Visualization of attributed hierarchical structures in a spatiotemporal context

Abstract

When visualizing data, spatial and temporal references of these data often have to be considered in addition to the actual data attributes. Nowadays, structural information is becoming more and more important. Hierarchies, for instance, are frequently applied to make large and complex data manageable. Hence, a visual depiction of hierarchical structures in space and time is required.

Although there are several techniques addressing specific aspects of spatio-temporal visualization, approaches that cope with space, time, data, and structure are rare. With this article we take a step to fill this gap. By combining various well-established concepts, we achieve a reasonably complete visualization of all of the aforementioned aspects, where our focus is on the hierarchical structures. We embed hierarchies directly into the regions of a map display using variants of the point-based layout. Layering and animation are applied to visualize temporal aspects. Based on the analysis goals, users can switch between representations that emphasize data attributes or hierarchical structures. Interaction techniques support users in navigating the data and their visualization. We demonstrate the usefulness of our approach by adapting it to implement a visualization for spatiotemporal human health data.

Keywords:

• hierarchy visualization
• geovisualization
• time visualization

Acknowledgment

We gratefully acknowledge the support by ‘VisMaster’ and ‘dIEM oSiRiS’.

Notes

1. The data hierarchies we consider here are not related to a hierarchical subdivision of space as commonly seen in GIS, but may be arbitrary hierarchical structures (e.g., structure of time, clustering hierarchies, or organizations).

2. These two factors describe the relation between the different regular grids shown in Figure 2 and guarantee a space filling placement. More information on this and other pairs of factors can be found in Schulz et al. (2010b).

3. Searching for subareas of size has proven practical.

4. We suggest a maximum depth of three to avoid too small subtrees at lower levels.