Lake Michigan Flow Field Visualization

In this study, data from a computational fluid dynamics (CFD) simulation was used to investigate the feasibility of VRML, the Virtual Reality Modeling Language on the World Wide Web, for visualizing three dimensional fluid flow. From a given data set, an animation representing the flow was created and was placed inside a virtual model of Lake Michigan.

The data set was supplied by the Great Lakes Environmental Research Laboratory (GLERL), and represented simulated particle releases and time-domain tracking locations in the south-western quadrant of Lake Michigan. The data described the release of three particles from three locations in the lake at three hour intervals. For a total simulation time of 60 days, there were 480 of these intervals, 1,440 particles in total, with locations given at each of these intervals for a sum of 691,200 three space locations. Lake Michigan was divided into a grid of 53 by 102 five kilometer by five kilometer squares. Each three space location was given by grid location and the particle's actual depth. In addition, GLERL also supplied elevation data for Lake Michigan. This data gave the depth of the lake at the center of each of the grid squares. Using all of this information, a VRML model of Lake Michigan (color-coded with respect to depth) was created with the particle animation placed inside at the correct location.

Interactive buttons on the screen allow for control of parts of this VRML model. These include an on/off switch for visualizing the source locations by spheres, a slider to control the transparency of the lake's geometry, and an on/off switch and control for a vertical plane that can be moved up and down. By adjusting the transparency, the flow animation can be observed from viewpoints outside the lake while allowing the lake to remain visible enough to provide a reference frame. The plane allows for part of the scene to be isolated for the study of details. As the plane is moved, the corresponding depth is displayed at the bottom of the screen.

Views from pre-defined viewpoints:

Web size (54K)
Full size (160K)

Web size (36K)
Full size (110K)

Web size (84K)
Full size (245K)

VRML models:

VRML (Virtual Reality Modeling Language) is an international standard that allows for the distribution of interactive and functional three-dimensional models over the Web. A user, connected to the Web, can access a VRML application and navigate through or interact with the model. To do this, a VRML plug-in for the Web browser in use (Netscape or Internet Explorer) is required. We recommend to download the CosmoPlayer plug-in to view the developed models.

Load the full version of the VRML 2.0 animation (9MB).
Shows the movement of all 480 releases (1,440 particles)

Load a smaller version of the VRML 2.0 animation (1MB).
Shows the movement of the first 30 releases (90 particles)

Additional Information:

The data set supplied by GLERL was in excess of 13MB, and far too large to be used on many computers. The data was edited, optimized and incorporated into a VRML file without loss of any information using a specially developed Perl 5 script.

The principles used in creating the VRML application are explained briefly on the separate page "Data Conversion Process", and additional information on data formats can be found at "Data Format Documentation."

Transparent lake geometry
Web size (41K) - Full size (136K)

Streak line visualization
Web size (15K) - Full size (75K)


The computational fluid dynamics model of Lake Michigan was developed by Dr. David J. Schwab at NOAA's Great Lakes Environmental Research Laboratory (GLERL) along with Dr. Dmitry Beletsky from the University of Michigan's Department of Naval Architecture and Marine Engineering and the Cooperative Institute for Limnology and Ecosystems Research (CILER).

The VRML application was created at the Virtual Reality Laboratory by undergraduate student Joshua Spain Levi with assistance from Lars Schumann and supervision from the Lab's director Klaus-Peter Beier. The development of the VRML model was supported in part by a mini grant from CILER.

Last Update: August 5, 1999 by jlevi
Return to VRL Home Page
Return to Project Overview