Preprocessing Open Data for Optimizing Estimation Times in Urban Network Analysis: Extracting, Filtering, Geoprocessing, and Simplifying the Road-Center Lines


Hacar M., Mara F., Altafini D., Cutini V.

Lecture Notes in Civil Engineering - Innovation in Urban and Regional Planning, Alessandro Marucci,Francesco Zullo,Lorena Fiorini,Lucia Saganeiti, Editör, Springer Nature, Zug, ss.551-562, 2024

  • Yayın Türü: Kitapta Bölüm / Araştırma Kitabı
  • Basım Tarihi: 2024
  • Yayınevi: Springer Nature
  • Basıldığı Şehir: Zug
  • Sayfa Sayıları: ss.551-562
  • Editörler: Alessandro Marucci,Francesco Zullo,Lorena Fiorini,Lucia Saganeiti, Editör
  • Yıldız Teknik Üniversitesi Adresli: Evet

Özet

Road networks are crucial datasets for urban and regional spatial analysis, yet their large size and complexity make them a computational challenge, particularly for methods such as Space Syntax’ Angular Analysis, which break the network according to road direction changes for more precise potential movement pattern estimations in each road element. This becomes difficult to apply in certain practical situations, for instance, digital twins, due to the considerable time required for modelling, which can pose computational resource limitations. In this study, we propose a rule-based approach for pre-processing OpenStreetMap (OSM) data to construct road networks suitable for Space Syntax analysis. Our approach extracts OSM data within an administrative area of interest from GADM database, filters by regarding tags, and then goes in two directions depending on whether the user prefers to keep semantic tags: preserving intersections or creating an arc-node structure, respectively. Finally, road networks are simplified to reduce the estimated time that the network will take to perform the analysis. We applied this approach to OSM data from six cities in Türkiye and Italy and evaluated in terms of computational time estimation in DepthmapX 0.8. Results show that simplifying road networks as a lane-width (2m) threshold in Douglas-Peucker algorithm maintains their basic structure (i.e., configuration, morphology, and movement patterns) and relationships between nodes, while reducing computation time significantly. Our approach provides a scalable and efficient solution for urban network analysis and has the potential to be applied in other fields that require the processing of large and complex datasets.