Pedestrian crossing compliance characteristics at-grade signalized crosswalks: case study in a downtown-university campus environment (07-0114)
Darcin Akin Department of City and Regional Planning, Gebze Institute of Technology Istanbul Cad. No:101, Gebze, 41400 Kocaeli, Turkey Tel:+90-262-605-1642, fax: +90-262-653-8495 e-mail:
[email protected], alternate e-mail:
[email protected] Virginia P. Sisiopiku Department of Civil, Construction and Environmental Engineering, The University of Alabama at Birmingham, 140 Hoehn Engineering Building, 1075 13th Street South, Birmingham AL, 35294-4440 USA Tel: +1-205-934-9912, fax: +1-205-934-9855 e-mail:
[email protected]
Date of Submission: 11/14/2006 11:16 PM Number of Words in the text: 5250 words Number of Tables & Figures: 4 Tables and 5 Figures (2250words) Total Paper Length: 5250 + 2250 = 7500 (max.7500 words) Submitted for presentation & publication at 86th Annual Meeting of Transportation Research Board, 21-25 January 2007, Washington, D.C.
Reviewed on 11/14/2006
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
2
ABSTRACT This paper presents the findings of an investigation of pedestrian crossing compliance characteristics at signalized crossings in a downtown-university campus environment. The study was designed to obtain information on pedestrian crossing compliance with "pedestrian green” (signal compliance) and with “crosswalk location” (spatial compliance). Pedestrian movements were observed along a 1-km section of a downtown boulevard adjacent to the campus of Michigan State University, East Lansing, MI, USA. The facility included several types of crossings such as midblock, unsignalized and signalized crossings. The results from the analysis indicated that a strong correlation existed between the presence of a positive type of traffic control and pedestrian crossing compliance. Pedestrians recognized and used the signalized intersection crosswalks properly in regard to their locations. However, they did not seem to be complying with the WALK signal in the same degree that they complied with the location. The collected data were used to model pedestrian crossing compliance characteristics based on various definitions of the compliance: spatial, temporal, and overall. Key words: Signalized crosswalks, pedestrian crossing compliance, temporal (signal) compliance, spatial (crosswalk) compliance, overall compliance.
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Akin and Sisiopiku
3
1. INTRODUCTION The purpose of establishing crosswalks at-grade crossings is to channel pedestrian traffic to designated crossing locations and separate pedestrian movements from vehicular traffic by using proper channelization and signal designs. Separation of pedestrians from vehicular traffic by the introduction of high quality service crosswalks and the provision of an acceptable level of service in crosswalks are expected to increase pedestrian safety as well as crossing compliance. More importantly, proper usage of crosswalks by pedestrians and motorists increases pedestrian safety (1, 2) and reduces potentials for pedestrian crash occurrences (3). Researchers (4, 5, 6, 7, 8, 9) have been interested in the behavior of pedestrians crossing at various types of environments in order to reduce the conflicts with motor vehicles by increasing spatial as well as temporal compliance of pedestrians (10, 11). 1.1. Pedestrian Crossing Compliance Many pedestrians crossing at signalized intersections tend not to comply with the signal indication of pedestrian green especially at times when low vehicular traffic flows are observed an/or vehicles platoon along the street. If one is interested in the assessment of the attractiveness of signalized crosswalks relative to unsignalized intersection or midblock crosswalks, spatial crossing compliance (compliance with location) should be the criterion to measure the effectiveness of such crossings because that creates a base to compare the two crosswalk types (3, 6, 11). In this way, the effect of signalization on the crossing compliance is excluded and the efficiency of the crosswalks is determined in a more comparable way. By this type of comparison, the design problems with signal timing and signal phasing might not conceal the real attractiveness of signalized crosswalks as compared to others. In addition, to improve the temporal compliance (compliance with signal timing), required signal timing and phasing adjustments can be done very easily. Signal (temporal) compliance is related to signal design (signal timing and phasing plans) and vehicular flow characteristics (gap, arrival type, speed, etc.), and presumably to pedestrian attitudes. On the other hand, spatial crossing compliance is a very important measure of effectiveness (MOE), which indicates whether or not the location of a crosswalk is properly selected relative to pedestrian paths, trip origins and destinations, and whether or not the crosswalk is well marked, safe and convenient for pedestrian crossing. A study of walking activities and the characteristics of pedestrians in Beijing, China defined illegal crossing as “crossing a road during the red signal” (12). Basically the temporal compliance is used as an MOE. However, if one really would like to examine how efficient the signalized intersection crosswalks are, overall pedestrian compliance (compliance with both location and signal) should be the criterion to assess the efficiency of signalized crosswalks since compliance with both location and signal timing are two main characteristics to describe signalized crossings (3, 6, 11). Compliance rates reported in the literature imply that pedestrian compliance varied from location to location, city to city, and neighborhood to neighborhood. For example, a study that observed the characteristics of Chinese pedestrians reported that 70 and 57 percent compliance rates were observed at two study locations in Beijing (12). Another study done in Ohio reported that pedestrian crossing compliance rates at signalized and unsignalized midblock crosswalks were 85.4, and 86.4 percent with pedestrian sign, respectively (84.2 percent without pedestrian sign) (10). In addition, crosswalk features, pedestrian characteristics, user preferences and habits are expected to affect compliance rates. Therefore, the comparison of crossing compliance rates from one place to another may not be very meaningful without knowing local pedestrian characteristics and behaviors, traffic conditions,
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Akin and Sisiopiku
4
and the location of crosswalks relative to major pedestrian paths leading to crosswalks. When possible, various crossing options should be evaluated in the same environment. This approach is preferred for more meaningful comparisons since the effects of pedestrian characteristics, traffic exposure, and geometric features of pedestrian paths and roadways can be isolated from the effect of crosswalk features. Moreover, pedestrian behavior and attitudes are very important determinants in crossing compliance. Thus, it is important to know of what type of facilities pedestrians favor and what the attitudes of pedestrians towards such facilities are at different crosswalk and environmental settings (3, 6, 11). 1.2. Pedestrian Behavior and Attitude A study by Forsythe and Berger (13) presented the results of interviews with pedestrians crossing unsafely during DON’T WALK signal indication or pedestrian red interval. It was reported that the reason for unsafe crossing was mainly time-related. A need to hurry or a desire to keep moving was the main reason behind the lack of compliance with pedestrian signals. In the study of pedestrian compliance in Beijing (12), the impatience of pedestrians and the lack of effective enforcement were reported as the main reasons for the low compliance with respect to pedestrian signals. Pedestrians were also less enthusiastic about using overpasses and underpasses than signalized crosswalks. However, when the safety aspects and other related attributes were appraised, pedestrians did not favor any type of crossing in particular. It was concluded that pedestrians in Beijing would accept any type of crossing facility as long as they were appropriate and sufficient (12). However, one should be cautious when applying conclusions from the behavioral studies conducted abroad to U.S. conditions since the pedestrian behaviors as well as attitudes might be very different due to differences in socioeconomic characteristics and physical conditions of the study populations. For example, it is reported that Asian pedestrians walk at slower pace compared to their Western counterparts (14, 15). 1.3. Research Motive Traffic engineers and planners have always been desirous of gaining insight into the compliance of pedestrian crossings at locations with various crosswalk designs and features. Information obtained from the analysis of pedestrians’ preferences, behaviors and attitudes can be used as a guide for future projects through the identification of treatments that are favored by pedestrians (5, 6, 7, 8, 9). The objective of this study is to evaluate the effectiveness of the design that was implemented through the analysis of spatial and temporal pedestrian compliance rates at signalized crossings in a downtown-university environment. The data for this analysis were collected along Grand River Avenue in downtown East Lansing, Michigan, U.S., a city of 46,525 residents in 2000 (16, 17). East Lansing is home of Michigan State University (MSU), with over 40,000 students, faculty and staff. Most of the university students live within the city, with 31.3% of all or 14.573 persons lived in group quarters (16). Sorority and fraternity houses located downtown host a great deal of students and the high student population, whose mobility mostly depends upon walking and bicycling, puts significant pressure on the pedestrian network of the campus as well as the city’s. In recognition of the importance of pedestrian and bicycle modes in urban transportation, the city of East Lansing with the collaboration of the state of Michigan initiated a project of redesigning downtown East Lansing. The renovations along Grand River Avenue from Abbott to Bogue Streets aimed at improving traffic flow through downtown, increasing safety of pedestrian movements along and across the boulevard, and enhancing aesthetics of the corridor.
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Akin and Sisiopiku
5
This paper presents the results from the analysis of pedestrian crossing compliance characteristics at signalized intersection crosswalks along the section of Grand River Avenue between Abbott and Bogue Streets. The collected data were used to model pedestrian crossing compliance in downtown East Lansing. The data and established relationships could be used as a basis for the development of pedestrian compliance models for the city of East Lansing and/or other large university campus cities in the U.S. with similar pedestrian characteristics.
2. SITE DESCRIPTION Grand River Avenue is an east-west corridor at the northern boundary of Michigan State University (MSU) campus. The selected section was a 1 km (0.63 mi) long divided boulevard between Abbott and Bogue Streets (32,000 AADT in 2000). This section was renovated from 1994 through 1995 to include well-marked intersection and midblock crosswalks, pedestrian signs and shelters at the median, brick paving at crosswalk medians and curbs, and physical barriers, 2-foot high concrete walls placed at some locations to separate vehicle and pedestrian traffic. With new crosswalk layouts, traffic conflicts between pedestrians and leftturning vehicles from Grand River Avenue onto cross streets are aimed to be reduced by locating the crosswalks only at the east side of the intersections (Figure 1a, b). To facilitate data collection the study site was divided into two sections, almost equal in length. The site had four signalized intersections, two of which are cross-intersections (Abbott and Collingwood), and the other two are T-intersections without the south leg (M.A.C. and Division, see Figure 1a, b). The Collingwood Street intersection was the only one that had crosswalks on both sides of the intersection. Crosswalk lengths varied from 27.8 to 37.9 m (91.2 to 124.3 ft). Median width also varied throughout the study site. To make the drawing of the study site’s scheme simpler, the median was drawn in a constant width. With respect to signal phasing design, the westbound traffic flow received the red signal a few seconds earlier than the eastbound traffic at the Abbott and Division intersections. The purpose of this arrangement was to allow the maneuver of eastbound left-turning vehicles onto the cross streets. During this signal interval, traffic signal remained red for pedestrians crossing the Grand River Avenue. However, pedestrians traveling southbound kept crossing the westbound portion of the roadway despite the steady DON’T WALK signal indication since westbound vehicles were stopped and pedestrians on the crosswalk observed no threat from the traffic.
3. METHODOLOGY To study pedestrian crossing compliance characteristics, several definitions of pedestrian compliance rates are introduced. In general, crossing compliance rate (percent) is defined as the number of pedestrians crossing at a crosswalk location divided by the number of pedestrians in the “crosswalk influence area” in a period of time, usually an hour. 3.1. Crosswalk Influence Area (CIA) The calculation of pedestrian crossing compliance rates (PCCRs) is based upon the assumption that each crosswalk has an influence area that attracts pedestrians crossing the street. In order to determine the crosswalk influence area (CIA) for each crosswalk, the distance between a pair of two consecutive crosswalks is divided into two by an imaginary …..
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
6
Akin and Sisiopiku
E
Division St Signalized x-walk
N
Student Book Store, SBS
113.7m (373’)
DIVISION ST
Charles St Unsignalized x-walk
Jacobson’s store
82 m (269’)
CHARLES ST
61.9 m (203’)
Jacobson’s Midblock x-walk
M.A.C. Ave Signalized x-walk
MSU Student
: signalized x-walks : stop sign
: signal : shelter
Union MSU Student Union Midblock x-walk 38.4 m (126’)
: barrier walls
132.6 m (435’)
M.A.C. AVE.
Abbott Signalized x-walk
ABBOTT ST
FIGURE 1a Scheme of the Study Site: Section 1 (not to scale).
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
7
Akin and Sisiopiku
E N
Orchard St Unsignalized x-walk 101.8 m (334’)
ORCHARD ST
39.3 m (129’)
Collingwood St-East Signalized x-walk
COLLINGWOOD ST
Good Times Pizza
70.7 m (232’)
Collingwood St-West Signalized x-walk
137.8 m (452’)
Bailey St Midblock x-walk
BAILEY ST
MSU Federal Credit Union Branch
stop sign
MSU Berkeley Hall
: shelter DIVISION ST
signal
78.3 m (257’)
Berkeley Hall Midblock x-walk
Division St Signalized x-walk
FIGURE 1b Scheme of the Study Site: Section 2 (not to scale).
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
8
Akin and Sisiopiku
line. Thus, each crosswalk is located between the two consecutive dividing lines serving as the boundaries of the so-called “crosswalk influence area.” Distances between three consecutive crosswalks (i-1, i, and i+1) are Li-1 and Li+1. Then, length of the CIA (LiCIA) for crosswalk i is the sum of Li-1/2 and Li+1/2 (Figure 2). 3.2. Full and Partial Jaywalkers Pedestrians complying with crossing location are those that cross within approximately 3.0 m (10 ft) from both sides of the crosswalk. This area is defined as crosswalk area (CA) in Figure 2. Jaywalkers are pedestrians who do not comply with crosswalk location; i.e., do not walk within CA while crossing the street. On the other hand, partial jaywalkers are those whose crossing paths are partially in CA (3, 6, 11). 3.3. Pedestrian Crossing Compliance Rates (PCCRs) Pedestrian crossing compliance rate (PCCR) is defined in three forms; namely, spatial, temporal and overall crossing compliance rates. 3.3.1. Spatial Crossing Compliance Rate (SCCR) Spatial compliance refers to the crossing compliance of pedestrians with respect to crossing location or “CA” and SCCR discloses information on the percent of pedestrians crossing the street at designated locations (i.e., crosswalks) in regard to CAs: SCCRi =
Pi L Pi CIA
(1)
where SCCRi = spatial crossing compliance rate at crosswalk i (percent); PiL = number of pedestrians crossing within crosswalk area, CA, (ped/hr); and PiCIA = total number of pedestrians in the crosswalk influence area (CIA) of crosswalk i (ped/hr).
Crosswalk i-1
Crosswalk i
Crosswalk area (CAi)
Crosswalk i+1
W
Li-1 (Li-1 / 2)
Li+1
(Li+1/2)
Crosswalk Influence Area (CIAi)
FIGURE 2 Definition of Crosswalk Influence Area (CIA) and Crosswalk Area (CA) for the Calculation of PCCR.
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
9
Akin and Sisiopiku
3.3.2. Temporal Crossing Compliance Rate (TCCR) At signalized crosswalks, pedestrian crossing compliance can also be linked to the compliance with pedestrian WALK signal indication. As shown in Equation 2, temporal or signal compliance, TCCRi, is defined as the ratio of the number of the pedestrians (per hour) in CA, who comply with the pedestrian WALK signal indication, PiT, over the total number of the pedestrians within CA, PiCA TCCRi =
Pi T Pi CA
(2)
The reason that the denominator in Equation 2 is different from the one in Equation 1 is because it is not meaningful to reason temporal crossing compliance when pedestrians do not comply with the crosswalk area at all. In other words, spatial compliance is the required condition for temporal compliance. Therefore, in determining TCCR only the pedestrians crossing in CA are considered. 3.3.3. Overall Crossing Compliance Rate (OCCR) To combine the SCCR and TCCR for signalized crosswalks, the concept of overall crossing compliance rate (OCCR) is introduced. As shown in Equation 3, OCCRi is defined as the ratio of the number of pedestrians (per hour) who comply with both CA and pedestrian WALK signal, PiLT, over the total number of the pedestrians (per hour) within CIA, PiCIA OCCRi =
Pi LT Pi CIA
(3)
4. DATA COLLECTION AND REDUCTION Pedestrian crossing patterns were studied through the analysis of field data collected by direct observation of pedestrian activities using video cameras set up at locations on the sidewalks along the study site. Pedestrian movements were observed and recorded for 30 min at each camera location during the noon- (10:30 am-1:00 pm) and afternoon-peak (2:30-6:00 pm) periods. Signalized crosswalks in the study area were observed over a total of thirty-six 30min periods, while crossing data at unsignalized crosswalks were recorded over fourteen 30min periods. Data collection was performed under various weather conditions (sunny, cloudy, snow sprinklers, cold, and warm). Eight video cameras were used simultaneously to record pedestrian movements in the site. Video cameras were consecutively located on both sides of the Grand River Avenue on the sidewalks in order to cover all possible pedestrian movements within the entire study area. The recording areas of consecutive cameras overlapped a little bit to ensure that all pedestrian walking activities were fully captured. Special efforts were made to avoid double counting of pedestrians that started to cross within a view of one camera and completed their crossing within the view of subsequent camera(s). Though this type of data reduction process was a tedious and labor-intensive task, it was very beneficial as it allowed detecting every pedestrian movement in the entire study site. Due to this data reduction approach, not only the information on pedestrian volume and crossing locations became available but also additional data about conditions during crossing (such as pedestrian signal indication, the presence of other pedestrians and motor vehicles,
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
10
Akin and Sisiopiku
etc.) were obtained through a careful processing of the videotapes in the laboratory environment. Following data were recorded for the analysis of pedestrian movements: a) number of pedestrians who started crossing the street during pedestrian WALK signal (regular users); b) partial jaywalkers; c) jaywalkers; d) number of pedestrians in the crosswalk area (CA) during “DON’T WALK” signal (sneakers); e) number of pedestrians who crossed the first portion of a crosswalk during “DON’T WALK” signal and then continued crossing during the “WALK” signal (partial sneakers); f) number of pedestrians who crossed from a curb to the median during the “flashing DON’T WALK” signal (late starters); and g) the total number of pedestrians within the CA and CIA. Data summary forms were developed to report summarized data and calculate PCCRs. An example of such data reduction forms for signalized crosswalks is given in reference (9). 5. DATA ANALYSIS AND RESULTS Spatial (SCCR), temporal (TCCR) and overall crossing compliance rates (OCCR) for the five signalized crosswalks (SICWs) in the study area were calculated and tabulated for each data collection session. Results are presented in the following sections. 5.1. SCCRs at the SICWs The SCCRs in the study site varied from 68.4% to 98.2% with an average value of 83.1%. Descriptive statistics of the SCCRs are summarized in Table 1. Mean values for Sections 1 and 2 are 82.78% and 83.40%, respectively. Overall, a little more than four-fifths of all pedestrians crossing in the CIAs complied with crosswalk location. The null hypothesis that the pedestrians do not comply differently with the location of the signalized intersection crosswalks throughout the study site is not rejected at the 5% level of significance (t= -1.666 and p=0.117>0.05). This means that the signalized crosswalks are equally attractive throughout the study site since they appeared to be well marked and equally recognized as proper crossing locations by pedestrians in the study site.
TABLE 1 Descriptive Statistics of Spatial Crossing Compliance Rates (SCCRs) at Signalized Intersection Crosswalks (SICWs) Study Crosswalks Section (1) (2) Abbott 1 M.A.C. Division 2
Total
Max (6) 90.3 89.1 93.6
SCCR (%) 85th Mean percentile (7) (8) 85.64 89.74 79.93 87.60 84.41 92.87
Std. Dev. (9) 3.914 6.497 6.213
70.0
98.2
88.38
98.04
10.380
68.4
91.2
77.07
91.90
10.228
Ave. PCIA N† (3) (4) 150 8 153 8 351 8
Min (5) 78.4 69.9 78.2
Collingwood-West
53
6
Collingwood-East
66
6
773
36
Overall mean =
Mean (85th percentile) (10) 82.78 (88.88) 83.40 (93.50) 83.13
Conclusion: t-value = -1.666 and p = 0.117 > 0.05. The difference between the two means is NOT statistically significant at the 0.05 level; thus, Ho is NOT rejected. †: N refers to the number of 30-min. observation periods.
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
11
Akin and Sisiopiku
5.1.1. Modeling of SCCRs at the SICWs The SCCRs are modeled using regression analysis. In this analysis, pedestrian volume (PV), pedestrian volume over the length of CIA (PV/LCIA), 2-way vehicular volume (VVtwo-way) and the length of crosswalk influence area (LCIA) are considered as potential independent variables. Two models were developed via the regression analysis using a forward stepwise procedure. Equation 4 shows the first model and the model fit to the observed data is shown in Figure 3. ln SCCR = 4.379 + 5.836 * 10 2 * ln(
PV ) LCIA
Model I
( 4)
The goodness-of-fit (R2) of model I in Equation 4 is 0.176. The model is statistically significant at the 0.05 level (F=7.068 and p=0.012
