Efficient Visualization of Geo-Physical Mass Flow Simulations
A computational tool called the TITAN2D has been developed by an interdisciplinary team from the departments of geology, geography, mathematics and mechanical engineering at the University at Buffalo to answer this question by modeling and simulating geo-physical mass flows from volcanoes. The system consists of a parallel hp-based FEM code that takes as its input Digital Elevation Models (DEM) of the area of interest and huge datasets are produced by the parallel simulation of the flow that often exceed gigabytes. To visualize these huge datasets in a meaningful and efficient manner, we have developed a viewer for the TITAN2D system which employs various techniques such as mesh decimation, image processing and level of detail algorithms.
The viewer is designed to aid end-users such as volcanologists with a clear representation of various properties of the mass flow such as pile height and velocity, and their relation to the terrain over which the flow is occurring. By overlaying the terrain mesh with satellite imagery of vegetation, population distribution, etc., hazard prediction is possible. In designing the viewer, we have used texture and color to visualize the attributes embedded in data elements which constitute the polygonal mesh. Attributes can be viewed as arrayed across the surface and can be appropriately color-coded to produce meaningful patterns.
The input to the TITAN2D system is a DEM from the GRASS (Geographic Resource Analysis Support System) data geographic information systems (GIS) data source. This GRASS data is coded into a grid and the grid is locally adapted during simulation to capture the flow features as the flow evolves. As the simulation advances, the adaptive mesh computational algorithm refines the grid into finer resolutions producing large datasets of the order of several gigabytes. Attributes such as pile height and velocity of the geo-physical mass flow are embedded in each element of the grid. An image is generated by mapping an attribute encoded in each element of the dense flow mesh, pixel by pixel to create a color contour of that attribute. Pixels in the image that do not correspond to an attribute are made transparent. The images are overlayed on the DEM.
Dr. T. Kesavadas (Dept. of Mechanical Engineering)
Padmanabhan MuraleedharanNair (Dept. of Computer Science Engineering)
Navneeth Subramanian (Dept. of Mechanical Engineering)
Amrita Chanda (Dept. of Computer Science Engineering)
UB Geophysical Mass Flow Group
National Science Foundation (NSF)
VR LAB 809 FURNAS HALL, THE STATE UNIVERSITY OF NEW YORK AT BUFFALO, BUFFALO, NY 14260 | (716) 645-2593
VR LAB (C) 2003-2004, ALL RIGHTS RESERVED