The story of this talk starts eight years ago, with the creation of a mapping-interface [1] to explore one of the fined grained statistical dataset on the french population (number of inhabitants / km², the average income per consumption unit, the percentage of low-income households, ...). These data were derived from localized tax revenues of households and were available in a very fine way i.e on pixels of only 200m x 200m, they were delivered at this time by the French National Institute of Statstics (INSEE) throught a massive spatial data file, which limited their use to a very limited number of people able to handle it. The proposed interface changed this and allowed almost any-body to visualize easily these data in their browser. In this presentation we will mainly discuss how to make such massive spatial data easily accessible to the public. The proposed solution is based on multi-scale aggregation: at high scales, the data are aggregated and at fine scales the final resolution of 200m by 200m is recovered. This interface, which heavily rely on zoom in/zoom out interactions allows users to deal and play with one of the well known problem in geography and spatial statistics namely the Modifiable Areal Unit Problem [2] (ie. Data tabulated for different spatial scale levels or according to different zonal systems for the same region will not provide consistent analysis results). This tool has leads to several interactions with new users (journalist, researchers, simple citizen,...) that seem to demonstrate its interest. Since this first project, several other massive datasets (one concerning an updated of the same data-set and one concerning a dataset on all french building) were made available with almost the same system [3,4]. From a technical perspective, this talk will covers the different evolution and open-sources projects that has made this project possible (d3 / postgres / postgis / mapLibre,Leaflet, ...), discussing vector tiles formats, map viewer evolution and a little bit of spatial databases and the development of a little python tool to help other build the same type of visualization.
[1] Francepixels, https://www.comeetie.fr/galerie/francepixels
[2] D.W. Wong, Modifiable Areal Unit Problem, Editor(s): Rob Kitchin, Nigel Thrift, International Encyclopedia of Human Geography, Elsevier, 2009, Pages 169-174,
[3] Francepixels2019, https://www.comeetie.fr/galerie/francepixels209
[4] FrancepixelsBati, https://www.comeetie.fr/galerie/francepixelsbati
Speakers: Etienne Côme