|Subject: Visualization Meeting|
|From: "John A. Quinlan" |
|Date: Thu, 01 Nov 2001 11:44:39 -0500|
|To: firstname.lastname@example.org, email@example.com, firstname.lastname@example.org, email@example.com|
This is about what I thought I heard. Let me know of any corrections, additions.
Minutes of the visualization meeting held in Swift House on 10/31/2001
Attended by: Dickey Allison, Bob Groman, Skip Little, John Quinlan
and Peter Wiebe
Summary: Skip and Peter called for a meeting to discuss directions
for GeoZui3D software development. The major focus right now seems to be
refinement and rapid development of a few tools. This approach is suggested
to expose individual GLOBEC researchers to the software as quickly as possible.
Individuals could then assess whether the software might be helpful in their
own work. They may also suggest productive areas for future development.
Skip will focus on cleaning up and packaging the three stand-alone matlab
utilities, development of a method to handle vector-valued data and refinement
of the current curtain plot techniques for acoustic data. Wiebe, Chu and
Quinlan were identified as the set of 'test-dummies' during development and
will try to break/improve the software whenever/wherever possible. Peter
intends demo the software at the Southern Ocean Principal Investigator's
Meeting on December 10-12. For this meeting we need access to Southern Ocean
data through the data management system. The software has already been presented
to both the Georges Bank and USGLOBEC SSCs. Continued interaction with the
UNH group is encouraged so that the product will change quickly in the areas
we need most. There should be a move towards more 'generic' software which
handles typical data from the network and provides good feedback to the user
when problems arise.
Details: For more extensive testing, a group of potential beta-testors
were identified. These are Peter Wiebe, John Quinlan, Jim Manning, Dezhang
Chu, Dave Mountain, Bob Beardsley, Glen Flierl, perhaps a representative
from Ted Durbin's group. (Can we add Carin Ashjian, Cisco Werner's lab and
maybe Dan Sosta?)
NetCDF is a viable method of storing data within the oceanographic community,
especially within DODS and WHOI's DSL. Developing a NetCDF compatibility,
while maintaining the ASCII interfaces, should be pursued in the future to
make the software more widely applicable. It's hoped that Glen Flierl will
be able to contribute to this development. He should be brought into discussions,
especially those surrounding JGOFS.
Dezhang Chu should be brought into GeoZui3D discussions due to his work with
kriging and various large data sets around the Institution.
Skip feels that all of his stand-alone matlab utility programs, designed
to interface with data downloaded from the GLOBEC data management system,
should be rewritten. Skip will take the lead on wrapping these three programs
within a single interface.
Skip will also work through a preprocessing solution of the relative scaling
problem. Once the scaling problem is addressed in preprocessing, the UNH
group will be approached to see if something similar can be implemented within
the software. It seems that underlying the relative scaling problem is a
more generic issue of relative motion within a scene, which has implications
for labeling and different sorts of probes one might use to explore features
of the data.
Due to the upcoming SO Principal Investigator's meeting, the curtain plot
routines must be upgraded to handle 'generic' acoustics data from GLOBEC
programs. Acoustics data from the SO Program should be moved into the data
management system. Peter, Skip and Bob will work to accomplish this. We will
also want to be sure that the data manipulation and plotting is done in a
manner which approximates what a typical user might experience. The motivation,
maybe from here on, should be something like 'Using the existing system,
this is what we can do....'
For the past week or so we've been moving toward developing a method to display
vector-valued data. As a test case, Skip currently has access to a modeled
velocity field (from J. Manning) and some finite element bathymetry (bank150
from the Lynch group) from the Georges Bank Program. He plans to continue
development of these methods. ADCP, meteorological and mooring data should
be considered along with model output. We will have a need to display data
from models, ship tracks and fixed stations. The ability to display the components
of a vector, as well as the vector itself, would be useful. As would the
ability to produce normalized, or unit length vectors, and map vector magnitude
to the color of the vector. Input into the routines should be able to handle
the x,y,z vector origin and either the u,v,w components or magnitude and
The UNH group has stated that they will pursue vector, animation and interpolation
capabilities for the software. We should talk to UNH about the December Southern
Ocean meeting and the potential DODS interface needs. We should also ask
them about the hardware/software GeoZui3D requires. This could be a critical
concern within the community. A comment was made during one of the recent
demos that if the software couldn't run on existing laboratory computers
(there were six recent machines in this lab), then it probably wouldn't be
used. Bob suggested the development of a test suite which would exercise
the system capabilities before the software is installed. If the test is
successful, the software would run.
We should get an update on Dale Kiefer's visualization work as well.
Sometime in January Georges Bank GLOBEC will have information regarding the
successful Phase IV proposals. We should try to gear GeoZui3D development
to the Phase IV time table and needs.
Additional Comments: I spoke with Skip for a few
minutes after the meeting. We discussed the video board on my laptop, which
may not be suitable for the software, and the availability of a few other
visualization packages. Glen Wheless' lab (http://www.ccpo.odu.edu/~wheless/)
at ODU is involved in fairly intensive oceanographic visualization
and I believe they are using Vis5D(?). The finite element modeling group
seems to have moved towards IBM's DX system. Here's a snippet from an email
by Dave Greenberg to the QUODDY users group:
DX is IBM's free 3d data visualization software. It is what we are
using as a replacement for AVS. I think this is easier to use, faster
and more stable than the last version we had of AVS (and a whole lot
cheaper!). We have had no difficulty installing it on a couple of
machines using different linux distributions, including Mandrake and
Red Hat on Pentiums and alphas. It has been tested on may other UNIX
systems as well. I even have it running on my home machine - a 266 mhz
K6 with 64mb ram.
Anybody familiar with AVS will recognize the structure of DX, with
graphical connecting of modules to make a network that produces 3d
pictures. Like AVS, there is a considerable learning curve, but I
think it is worth it. DX includes a very good tutorial. With the
tutorial and the example networks Jason has included, it should not be
difficult for anybody to get up to speed in using it with any NML
Also, we should be aware that the EPA (http://www.epa.gov/vislab/)
has made a heavy investment in visualization. They are involved with meteorological
and watershed research and we may have a great deal to learn from their experiences.
I do know that they've produced a number of modules for AVS and it seems
they've branched into ARC products as well.
A general resource for visualization is:
Finally, is it worth the time to try and get something like the EPA lab,
or perhaps a manufacturer like AVS, involved in this at the Program level?
I've wondered if a manufacturer might not be willing to allow access (free
or cheap) to the software, and perhaps training, if the product is made available
throughout the Program. They would essentially be building up a base of users
from across the country. How many potential users are there within USGLOBEC?
Microsoft does precisely this on college campuses everywhere. I don't know
what the benefit to the EPA might be except that it would represent a technology
John A Quinlan MS-#38 | Email: firstname.lastname@example.org
Woods Hole Oceanographic Institution | Tel: (508)289-2238
Woods Hole, MA 02543 | Fax: (508)457-2169