Assessing bicycle safety in multiple networks with different data models Martin Loidl Department of Geoinformatics, Z_GIS University of Salzburg [email protected] GI-Forum 2014, Salzburg
Jun 22, 2015
Assessing bicycle safety in multiple networks with different data models
Martin LoidlDepartment of Geoinformatics, Z_GIS
University of Salzburg
GI-Forum 2014, Salzburg
2
Context
Why bicycle safety?!
www.zeitpunkt.ch
“Road safety: Second good year in a row puts Europe firmly on track towards target” (EC - IP/14/341; 31/03/2014)Another worrying feature of the statistics is the situation of vulnerable road users: The number of pedestrians killed is decreasing to a lesser extent than expected and the number of cyclists killed has recently even been increasing. This is partly due to the fact that more and more people cycle; the challenge for Member States is to encourage people to use their bicycles rather than their cars more often, but to make sure that the shift from car to bicycle is a safe one.
3
How can we contribute to safety improvements for bicyclists?
Numbe
r of b
icycli
stsNum
ber of accidents
4
Assessment
Status-quo analysis Weak-point analysis Corridors etc.
Planning & Infrastructure Priority of measures Budget allocation etc.
Information Routing etc.
www.stadt-salzburg.at
www.stadt-salzburg.at
www.radlkarte.eu
Assessm
en
t of N
etw
ork
5
How to assess road networks
6
Assessment Method I
Fuzzy results From grades to verbal description
Assessment standard hardly to establish Person, time, location
Requires much effort E.g. re-assessment after every physical
modification of road space
Expert knowledge as input for assessment procedure Sub-optimal for global assessment approach
7
Assessment Method II
Conceptual fallacy Absolute (!) number of accidents no indication
for unsafe roads Function of N External effects
Accident analysis for detection of risk factors or hot spot analysis
http://gicycle.wordpress.com
8
Assessment Method III
Not equally distributed over space Coverage in remote vs. central areas
Data Representative only for specific sub-groups
Semantic challenge Ontologies for feedback assessment
For punctual validation/calibration For group-specific analysis, qualitative assessment Not yet applicable for global assessment = research topic!
9
10
Indicator-basedAssessment Model
11
Loidl & Zagel (2010)
12
Problem Statement
Bicycle routing service with „safety“ as routing criteria Different data sets from different sources Different data models
Geometry and attributes
Different environments to adapt the model for Urban vs. rural
type = roadbicycle_infra = cycleway
type = road
type = cycleway
13
Study Area & Data
Authoritative data for city of Salzburg 11,458 edges 1,119.7 km net length
OpenstreetMap extract for adjacent municipalities (1A, 4G) 9,601 edges 941.3 km net length
Data sets with different data model and attribute structure
14
Data models
Authoritative graph Center line Complex attributive formulation of bicycle infrastructure Complete and homogeneous
DB-Name Attribute Value Value descriptionSIC_STRAS4 Road category 400 Municipal roadSIC_STRAS5 Max.speed, km/h 50 SIC_STRA9 Buslane 1 Yes, FTSIC_STR13 Oneway 1 Yes, FTSIC_RAD_RE Bicycle infrastructure (right side, FT) 10 Yes, undefinedSIC_RAD_R1 Bicycle infrastructure (left side) 5 Cycle-/footway mixedSIC_RAD_RI Direction bicycle infrastructure (FT) 0 NoSIC_RAD_R5 Direction bicycle infrastructure (TF) 4 Both directions, independet from onewaySIC_STR24 Motorized traffic load, V/24h (TF) 0 SIC_STR25 Motorized traffic load, V/24h (FT) 17698
15
Data models
OSM graph Edge for every physically separated
lane Very simple tag (key = value)
structure Up-to-date Gaps, inconsistencies, errors,
heterogeneous attribute structureKey Valuehighway pathbicycle designatedfoot designatedsurface paved
Loidl et al. (2014), Aufbereitung von Open Street Map Daten für GIS-Modellierungen und Analysen. AGIT
16
Workflow
Define transition points between networks
Line matching and integration of transition
edges
Identify indicators for urban and rural
environment
Find corresponding attributes in digital data
Run indicator-based assessment model for
both environments
Z-transformation for index values
Rivers » bridgesHighway » ring underpasses
17
Geometry & Topology Transfer points
Bridges, under-/overpasses Transition edges Line matching
18
Assessment Model
19
Result
20
Application
www.radlkarte.eu
21
Conclusion
Indicator-based assessment model Transparent Comparable Adaptable and scaleable Reproducable
Applicable in several contexts Proof of concept: safe bicycle routing Planning and simulation tool
GIS „intelligence“ helps for modelling in heterogeneous environment (data, geography, …)
@gicycle_
gicycle.wordpress.com