HexTiles and Semantic Icons for MAUP-Aware Multivariate Geospatial Visualizations

TL;DR

This paper introduces HexTiles, a hexagonal tiling visualization method for multivariate geospatial data that incorporates semantic icons and confidence encoding to improve interpretability and address MAUP effects.

Contribution

HexTiles is a novel multivariate geospatial visualization design that uses semantic icons and confidence encoding to enhance interpretability and mitigate MAUP effects.

Findings

Positive user study feedback on HexTiles effectiveness

Domain experts found HexTiles improved data interpretation

HexTiles effectively visualizes data variability within geospatial areas

Abstract

We introduce HexTiles, a domain-agnostic hexagonal-tiling based visual encoding design for multivariate geospatial data. Multivariate geospatial data have presented a challenge due to the graph schema associated with geospatial maps, on which most geospatial data is presented. With HexTiles, we design a multivariate geospatial visualization design that leverages semantic icons to (1) simplify the process of interpreting interactions between multivariate geospatial data, and (2) put the visualization designer in the driver's seat to guide user attention to specific variables and interactions. Additionally with HexTiles, we attempt to explicitly mitigate effects of the Modifiable Areal Unit Problem (MAUP) for interpreting geospatial data, by proposing a confidence encoding for each of the information channels in HexTiles. We calculate weighted variances of the variables in each HexTile to provide a confidence value for each tile, which can be used to interpret the variability of the data within the corresponding geospatial area, an information that can be lost in geospatial visualizations. To validate our approach, we gather quantitative and qualitative feedback from a user study and document domain expert feedback from ecologists and hydrologists experienced in designing geospatial visualizations.