Geovisualisation Beyond Maps: Addressing Challenges in Visualising Geospatial Data Using Non-map-based Geography Encodings

Typically, visualisations of geospatial data are based on maps. There are many established types of thematic maps, e.g., choropleth or proportional symbol maps. However, particularly for complex datasets, such as network data [1] and large multivariate data (in which many variables in addition to geographic locations need to be visualised), maps suffer from various challenges such as overlapping symbols or markers, visual clutter, and geographic distortion. Beyond these visual issues, there may be additional considerations regarding the use of maps, for example, maps of COVID-19 case numbers have been shown not to increase public knowledge and behaviour [2].

This PhD project is based on the hypothesis that, instead of trying to mitigate and reduce these issues on map-based representations, they can (for many applications) be avoided entirely by using alternative visual encodings not based on conventional maps. Such non-map-based visualisation designs do not encode geospatial information as horizontal and vertical positions on a screen, but rather use alternative encodings based on distortion, abstraction, spatial ordering or grouping, colour-coding, and more. As such, they can use the position of elements on the screen to visualise attributes other than geolocation. This opens up opportunities for more flexible visual encodings that centre other aspects of the data, instead of the map-based layout dominating the visualisation. Therefore, they are mainly suitable for applications in which the geographic context is required as a reference, but not the primary information to be communicated. One major category in this space are schematized maps, which "[transform] cartographic maps to be more visualization-like, emphasizing the display of data over geographic accuracy" [3, p.1]. A number of examples of such techniques are included in [1] (geospatial networks) and [3] ('map-like' visualisations).

The first step of this project is to get an overview of existing non-map-based visualisations and visualisation techniques and to identify avenues for creating new methods. This could lead to a taxonomy or structured design space of non-map-based visualisations. Based on this, I plan to develop novel non-map-based visualisation techniques in the context of a case study and evaluate them using controlled quantitative and qualitative studies with experts and non-experts. Furthermore, I will explore strategies to teach users how to read these novel visualisations, as they may require support to understand unfamiliar visual encodings.

[1] Schöttler, S., Yang, Y., Pfister, H., and Bach, B. (2021) "Visualizing and Interacting with Geospatial Networks: A Survey and Design Space". To be published in Computer Graphics Forum.
[2] Thorpe, A., Scherer, A.M., Han P.K.J., Burpo, N., Shaffer, V., Scherer, L., Fagerlin, A. (2021) "Exposure to Common Geographic COVID-19 Prevalence Maps and Public Knowledge, Risk Perceptions, and Behavioral Intentions". JAMA Netw Open. https://doi.org/10.1001/jamanetworkopen.2020.33538
[3] Hogräfer, M., Heitzler, M. and Schulz, H.-J. (2020), The State of the Art in Map-Like Visualization. Computer Graphics Forum, 39: 647-674. https://doi.org/10.1111/cgf.14031