Visualizing time series using small multiples in a spatial context is a challenging task, which requires careful balance between the amount of depicted data and perceptual precision. A well-known technique for compactly representing time-series data are horizon graphs, which provide fine details while giving an overview of the data where extrema are emphasized. They compress the vertical resolution of the individual line graphs, but they do not affect the horizontal resolution. We present two variations of a new visualization technique called “collapsed horizon graphs” that extend the idea of horizon graphs to two dimensions. Our main contribution is a quantitative evaluation that experimentally compares four visualization techniques with high visual information resolution (compact boxplots, horizon graphs, collapsed horizon graphs and braided collapsed horizon graphs). The experiment investigates the performance of these techniques across tasks addressing both individual graphs as well as groups of adjacent graphs. Our results show that collapsed horizon graphs are similarly effective while providing an increased horizontal resolution. Moreover, they indicate that the visual complexity of the techniques highly affects users’ confidence and perceived task difficulty. The collection bundles supplement materials and research data from the conducted user study to ensure reproducibility. Specifically, the collection includes a user study prototype, exemplary pictures from the study design, the collected results as well as the analysis scripts.

visualization, user study, evaluation, time series, spatial data, collapsed horizon graphs, compact boxplots, braided graphs