Transcript Slide 1

Page
Request
Plot Page
Generation
Incident Record
Table
Query
Result
Incident on I-205N at the off ramp
for Hwy 212/214. A log truck rearended a truck carrying nursery
stock; two cars also involved.
Incident lasted just over 4 hours.
Aggregation
Query
Manual
Transfer
Incident data architecture in PORTAL
Display of incident location and
duration in PORTAL
Right
Shoulder
44%
Construction
2%
All Lanes 5%
Off Road
Left 0%
Left
Shoulder 4%
2 Lanes
3%
Crash
15%
3+ Lanes
1%
4.5
1 Lane
31%
Off Road
Right 1%
0 Lanes
65%
Stall 57%
Debris
12%
Incidents by location for
1999–2005 (N = 17,855)
Total
Stall
Crash
Debris
Construction
4
3.5
3
2.5
2
1.5
1
0.5
0
Incidents by Type and Number of Lanes
Affected for 1999–2005 (N=128,886)
Incidents by hour and type (1999-2005)
9:00
12
Average Number of
Incidents Per Day
28
Crashes on Dry Days
Crashes on Wet Days
Wet Days
14
8:00
24
20
10
16
8
12
6
8
4
2
4
0
0
7:00
Accident
6:00
Stall
Debris
5:00
Tow
4:00
Construction
Congestion
3:00
Other Closure
2:00
Other Incident
1:00
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
0:00
Dec
1999
700
MVMT
1:00
5
400
4
300
3
200
Incidents/MVMT
500
2
100
1
0
0
Jan Feb Mar Apr May Jun
Jul
Aug Sep Oct Nov Dec
Incidents and MVMT by month (99-05)
Duration (H:MM)
6
Incidents/MVMT
2001
2002
2003
2004
2005
1:10
7
Incidents
600
2000
Average incident duration by type
Crashes and precipitation (2005)
Number of Incidents and MVMT
The authors thank the Jack Marchant and Dennis Mitchell at the
Oregon Department of Transportation for providing the data and the
National Science Foundation for funding this research.
Incidents By Hour and Type (1999-2005)
Tow 6%
Month
Acknowledgements
Other
Incident
7%
10:00PM
Gore Area
(left) 0%
Other
Closure
1%
Congestion
0%
8:00PM
Right Lanes
16%
Center Lanes
4%
6:00PM
Gore Area
(right) 1%
4:00PM
Left Lanes
11%
None 10%
Average Incidents Per Day
Unknown 4%
Incident on I-205 Southbound.
One lane closed.
2:00PM
ODOT TMOC
12:00PM
Client Machine
Incident Entries
Table
10:00AM
Web Browser
Operators
Create
Incident
Entries
8:00AM
Incident Plot
Display
6:00AM
File
Request
4:00AM
Web
Page
Plot
SQL
Query
2:00AM
Generated
Web Page
Database Server
12:00AM
Web
Server
Conclusions
This paper has described an automated data cleaning and
aggregation process for incident data from the ODOT ATMS
database. By automating the data cleaning, it is now possible to
quickly produce incident records from ATMS database incident
entries, which are stored in the PORTAL data archive. Having such
incident records in PORTAL allows incident statistics to be
generated with minimal effort. In addition, any PORTAL user can
easily access the incidents for a particular freeway during a time
interval so that traffic patterns can be compared with associated
incident data. In particular, this research has demonstrated how
incidents can be displayed on speed contour plots to allow users to
distinguish between recurrent and non-recurrent congestion. Finally,
since PORTAL archives weather data, weather conditions can be
easily compared to incident records. This flexible, automatic incident
cleaning and archive technology provides a new and useful tool to
traffic practitioners and researchers.
 Incident data from ODOT ATMS
 139,484 incidents from 1999 – 2005
 Data entered by TMOC operators
Central Server
Duration(H:MM)
The Portland Oregon Transportation Archive Listing (PORTAL)
archives high resolution traffic data including speed, volume, and
occupancy collected from 500 freeway loop detectors in the
Portland metropolitan area. PORTAL currently provides measures
related to total congestion that occurs on the freeway network,
but cannot presently distinguish between recurrent and nonrecurrent congestion effects. In response to the need to make
such a distinction, the objective of this paper is to describe the
incorporation of freeway incident data received from the Oregon
Department of Transportation (ODOT) into PORTAL. ODOT’s
freeway incident database includes information about vehicle
crashes and stalls, debris on the road, construction and other
random events. The paper describes the incident archive
architecture and demonstrates how users can view incident data
associated with a particular time frame and location. For example,
a user analyzing data from a particular day will be able to
immediately access associated incident data so the user can
determine if the traffic pattern is related to an incident. In addition,
freeway incident performance metrics will be described, including
tracking of incident trends over time and location and numbers of
incidents by incident type. Finally, the paper will describe how
comparing incidents with weather data archived in the PORTAL
database can be used to determine how weather has influenced
incidents. The above-described incident performance metrics are
useful to traffic researchers and practitioners and may contribute
to incident-reduction measures in the Portland area.
Incorporating Incident Data into a Freeway Data Archive for Improved
Performance Measurement
Jessica Potter, Suman Tasnim, Dr. Kristin Tufte and
Dr. Robert Bertini, Portland State University
Number of Wet Days
Abstract
30,000
All located incidents
Exclude hazmat, fatal
Exclude construction, congestion
Number of located incidents
Incident Tree 2005
One Lane 71.8%
Crash 41.1%
25,000
Multi-Lane 28.2%
0:50
In-Lane 37.3%
20,000
One Lane 98.8 %
0:40
Stall 58.9%
15,000
N = 14253 *
Multi-Lane 1.2%
0:30
Crash 10.3%
10,000
www.its.pdx.edu
0:20
Shoulder 62.7%
5,000
0:10
Stall 89.7%
0:00
0
1999
2000
2001
2002
2003
2004
2005
* Incidents with complete lane location data
Average incident duration (July 99 – Dec 05)