direction from these.  Also, most of the parameters are
defined at "isobaric levels" such as 500 millibars (also called
hectopascals), which means the surface on which the pressure is a
constant 500 millibars, so plotting the constant pressure on one of
these surfaces would not be very interesting :-).  A more interesting
parameter might be "relative humidity".

> Since you say these are converted from GRIB, does that mean all these models
> output GRIB?  I thought NetCDF was the most important format for you guys.

GRIB is an international standard that's good for data transport,
because it emphasizes data compression and it has no language or
character-set bias (it's all numbers).  Almost all operational models
output GRIB if the output will be archived or distributed to other
countries.  It's the blessed standard of the World Meteorological
Organization (WMO).

But it's not particularly good for analysis and visualization
applications to use because it only represents two-dimensional
horizontal fields, has no standard API for data access, and requires
external centrally-maintained tables for interpreting the meaning of
the data.  So the GRIB model data is often stitched together into
multidimensional netCDF files that have a supported API for data
access.  (But there are also some applications that can use GRIB data
directly, such as GrADS and GEMPAK.)

> To download one of these files, do I need a password?  I'm getting an error
> message "The password was not allowed" when I just try to copy it through
> Internet Explorer. Do I need to ftp it?  Can't do that from home through the
> firewalls right now.

Sorry, I didn't remember those files required a password.  Since this
is being archived in our support system, I'll have to send that in a
separate email.

> Thanks for the link to the Redbook graphics.  Would it be useful or
> interesting to the atmospheric community if for example after a WRF run, a
> collection of predefined map layers were generated and available as part of
> a mapping service on the web where you could combine several layers and
> change symbology?  Or are the consumers of this data happy with this and
> just use it to make value added products?   

That's a complicated question.  Some of the people we work with are
more interested in data than pictures, often so they can use the data
to create their own web sites with their analyses and forecasts.  Also
the reason users in the research and education communities need
applications like the IDV and GEMPAK to look at the data is that no
set of predefined layers can adequately capture the complexity of the
atmosphere, address the diverse interests of researchers, or help
forecast what will happen as well as being able to interactively
explore the data.

On the other end of the scale, there are users who are most interested
in graphics, and for them a small set of predefined layers might be
fine.  In the U.S., a commercial weather sector provides such services
to various niches of users, such as aviation, agriculture, emergency
response organizations, the marine sector, etc.  So the set of
graphics needed differs for each set of users.

I have to emphasize that my area is software engineering, not
atmospheric sciences.  Other Unidata staff are better qualified to
answer some of your questions, but most of them are trying to get
ready for the annual AGU meeting in San Francisco next week, which
will be followed shortly by the annual meeting of the American
Meteorological Society in which we're quite active.  But I'll CC: Jeff
Weber also, in case he can correct anything he sees wrong in this
reply.

--Russ