Report of the Data Visualization Meeting

December 20, 2000
Smith Conference Room
WHOI, Woods Hole, MA

Attending:

From the University of New Hampshire - Colin Ware, Roland Aresenau, Matt Plumlee

From Woods Hole (WHOI and NMFS) - Dezhang Chu, Peter Wiebe, David Mountain, Bob Groman, Skip Little, Maureen Taylor, Jim Manning, Cindy Chandler, Chris Hammond, Larry Madin, Warren Sass

Report prepared by David Mountain (Rapporteur) and Peter Wiebe

The Meeting:

Bob Groman introduced the meeting and the agenda. Introductions showed a mix of researchers and programmers. Peter Wiebe described the motivation behind the meeting. The data efforts in the GLOBEC and JGOFS programs are moving from managing data to visualizing it. Open market products do not meet the anticipated needs for visualization. The UNH group has developed an impressive set of visualization tools. A series of discussions have occurred over recent months between investigators at UNH and at WHOI, which resulted in planning this meeting. The intent of the discussion is to explore the needs and the capabilities required to meet those needs. In addition, DODS, which bears considerable similarity to JGOFS/GLOBEC data system, is becoming a national method for sharing/accessing data sets - and visualization likely will be an issue there, as well.

Peter gave an example of a visualization need. Given the GLOBEC Broad Scale data sets, he would want the ability (1) to see values on a section across the Bank, (2) to see a 2-D contour plot over arbitrary areas at an arbitrary depth level, and (3) to make 3-D realizations of selected variables, as well. We need a system that allows these type of visualizations be done easily from within the GLOBEC data system.

Bob Groman provided a summary of data that are now on line in the GLOBEC data system. A major point is that these represent a wide range of data types collected on a range of spatial and temporal scales. The data are irregularly spaced and are distributed on a number of servers.

Colin Ware gave overview of GeoZui3D (a "Fledermouse" deriviative). The system's features include -- rapid zooming, support for multiple objects in 3D; support for stereoscopic viewing; multiple inter-linked windows; multi-platform (Windows, Linux, Irix); OpenGL, C++, TclTk; easy to add new objects, real time input capacity (through CORBA).

He showed example of data from Juan de Fuca ridge segment -- with bathymetry in 3D, vehicle path as a tube in 3D, and other data objects. The system provided an easy, very quick ability move around the 3D space and to change angle of view. The system uses gridded data -- and can handle different spatial resolution in overlapping objects (data sets).

He also showed an example of application that used non-gridded data and gridded it on the fly as the data was read in. This seemed a particularly relevant capability for the GLOBEC data, which generally is not gridded. Another feature being added is an adjustable color pallette for the objects displayed.

The system does not do volume rendering.

The system does provide a "recorder" that remembers all the steps done, so that one can recreate what has been done.

A general discussion of needs, capabilities, and the difficulties getting the two to match then took place through lunch.
Some points from that discussion were:

• For 3D gridding, Chu is developing a 3D kriging algorithm which could be used to develop 3D gridded data. This then could be used for making planes through the data, and the planes (and multiple planes) could be viewed by GeoZui3D.
• Jim Manning identified two different needs. The first is to see in 3D what data are available for an area/time of interest. With that, the data to be analyzed could be selected. Then the second need is the visualization of the selected data -- the actual analysis.
• Another type of data is vector data -- such as ADCP current vectors. The system will need the capability to display vector data.

To identify where we want to go and the steps that need to be done, an outline was developed:

Need 1:

• Looking at where data are in time and space.


• Geographical representation of data locations - 3D.


• Temporal position of data.

Need 2:

• Kinds of visualization.


• Point source data.


• Section data.


• Plan view data.


• 3D data.

At present, the UNH system can provide 3D visualization of data provided in a gridded format. The major tasks are determining what data are of interest and getting them into a gridded format that would allow the desired visualization. UNH might focus on a small set of object types and WHOI/NMFS could focus on how to get data into form of one of these types.

Next Steps (for Need 1):

1. Improve time/space selection of the data (WHOI job).
Improve the "object of objects" capabilities of the JGOFS/GLOBEC system.
(Exactly what is needed here will need to be defined.)

2. Link selected data to visualizer (WHOI/UNH job).
Proof of concept first.
Iteration steps to select more/less data and visualize the result.

3. Refinement of 3d visualization capability (UNH/WHOI job).
Volume rendering - transparency.
Axes generation and data scaling.

4. Oncedata selected, then:
Data manipulation/transformation/interpolation.
Data projection - point or continuous.
Feature mapping.

5. Highresolution output for hardcopy (UNH job).

6. Save parameters to recreate image (state variables, scene file) (UNH job).

7. Color maps (UNH job).

Items 5, 6 and 7 all relate back to item 3.

For a time line, a preliminary estimate would be 3-6 months to deal with item 1 and 2, with progress being made along the way on the other items.

The specifications for computer requirements to use the system will need to be developed.

It was agreed to plan another meeting in late March, perhaps at URI, to show where we are to a larger group.

A final point was that some funding will be required for UNH collaboration and we need to look for a proper source for that funding.

Data Visualization Software Planning Meeting Participants

Meeting Agenda

1100     Introductionand background (P. Wiebe)

1115     GeoZui3D:System for rapid interaction with 3D data spaces (Seewww.ccom.unh.edu/vislab/GeoZui3D.html) (C. Ware)

1145     Summary of U.S. GLOBEC Georges Bank data sources (R. Groman)

1200     Lunch

1245     Discussion

Outline data groupings/characterizations

What data will investigators want to visualize?

What are the next steps?

1430     MeetingClose.

Meeting Product

Meeting Summary

Visualization needs identified.

Action items to address needs.