Using Vis5d+ to Visualize Five Dimensional Data on a Three Dimensional Grid
Prepared for: Carmen Benkovitz
By: Ron Huegel
August 10, 2002
Data visualization is a technique for visually exploring physical (temperature, pressure, viscosity, etc.) processes in science and engineering. The development of efficient visualization software tools and techniques (algorithms) continues to be an active area of research. Computers continue to grow in speed, memory, and functionality in accordance with Moore’s Law, which states “chip density (roughly) doubles every eighteen months” (Boyd). This fact should suggest that the software used by computers must grow at a proportional pace. Moore’s Law has held true for over twenty years despite skepticism from some physicists. There is no law or equation governing the growth of software; however, the graphics capabilities of PC’s and video games have grown rapidly over the past several years.
Software programs are a very necessary component of any experiment in modern day science. Without any way of reading and displaying data, science would slow to an absolute crawl. There are currently countless numbers of software programs available on the open market, and there are also many programs that can be downloaded from the Internet at no cost. These cost-free downloads are known as “open source” software.
Open source is not a well-supported means by which to operate a software package. Much of the time software that is open to the public is left unsupported or poorly supported by the author or authors. Then again, much of the software that we buy in the stores is not very well supported either. That is not to say that open source is not quality software. There are plenty of high quality programs ready to be downloaded from the Internet. Sometimes the people currently using the program may even support a software package by communicating problems and insight via e-mail or through web site bulletins.
Vis5D is an open source software package currently available for download from the Internet. Some scientists who display weather data are currently using Vis5D. The Visualization Project at the University of Wisconsin-Madison Space Science and Engineering Center (SSEC) by Bill Hibbard, Johan Kellum, and Brian Paul wrote the software. Their goal was to “design a means by which to interactively visualize large 5-D gridded data sets such as those produced by numerical weather models” (“Vis5D”).
The goal of this project is to determine if Vis5D+ is a feasible software program for better analyzing model output data (MOPUT) being mapped at Brookhaven National Laboratory. The data comes from Carmen Benkovitz of the Brookhaven Department of Environmental Sciences division of Atmospheric Sciences.
The National Weather Service has been using Vis5D+ to visualize various weather variables such as humidity, wind vectors, and cloud formations for the past several years. This data is similar to the data processed by Carmen Benkovitz and her team, so Vis5D+ may be a good tool for visualization of MOPUT.
Materials and Methods
Downloading and compiling the Vis5D+ software proved to be quite a challenge and required the help of a computer programmer. The downloading and uncompressing went smoothly; however, when it came to compiling the program, several errors were encountered. With some help, it was determined that a missing library, which was not installed on our system, caused the problem. Since this library was not critical to the application of Vis5D+, the library was excluded from the compilation process. Subsequently, there were no other problems compiling Vis5D+.
At this point, Vis5D+ was run with several sample data sets that were downloaded from the Internet. These samples ran well; the only glitch at this point was a “reverse” function in the Vis5D+ user interface. This function is supposed to reverse white and black colors. This feature is not a necessity for the experiment and has not been addressed to date.
The next obstacle was to convert data to a format that can be read by Vis5D+. With help from Robert Bennett, who determined the data ordering and offsets, the MOPUT was successful converted to Vis5D+ format.
Once the data was converted to a format that could be read by Vis5D+, several tests on the program were run to exercise the capabilities of visualizing the converted MOPUT data. The “reverse” function was already lost and these tests revealed a problem with “cloning” variables. The “cloning” function is an important function for this project and therefore required further investigation. Upon further investigation it was determined that there was an alternative means by which we could clone variables. This was accomplished by writing a mathematical expression using the “NEW VAR” function. By setting the new variable to be equal to the existing variable (i.e. VARIABLE = VARIABLE*1.0)the limitation could be overcome. All other functions worked as described in the documentation.
During this testing period, different versions of Vis5D were explored/tested to determine which one is best for our data and computing platform. Preliminary results showed better compatibility between Vis5D-5.0 and our SGI machines. This changed since finding a way to clone variables on Vis5D+. Vis5D+ also has a function for creating VRML files; this feature is non-existent on earlier versions.
A preliminary evaluation by Carmen Benkovitz led to the next experiment. Carmen felt that a logarithmic color scheme would show her data better. Again with Robert Bennett’s help, it was found that there was a way to incorporate such a color scheme into Vis5D+. Since the science team was already using a software program called PV wave, it was determined that the color scheme from that existing program would be integrated into Vis5D+. Although the color scheme worked, it was found that the default linear color scale could be manipulated to produce an acceptable range of colors.
Recommendations and Conclusion
This research demonstrated that Vis5D+ is an effective tool for visualizing large data sets. Vis5D+ provides the basic functionality for scientific visualization. In addition to volume rendering capability (Figure 1), Vis5D+ also supports iso-surface generation (Figure 2) and volume slices in the vertical and horizontal directions.
Figure 1 shows mixing of sulfates produced by North America and present over approximately a 4-day period. The color table gives values to the colors (data from Benkovitz et al in preparation).
Figure 2 shows four cylindrical views of the northern hemisphere with iso-surfaces representing sulfates by volcanoes (Data from Benkovitz et al in preparation).
Important for scientific evaluation of data is the “probing” function (Figure 3) that allows the user to display all variable values for a point in time and space.
Figure 3 shows a volumetric view of H2O2 being probed by the Vis5D probing function. Notice the list of values with their corresponding values (data from Benkovitz et al in preparation).
Vis5D+ also incorporates a few different map projections (figures 1, 2, & 3) that help to account for the curvature of the Earth, which can distort data. Also, tests were able to demonstrate the flexibility of Vis5D+ by visualizing a non-meteorological data set, specifically, structural MRI data sets. Both programs generated volume and iso-surfaces from the MRI data, thereby illustrating the use of the software outside of the MET domain. “Vis5D allows users to visualize data in a straightforward GUI environment. The program gives users freedom to customize the visualization so every aspect of the simulation can be explored” (“Visualizing”).
Boyd, Charles. “Moore’s Law.” 30 April 2002. Strategic Management. 9 Aug. 2002
“Vis5D.” 17 September 1998. University of Wisconsin-Madison Space Science and
Engineering Center. 10 Aug. 2002 .
“Visualizing High-Resolution Climate Data.” 2 Aug.2002. Springer. 10 Aug. 2002