Bleile, Antonio (2002)
Semantics Based Management and Visualization of Dynamic Atmospheric Data.
Technische Universität Darmstadt
Diplom- oder Magisterarbeit, Bibliographie

Kurzbeschreibung (Abstract)

When the semantics of time-variant three-dimensional atmospheric data is correctly considered and described by mathematical means, new methods of efficient data compression, time interpolation, and visualization can be developed and applied. This thesis will investigate the nature of simulated viscous masses, take such a semantic approach, demonstrate its benefits, and compare its results to currently comparable solutions. Visualizing dynamic atmospheric data has its very own challenges. Mostly, large data-sets must be stored in main memory to access them quickly for interactive graphics. Fast mapping algorithms are applied to these data-sets in order to produce geometric objects for direct viewing or for the offline production of animations. All currently applied methods of data storage (arrays with high or reduced numerical accuracy) and of visualization (iso-surfaces or direct volume rendering) as well as time-interpolation (linear or higher order) have significant and well-known shortcomings. The reason for this is the fact that they do not take advantage of the specific characteristics of dynamic atmospheric data. It must be expected that correctly considering the semantics of this special class of data will lead to new and appropriate solutions. When this data is understood more as moving and at the same time dynamically changing air masses rather than as discreet and seperately treated time steps of large arrays, new algorithms can be developed that follow these approaches: - compress the data with algorithms found in the fields of image or movie compression which take a while for the compression step but are very efficient and support rapid decompression during the visualization phase - interpolate data in time for smoother animations by following moving air mass "volumes" through the atmosphere. Without applying the full simulation model with smaller increments in time, in-between time steps can be computed by considering the specifics of animated viscous masses. - visualize with regard to the three-dimensional space and the dynamic nature of the data by effectively applying modern culling methods during the data filtering and data mapping stages of the visualization pipeline. Thus, this new approach in the treatment and internal representation of the data leads to innovative new possibilities for the visualization of dynamic atmospheric data. This thesis will develop such an approach and implement prototypes of software modules for the areas listed above. The results are compared to the state of the art.

Frage zum Eintrag

Redaktionelle Details anzeigen