(GTFS)-based GIS Tool for Creating Practical Applications

Report
General Transit Feed Specification (GTFS)-based
GIS Tool for Creating Practical Applications
East-West Gateway
Council of Governments
Sang Gu Lee
GIS in Transit Conference
October 16, 2013 │ Washington, DC
Introduction
• Propose use of Google’s General Transit Feed Specification
(GTFS) for a transit stop aggregation model (SAM)
• The idea of using GTFS has been drawing attention in the
public transit planning area these days
• One area in which GTFS can be very useful is in developing
and updating transit networks used in service planning
• We explore how to use this innovative data source in
various areas by proposing a SAM
General Transit Feed Specification (GTFS)
• Open data format for transit schedules
• First released with TriMet (Portland, OR) in 2005
• Incorporating transit information in the Google Maps
application
• A de facto standard for data describing transit stops,
schedules, and route geometry, …
• Currently, many transit agencies in the US have made their
GTFS data publicly available, which helps developers and
transit agencies efficiently share and retrieve GTFS data
(e.g., http://www.gtfs-data-exchange.com)
General Transit Feed Specification (GTFS)
Previous Research
• Stop-level boarding and alighting counts aggregated to
the segment level for generating a transit route origindestination matrix (Furth and Navick, 1992)
• The need for relevant stop aggregation was discussed
to match the scheduled time between bus stops from
the transaction data collected (Barry et al. 2002)
• Each pair of stops on the opposite sides of a road at
the same general location might be combined for
predicting transit-related activities (Chu, 2004)
Conceptual Approach
Transit users’ activity may not be originated
from or destined to an individual stop per se
The activity is associated with a specific
location in the vicinity of the stop
This location may be “covered”
by several adjacent transit stops
Three parameters:
Distance, Text, and Catchment area
Developing Stop Aggregation Model (SAM)
Stop Aggregation Model (SAM)
Parameter
Distance-based
Text-based
Catchment-based
Similarity
Spatial
Textual
Land use or activity
Stage
Lower-level
Lower-level
Upper-level
Scope in algorithm
Regional
Regional
Route-level
Data in GTFS
Stops.txt
Stops.txt
Stop_times.txt
Implementation
ArcGIS
Geographical
proximity
Microsoft SQL server
Microsoft SQL server
Service characteristics
in the catchment area
Advantage
Drawbacks
- Distance threshold
dependent
(e.g., individual or
overlapping)
Textual comparison
- Unique and various
types of text
- Geographical
location issue (e.g.,
curves in transit line)
- Not easy to extend
to a regional scope
Stop Aggregation Model: Development and Application (Lee et al. 2012)
Distance-based SAM
Stop Aggregation Model: Development and Application (Lee et al. 2012)
8
Distance-based SAM: Sensitivity of Distance
200 150 100 70 50
50 70 100 150 200
SB Trip
200 150 100 70 50
50 70 100 150 200
SB Trip
NB Trip
One-way
CBD
Downtown
Minneapolis
University of
Minnesota
NB Trip
SB Trip
NB Trip
Same direction
Study route
Opposite direction
Stop Aggregation Model: Development and Application (Lee et al. 2012)
Case Study
• Minneapolis /St. Paul (MN) and Sacramento (CA)
Number of Groups
Minneapolis/St. Paul Area (Total: 14,601)
Number of
individual stops
in the group
DBSAM
(50 m)
TBSAM
Integrated
DBSAM and
TBSAM
Sacramento Area (Total: 4,366)
Integrated
DBSAM and
TBSAM
DBSAM
(100 m)
1
2,383
4,422
2,135
1,107
665
2
5,225
4,978
5,301
1,248
1,272
3
265
43
262
116
142
4
192
11
213
73
106
5
23
5
27
13
27
6
8
8
4
17
7
2
1
1
1
5
8
1
1
2
1
2
1
1
9
10
Total
1
1
2
1
1
8,101
9,462
7,951
2,565
2,238
Stop Aggregation Model: Development and Application (Lee et al. 2012)
Use of a Stop Aggregation Model
AFC
Transit demand
GTFS
Parcels
Stop
Aggregation
Model
(SAM)
Land-Use pattern
Aggregate-level
O-D estimation
Measuring
accessibility
Identification of
boarding and
alighting locations
Observing land use
and activity location
Network Development
Intersection-level
Transit Network
Intermodal Network
(e.g., Park-n-Ride)
Mutually Exclusive
Service Areas
Are Transit Trips Symmetrical in Time and Space? Evidence from the Twin Cities (Lee and Hickman, in press)
Integrating Transit Demand and Land Use
General Transit Feed
Specification (GTFS)
Automated Fare
Collection (AFC) Data
Stop Aggregation
Model
Street
Network
Determination of
Transit Service Area
Parcel-level
Land Use
Measurement of
Land Use Types
Time-varying
Transit Demand
Linkage
Development of a Temporal and Spatial Linkage between Transit Demand Land Use Patterns (Lee et al. 2013)
Developing Intermodal Network
Street
Junction
Vehicle
SAM
Access point
to P&R
Auto
Vehicle
P&R Centroid
Transit
Walk
Bus Stop
LRT Station
Sunrise Park-and-Ride at Sacramento, CA
An Intermodal Shortest and Optimal Path Algorithm using a Transit Trip-based Shortest Path (Khani et al. 2012)
Intersection-level Origin-Destination Estimation
• Using AFC data
B2 T2
A4
B4
T4
A1
B1 T1
A3
B
Boarding stop
A
Alighting stop
Stops serving by Orange Route
Stops serving by Red Route
A2
Location of Transaction
T3
B3
Stop Group of SAM
Stop Aggregation Model: Development and Application (Lee et al. 2012)
Linkage with On-Board Survey Data
• Spatial references are typically asked of each respondent
about where they are coming from and going to
Boarding and alighting information
from the on-board survey data
Stop names in SAM
Stop IDs only
along Route 25
10513
…
…
SurveyID
4334
Silver Lake Rd & 36 Av NE
14157
Route
25
Silver Lake Rd & 37 Av NE
14155
ServiceType
Local
Silver Lake Rd & 39 Av NE
14154
TimePeriod
Peak
…
BoardIntersect
Silver lake & 39th NE
Hennepin Av E & 4 St SE
42008
BoardCity
Minneapolis
Hennepin Av E & 5 Av SE
14943
AlightIntersect
Hennepin & 6th St
Hennepin Av E & 6 St SE
14955
AlightCity
Minneapolis
Hennepin Av E & 8 St SE
14946
…
…
…
…
Stop Aggregation Model: Development and Application (Lee et al. 2012)
…
Record
Generating Mutually Exclusive Service Areas
• Combination of Thiessen Polygon and Buffer (CTPB)
• CTPB approach improves the capability of spatial data
integration in direct demand models
Case Study: Route 6
Stop Group by SAM
CTPB
Comparative Study of alternative methods for generating route-level
mutually exclusive service areas (Lee et al., in press)
Route-level Mutually
Exclusive Service Areas
Accessibility: The Nth Nearest Stop Group
1st
2nd
3rd
Parcel ID
Nth Nearest
Stop ID
Stop Group
053-0311722XXXXXX
1
2
3
4
5
6
7
8
44952
45006
7049
7059
45007
44951
52924
52923
1
1
2
2
3
3
4
4
Length
(in meter)
662
668
981
987
1026
1042
1082
1105
4th
Express Service Available
7 - 8 am
12 - 1 pm
5 - 6 pm
O
O
O
O
O
O
What if 4th stop is better choice with express service at a specific time?
Measuring Transit Accessibility
Meters
1,700 ~ 1,800
700 ~ 800
The 1st nearest stop group
The 2nd nearest stop group
0 ~ 100
Arbitrary points assigned as
facilities in Network Analyst
in GIS due to the
observance of isolated
street network
The 3rd nearest stop group
The 4th nearest stop group
East-West Gateway Travel Demand Model
Passenger Behavior Data
Quantity of Data
Passenger
Counts
Farecard Data
On-board
Surveys
Household Surveys
Behavioral Richness
Enhancing the Modeling Capabilities
Travel Behavior Analysis and Accessibility Measure
Travel Pattern Analysis (Lee and Hickman 2011)
Modified Empty Space Distance for Measuring Transit Accessibility (Lee et al. 2012)
Trip purpose inference using AFC data (Lee and Hickman, in press)
Generating Mutually Exclusive Service Areas (Lee et al., in press)
Symmetry of Boardings and Alightings (Lee and Hickman, in press)
Relational Database Modeling
Integration with GTFS (Nassir et al. 2011)
Integration of Land Use and Transportation
Temporal and Spatial Linkage between Transit Demand and Land Use Patterns (Lee et al. 2013)
Stop Aggregation Model: Development and Applications (Lee et al. 2012)
An Intermodal Shortest and Optimal Path Algorithm (Khani et al. 2012)
Demand Modeling
Time-varying Transit Patronage Models (Lee et al. 2013)
Transit O-D Estimation
AFC data: Stop-level (Nassir et al. 2011) and Aggregate-level (Lee et al. 2011)
APC data: Time-varying Alighting Probability Matrices (Lee and Hickman, under review)
Conclusions
• Provides the development and application of a stop
aggregation model for a transit network based on
Google’s General Transit Feed Specification (GTFS)
• Aggregate representation of transit stops
– Stop groups that serve common or similar land use patterns
and activities can be represented by a single node, which is
able to reduce the complexity of the transit network
– Easily applicable to model passenger transfers, and access
time and distance within these stop groups
• Utilization of Google’s GTFS
– Frequently updated by transit agencies, as it provides
detailed information on transit supply-side characteristics
Acknowledgements
•
•
•
•
Dr. Mark Hickman (University of Queensland, Australia)
Dr. Daoqin Tong (University of Arizona)
University of Arizona Transit Research Unit (UATRU)
East-West Gateway Council of Governments

similar documents