Volume probes: interactive data exploration on arbitrary grids

Abstract

A taxonomy of computational grids used in scientific and engineering practice is presented and a technique for cutting them by, and displaying data on, 2D surfaces is developed. When sliced by a surface, these grids give rise to a graph G(C, F) where C, the nodes, are the intersected cells and F, the arcs, are their connectivity across faces. Starting from any cell known to be intersected by the surface (a seed), G is traversed breadth-first and is constructed locally on the fly, that is, only the spreading "front" explicitly exists at any time. Only sliced cells are visited, shared computed values such as edge intersections are passed to neighbors, and most of the geometric work is done via table lookup. A seed cell is found by fence-hopping from any cell to a distinguished point on the surface.This means of slicing grids is then utilized in an effective visualization tool. Concentrating on planar surfaces, local coordinate systems are defined for constructing clipping windows and linear transformations within the planes which further reduces display time and allows effects such as zooming within the windows. Several of these planar windows are then organized into various objects, called probes, that can exploit the mind's "retinal memory" when repeatedly swept through amorphous data.