1 An evaluation of the OSM dataset for sustainable transport planning RGS-IBG 2015 02/09/2015 Robin Lovelace 18 1

2 Contents ● Ordnance Survey data ● Open Street Map data ● Comparisons and issues ● Applications ● Conclusion and further work

3 Context ● Open Street Map (OSM) data found to be approaching Ordnance Survey (OS) data quality on UK road network (Haklay, 2010) ● Focus on augmentation official datasets (Du et al., 2012) ● OSM cycling data found to growing rapidly, outperforming Google Bicycling layer in many locations (Hochmair et al., 2012) ● Intersects with wider debate about merits of Volunteered Geographic Information (VGI) (Goodchild, 2007)

4 1: Ordnance Survey (OS) data “OS MasterMap ITN Urban Paths is a new data Theme depicting man-made path networks in urban Great Britain ... designed to work with OS MasterMap ITN’s established Road Network Theme.” http:/ /www.ordnancesurvey.co.uk /busin ess-and-government/help-and-support/products/urban-paths.html 1.5 Gb of data provided by OS under strict academic license

5 OS data: key attributes ● Hard to load in anything other than ArcMap (comes in hundreds of .gz files -> .mdb format) ● Mostly walking paths: 98% are 'Footpath' ● By distance, cycle paths constitute 4.3% of dataset: 2,500 km ● Almost exclusively concentrated in urban areas ● Very few attributes: Footpath, Cycle Path, Canal Path, Bridleway

6 Milton Keynes Very good coverage in some areas...

7 Birmingham No cycle paths at all in UK's 2nd largest city

8 Open Street Map (OSM) data

9 Accessing OSM data ● British Isles downloaded in .pbf format (<1 Gb) from Geofabrik ● osmconvert used to convert to .o5m file type (15 Gb) ● osmfilter extracted cycleways, bicycle=yes and cycle path relations, e.g.: – Osmfilter map.05m –keep=”highway=*” >ways.osm – Osmfilter map.05m –keep=”highway=cycleway” >cways.osm ● Overpass API can be used to access data directly ● .osm > .db sqlite database. Filtered tags w. QGIS.

10 Overview of OSM data Tag T. Length (1000 km) Av. Length (m) Colour highway='cycl eway' 13.6 260 Black bicycle=... 6.3 390 Blue lcn='yes' 5.4 260 Red ncn='yes' 1.1 320 Green rcn='yes' 0.3 530 Yellow Total merge 24.6 280 - • ~80,000 multi-node lines • 1.5 Gb storage, including tags • 14 Mb shapefile (no attributes) • 879,062 nodes • ~30 mins to load on 4th gen i7 processor (in R)

11 Attribute data in OSM ● Total number of 'highway=cycleway's: 52,700 ● Foot= 22,639 ● Surface= 16,312 ● Bicycle= 14,035 ● Segregated= 10,744 ● Lit= 10,545 ● Many more, relating to quality, law and use

12 OSM and OS data compared Attribute Ordnance Survey OSM Accessibility Unavailable to public, available for sale (price unknown) Freely accessible, provided skills Network size 2,500 km 13,600 km (just 'highway=cycleway') Coverage Very patchy, lacks continuity, only urban areas Worldwide but dependent on volunteer labour Attributes None Very wide range of attributes allowed

13 Issues: diversity and inconsistency ● Towpaths, paths, ncn, lcn, bicycle=... ● Dozens of tags related to cycling ● Solution: moves towards consolidation

14 OSM data for Cycle Campaigners

15 OSM data for transport planning Import of OSM data into GIS/CAD systems Integration of how public 'sees' infrastructure into public/private plans Huge opportunities for community engagement Example: Leeds-Bradford Cycle Superhighway Custom OSM map with Leaflet interface used Collection of opinions about proposed route Stopped short of encouraging altering vector data Field testing needed for full evaluation Use in CycleStreets.net for DfT project (current project)

16 OSM data for public consultation

17 Conclusions and further work ● Open Street Map outperforms Ordnance Survey for provision of cycle path data ● Wider point: crowd sourced data as THE standard? ● Encourages public participation ● Plans for further analysis and publication: – Spatial comparisons where OSM/OS overlap – Framing as crowd-sourced data becoming mainstream – Aiming for 'public engagement' as well as academic outputs – E.g. Lovelace (2014), CycleStreets.net blog, CTC – Making OSM data accessibe (see next session!)

18 References ● Du, H., Anand, S., Alechina, N., Morley, J., Hart, G., Leibovici, D., … Ware, M. (2012). Geospatial Information Integration for Authoritative and Crowd Sourced Road Vector Data. Transactions in GIS, 16(4), 455–476. doi:10.1111/j.1467- 9671.2012.01303.x ● Goodchild, M. F. (2007). Citizens as sensors: the world of volunteered geography. GeoJournal, 69(4), 211–221. doi:10.1007/s10708-007-9111-y ● Haklay, M. (2010). How good is volunteered geographical information? A comparative study of OpenStreetMap and Ordnance Survey datasets. Environment and Planning B: Planning and Design, 37(4), 682–703. doi:10.1068/b35097 ● Hochmair, H., Ziestra, D., & Neis, P. (2012). Assessing the Completeness of Bicycle Trail and Designated Lane Features in OpenStreetMap for the United States and Europe, 1–21. ● Lovelace, R. (2014). Harnessing open street map data with R and QGIS. EloGeo. Interested? Contact: r.lovelace@leeds.ac.uk