Understanding Spatiotemporal Patterns: Visual Ordering of Space and Time

Abstract

Deriving patterns and relations from large multivariate and multi-temporal data sets to acquire knowledge about real-world processes is not a trivial task. To understand the content of such data sets, current analytical tools do offer interesting solutions, but an approach combining the above different types of data is lacking. This article introduces a visual integrated solution that allows the user to explore and analyse the data at hand. The approach introduced consists of a dynamically linked multi-view environment that offers different interactive visual representations to “look at and play with” the data. For the time component, the temporal ordered space matrix (TOSM), which schematizes the temporal nature of the data set, is introduced. The rows of the matrix represent time and the columns the geographic units. A preliminary usability test has been conducted to see how the multi-view approach in general performs when considering specific tasks oriented toward the understanding of spatiotemporal patterns. The TOSM functions well for naturally ordered (linear) phenomena such as rivers and coastlines. The article also discusses the use of the TOSM for non-linear-ordered phenomena such as administrative units. The method is based on directional ordering and is compared with other ordering approaches, such as space-filling curves, the travelling salesman problem, and plane-sweeping algorithms.