| With the continuing development of sensor and computer technology, collected data are becoming richer and with increasing complexity. Data produced from remote sensing, geophysics, demography, meteorology and bioinformatics are often presented so as to show the variation in values taken by some variables. In this document, we define this emerging new class of data as multi-valued data, in which there are multiple values about a variable at each location and time. It is important to draw the attention of the visualization community to such data and address the challenge of handling them with efficient solutions.; Existing visualization techniques have limited power in handling multi-valued data. We propose an operator-based approach for visualizing multi-valued data. Instead of developing specialized visualization techniques, we apply operators to the data so that they can be easily visualized. This operator-based approach acts as a bridge between the data and available visualization tools. With the assistance of appropriate operators, isosurfaces, contour lines, pseudo-coloring, streamlines and pathlines on multi-valued data become straightforward. In addition, guidelines on how to select operators for particular tasks are provided. Operators discussed in this document are by no means exhaustive, in that different features from different applications may call for different operators.; We are the first to define spatial multi-valued data, which opens up a new area into which scientific visualization techniques can expand. By drawing the data into a class of its own, we simplify visualization problems on handling such data. Furthermore, we address the challenge and provide a systematic solution to the problems. This work initiates an exciting research topic and it is a primary step on the road toward advanced data exploration through scientific visualization. As a consequence of this work, many tasks associated with this new type of data become feasible, for example, comparisons and feature extractions on multi-valued data are demonstrated in this document. |
