Urban monitoring using NetKDE and VGI: network based kernel density estimation on volunteered geographic information applied to Baghdad, Iraq


  • Nicolas Lachance-Bernard
  • Timothée Produit
  • Loic Gasser




This paper presents a methodology for urban monitoring using volunteered geographic information (VGI) and journalism data Iraq war logs with network based kernel density estimation (NetKDE). It investigates, using spatio-temporal analysis, the evolution of urban events in Baghdad between 2004 and 2009. The extracted street network is based on the data distributed by OpenStreetMap (OSM). A total of 21,876 logged events, 66,648 network segments, 22,644 gridpoints (200m resolution grid) and 362,304 gridpoints (50m resolution grid) are used for the analysis. The methodology combines and adapts these VGI data and is mainly based on open source and/or publicly available software. It handles very large datasets with multiscale, multi-resolution and temporal perspectives. Fuzzy-set map comparison (FMC) is used to identify level of changes between each period of time. The methodology is already used in other fields of research being biology, urban planning, criminology or economic evelopment. It should help stakeholders in respective domain to analyze the evolution of network constrained events in multiple contexts. This paper is divided in three parts. Firstly, conceptual background of VGI, NetKDE and FMC is presented. Secondly, the methodology is illustrated using data Iraq war logs, OSM data and grids with two different resolutions. Thirdly, spatio-temporal analysis results are presented and discussed.