[Wrf-users] IDV and VAPOR

[Leigh Orf]

"Mark Stoelinga" wrote:

|   I'm curious about IDV and VAPOR, two relatively new
|   visualization and analysis packages from UCAR (IDV from
|   Unidata, first released in 2004, and VAPOR from NCAR/CISL,
|   first released in 2006). I admit I haven't tried either of
|   them, but from the web pages, they both look like great
|   software packages, with much effort poured into them. I have
|   seen a increasing discussion on the users' email list about
|   both of these packages recently.
|
|   My first question is from a policy perspective: Why is UCAR
|   developing two major visualization/analysis packages that, at
|   first glance, seem to have many of the same capabilities?
|
|   I'm also curious about WRF users' experiences with these
|   two packages. How seamless are each of these packages with
|   ingesting native WRF system data? What are users' experiences
|   with using these two packages for analyzing and displaying
|   WRF system data? Has anyone used both, and has an opinion on
|   relative strenghts and weaknesses?
|
|   Mark

IDV is more of a swiss army knife for viewing weather data but its
performance can be rather slow in a lot situations. It is written in
Java and you can tell. It's resource intensive and has a tendency to
eat up a lot of memory. It has a bazillion menus/options and can be
difficult to navigate until you spend a lot of time with it. IDV can do
3D isosurfaces but not volume rendering. It also has built in mapping
capabilities. It can read just about any meteorological data format
you can throw at it, point source data, level 2 and 3 radar data, GRIB,
GEMPAK, netcdf, etc. It also connects seamlessly to Unidata datastreams
(THREDDS servers, ADDE servers).

VAPOR is more of a 3D scientific analysis tool for numerical model
output. It is written in C++. It is in rapid development right now.
I have only limited experience with it (that should change soon).
VAPOR uses a wavelet transform function to create data files from raw
weather data. This means the software only reads what it needs for a
given vantage point / chosen granularity. This can increase performance
markedly over a tool such as Vis5D, which is probably its closest
analog. You must convert your data to VAPOR'S format before you can view
your data, unlike IDV which is truly an integrated data viewer as its
name implies.

WRF output data can viewed relatively easily with VAPOR, as it is one
of the main targets the developers are aiming for with regards to
sources of data to visualize. In addition, IDV can read WRF netcdf
files just fine. I believe raw netcdf output from WRF does require some
massaging to get in what most would consider a useful form (hence all
the post-processing software for WRF).

In closing I would say if you are looking to do research with the WRF
model, VAPOR is probably is a more appropriate tool. If you want to be
able to take a quick look at model output without having to convert
anything, give IDV a whirl. When you get around to publishing your
results, however, I'd say neither is really the right tool - go with
NCAR graphics or matplotlib (my recent favorte).

Leigh

--
Leigh Orf
Assistant Professor of Atmospheric Sciences
281 Dow Science Complex
Central Michigan University
Mount Pleasant, MI 48859
(989)774-1923
Amateur radio callsign: KG4ULP
Research journal: http://research.orf.cx/journal