Certain Uncertainty - Mapping Global Water StressA Physical Data Experience


Featured at Re:publica 2023
Nominated for the Information is Beautiful Awards
Selected for the 2023 Museum of Wild and Newfangled Art Biennial
Featured at CHI 2023 – WS 10: Data as a Material for Design
Featured at Datavisualization Society's Lightning Talks series
Featured at the Exhibition 'Loosing Earth' at Kunsthalle Düsseldorf

The Project

Within the certain uncertainty project, we developed a novel approach to visualize water stress in a geospatial context in relation to population density. Water stress or scarcity does always need to be reflected in context. While some parts of the world are only sparsely populated, the impact and mitigation of water stress in densely populated areas are potentially critical. While most water stress mappings focus on communicating the water stress within a tempo-spatial context, this project aims to map the water stress of selected capitals within the context of the global population density to enable the viewer to explore the interaction between both dimensions in a meaningful way. By focusing strongly on the data and removing all cartographic borderlines, an abstract space of mountain ranges remains, showing the world as a spatial accumulation of humans confronted with increasingly changing environmental conditions. In this paper, we describe the stepwise development process of the artifact starting data processing and how the individual physical layers were created.

Related Publications

Certain Uncertainty – A Geospatial Data Physicalization of Water Stress and Population Density
K Schroeder, S Jules
CHI’23 WS 4: Data as a Material for Design: Alternative Narratives, Divergent Pathways, and Future Directions, Hamburg 2023


In order to realize the installation, population density mountains must be created. This first requires data that can be used to design this part of the model. A render based on data from Nasa was used. This visualization shows the population density in the world by color (heat map). The image as a data file was used as input for generating the 3D model with Houdini. The figure of the data tree in Houdini shows how the image is directed to the mountain landscape; this tree can be divided into four segments. In the first part, the visualization is converted into points that have the color of the underlying pixels, this color is a vector.

Based on the length of this vector, the points are moved in the normal direction, creating a mountain landscape. Then the mesh of this mountain range is rearranged so that the vertices are more evenly distributed, which is necessary for a clean result. The mountain range is then sliced to reveal the layers. In the final stages, the flat layers (cuts) only need to be extruded according to the thickness of the layers. After that, the model is ready to be produced. Before the model can be cut out, it has to be reorganized so that the CNC cutter understands how to cut it out. Each layer is cut out of a separate plate, for which the Fusion 360 software was used.

This software wrote a gcode that allowed the cutter to cut out the model. After all the models were cut out, the parts could be assembled into the landscape. We used a overhead projection of the geodata mapping to accurately position the individual layers of the population density map on their respective positions. In a next step all individual parts were glued together with wood glue.