Visualising Attribute and Spatial Uncertainty in Choropleth Maps using Hierarchical Spatial Data Models

Attribute and spatial uncertainty are defined and put into context for this research. This paper then extends on a research programme which has designed a visualisation of attribute and choropleth spatial uncertainty using the Hexagonal or Rhombus (HoR) hierarchical spatial data structure. Using the spatial data model in this fashion is termed – the trustree. To understand this progression, a brief explanation of this research programmes past history must be covered. The New Zealand 2001 census is used as an exemplarity dataset to express attribute uncertainty and choropleth boundary uncertainty (termed spatial uncertainty). An internet survey was conducted to test the usability of the trustree, which was used as a transparent tessellation overlay and a value-by-area (VBA) display within a population choropleth map. Two other visualisation of attribute uncertainty methods – blinking areas and adjacent value were also incorporated into the survey. Participants were required to rank, from 1 to 6, six grid cells which overlaid the uncertainty visualisations, in order from the most accurate to the most uncertain cell, respectively. These ranking results were correlated with the actual ranks, providing a metric of usability for each visualisation method. The blinking areas method was the most effective, followed by adjacent value, VBA trustree and the transparent HoR trustree. The time taken for a participant to rank each visualisation’s cells was collected – there is an 82 % correlation between the time taken and the final usability results obtained