Incorporating Uncertainty into 3D Spatial Heatmaps for Risk Visualizations in the Built Environment

Abstract

When analyzing the distribution of climate, health or similar risk-related data in the built environment, we often involve spatial heatmaps that are placed over or between existing environmental geometry. Common examples of this are indoor air-quality visualizations or city-scale maps of flood risk. These heatmaps can be based on simulations, interpolated measurement data, or other probabilistic methods that turn limited data into full spatial coverages. This means that beyond the visualized risk factor, there is always a measure of uncertainty in the data. While there has been research into showing this uncertainty on spatial heatmaps, such techniques have rarely been applied in urban scenarios with detailed building geometries. These environments introduce occlusion and other viewpoint-related issues and thus make existing cartographic techniques less effective. In this paper, we want to develop visualization strategies that effectively show uncertainty on spatial heatmaps, while circumventing issues of occlusion and viewpoint-dependency. To do so we collect common uncertainty visualization methods from the literature and conduct a preselection for this use case. We then evaluate the effectiveness of each method based on an example scenario, discussing any performance and readability issues that arise. Finally, we recommend certain configurations of methods that strike an appropriate balance between the chosen quality measures.