Physically Plausible Weather Visualization
Master Degree Thesis
by Gisle Nes (gislenii.uib.no)
supervised by Ivan Viola
Abstract
Physically plausible visualizations of weather-phenomena such as hurricanes or cloud structures are difficult to achieve at interactive frame rates. The visual properties of clouds depend on how light is scattered through the media, as well as the amount of light that is scattered towards the viewer. The fact that all particles can scatter light in all directions makes the light calculation very complex.
Visualizations of weather phenomena are desirable for several applications, but this project has been carried out mainly with television broadcasters in mind. A big interest for television broadcasters is weather visualizations during the hurricane season. Because of the large interest for hurricane visualizations, the main scope of this thesis is to visualize simulated hurricane data sets.
This thesis presents a weather visualization system that utilizes knowledge about the physics of clouds in order to render physically plausible weather visualizations. The interaction between light and clouds is imitated by considering self shadowing and multiple forward scattering. The speed and parallelism of modern graphics hardware enables visualization of large weather data sets at interactive frame rates.