SUMMARY
Accidents involving pedestrians on crosswalks are a common cause of road fatalities. In-pavement flashing warning lights have been proposed as a means of increasing the conspicuity of a crosswalk when a pedestrian is using it. Evaluations in California and Washington have demonstrated the effectiveness of such in-pavement flashing warning lights on moderating drivers’ behavior when approaching a crosswalk. However, in-pavement flashing warning light systems are more expensive to install than striping, the conventional way of identifying crosswalks. This report describes work undertaken to determine the effect of an in-pavement flashing warning light system installed on a crosswalk on pedestrian safety, relative to striping.
The site for the work described was in Denville, New Jersey. This site was chosen for study for two reasons. First, the site forms part of a pedestrian route between a residential area and an extensive recreational area. Consequently, a significant proportion of the pedestrians are children. Second, vehicles can approach the site from several different directions, making it difficult for a pedestrian to know where to look, and the driver’s view of the site is often obstructed by stationary, back-up traffic.
At the time of the first evaluation, in 1999, there was only one marked crosswalk, delineated by eroded minimal striping. In 2000, another crosswalk was added, both crosswalks were striped and ADA ramps were provided. The second evaluation was then made. In September 2000, the in-pavement warning light system and the associated pedestrian detectors were installed. Evaluations were made shortly after the installation of this system, nine months later and one year later. Comparisons of the same crosswalk before and after striping and after installation of the in-pavement flashing warning lights enable the benefits of these actions to be determined. From the data collected it is concluded that:
· Clear striping of a crosswalk enhances the noticeability of the crosswalk to drivers who are not familiar with the location.
· Clear striping of a crosswalk reduces conflicts between pedestrians and vehicles, a conflict being defined as an occasion when a driver moves over the crosswalk while a pedestrian is on the crosswalk.
· Clear striping of a crosswalk does not reduce the mean speed at which vehicles approach the crosswalk.
· Clear striping of a crosswalk does not reduce the mean number of vehicles passing over the crosswalk while a pedestrian is waiting to cross.
· Adding an in-pavement flashing warning light system to a crosswalk that is already clearly striped enhances the noticeability of the crosswalk to drivers who are not familiar with the location.
· Adding an in-pavement flashing warning light system to a crosswalk that is already clearly striped reduces the mean speed at which vehicles approach the crosswalk.
· Adding an in-pavement flashing warning light system to a crosswalk that is already clearly striped reduces the mean number of vehicles that pass over the crosswalk while a pedestrian is waiting.
· The impact of adding an in-pavement flashing warning light system to a crosswalk that is already clearly striped on the mean speed at which vehicles approach the crosswalk tends to diminish over time.
Recommendations for the future use of in-pavement flashing warning light signals at crosswalks are given.
INTRODUCTION
Accidents involving pedestrians on crosswalks are a common cause of road fatalities(1). A number of companies are now promoting systems for increasing the conspicuity of a crosswalk when a pedestrian is on it. These systems consist of a series of high-intensity luminaires buried in the pavement on both sides of the crosswalk that direct light along the road towards oncoming traffic (2, 3). When activated, either by a pedestrian pressing a signal button or by some form of automatic pedestrian detection system, the lamps in the luminaire flash at a set rate for a fixed time. The bright flashing warning lights lining the crosswalk draw driver's attention to the crosswalk making it more likely that drivers will pay attention to what is happening there and act appropriately. Such systems can be integrated with other traffic signal lights if required. The 2000 Millennium edition of the Manual of Uniform Traffic Control Devices contains language that makes the use of in-pavement flashing warning lights at crosswalks acceptable and gives guidance for their application (4).
Previous evaluations of such in-road warning lights have been carried out in the states of California and Washington (5). The evaluations are based on observations of driver and pedestrian behavior and the opinions of drivers and pedestrians. However, these studies were all before-and-after studies and did not systematically compare the relative effectiveness of the in-pavement warning lights to the conventional approach of striping. As striping is less expensive to install than in-pavement flashing warning light systems, the use of the latter will only be justifiable if it produces a marked improvement in drivers' behavior and fewer conflicts between pedestrians and vehicles than does striping alone.
The objective of this project is to determine the effect of an in-pavement flashing, warning light system installed on a crosswalk on pedestrian safety, relative to striping.
RESEARCH APPROACH
The research approach is a field evaluation of the impact of successive improvements to an existing crosswalk. The evaluation design consists of an escalating series of before-and-after comparisons. The starting point was the crosswalk as it existed in 1999, with only one marked crosswalk delineated by eroded minimal striping. The first evaluation was made in these conditions. In 2000, another crosswalk was added, both crosswalks were striped and ADA ramps were provided. The second evaluation was then made. In September 2000, the in-pavement flashing warning light system and the associated pedestrian detectors were installed. An evaluation was made shortly after the installation of this system. Two further evaluations were made nine months and one year after the installation. Comparison of the same crosswalk before and after the striping and after the installation of the in-pavement flashing warning lights enables the impact of these modifications to be examined.
The evaluation was based on measurements of the behavior of drivers and pedestrians using the crosswalk, opinions of the pedestrians using the crosswalk and the conspicuity of the crosswalk to unwarned drivers. In addition, details of the reliability of the in-road, flashing warning light system over a year of operation were collected.
Evaluation Site
The Department of Transportation of the State of New Jersey identified a site for the evaluation in Denville, New Jersey. Figure 1 shows a plan of the site. The site is adjacent to a major traffic-signal-controlled intersection between US Route 46, a four-lane, divided highway, and Franklin Road, a two-lane road. There are crosswalks across US Route 46 and across Franklin Road but these are linked to the traffic signals via pedestrian operated push-buttons. The actual site of the evaluation is the T-junction between Franklin Road and Savage Road. At this junction, Franklin Road, which forms the stem of the T, has a large median, there being three lanes for travel in a southwesterly direction and one lane for travel in a northeasterly direction. As for Savage Road, which forms the cross of the T, there are two lanes for travel in the northwesterly direction for the part to the southeast of Franklin Road, and one lane each for travel in the southeasterly and northwesterly directions on the part of the road to the northwest of Franklin Road. To the North of Savage Road is Gardner Field, an extensive recreational area containing baseball diamonds, tennis courts, soccer and football fields, and a children's playground. Access to the parking lot of Gardner Field is off Savage Road, just to the west of crosswalk 2. Egress from the parking lot is actually into the junction adjacent to crosswalk 1. To the South of Savage Road, the area between Savage Road and US 46 to the southeast of Franklin Road is open grassland. The area between Savage Road and US 46 to the northwest of Franklin Road is dense scrubland.
Figure 1. A plan of the evaluation site.
This complicated pattern of lanes means there are six possible routes for drivers to pass through the junction between Franklin Road and Savage Road. They are shown in Figure 2. A triangular island is placed in the middle of the junction to guide drivers turning from Savage Road into Franklin Road. For pedestrians, the crosswalks being evaluated cross Savage Road from either side of Franklin Road. There is no crosswalk across Franklin Road until the intersection between Franklin Road and US Route 46.
Figure 2. The six directions of traffic flow through the evaluation site.
Throughout the series of evaluations the lighting of the junction of Franklin Road and Savage Road has been provided by two drop-lens cobrahead luminaires, each one containing a 250 W high-pressure sodium discharge lamp (Figure 3). Each luminaire is mounted on an extension arm fixed to a wooden utility pole. The two poles are located close to the two crosswalks.
Figure 3. The cobrahead luminaires used to light the evaluation site at night.
Changes to the crosswalks
Original state of the crosswalk
The first evaluation was made on Saturday, 12th June 1999. At this time, there was only one marked crosswalk at the site, that being crosswalk 2 on the northwest side of the junction of Franklin Road and Savage Road. This crosswalk was linked to a sidewalk on the northwest side of Franklin Road. There was no sidewalk on the southeast side of Franklin Road so there was no marked crosswalk 1 at this point. Figure 4 shows the marking of crosswalk 2 from the sidewalk on Savage Road on the day of the evaluation. It is clear that the original marking of crosswalk 2 consisted of two parallel lines across the road, although by the time of the evaluation, the lines had been badly eroded by traffic.
Figure 4. Crosswalk 2 at the time of the first evaluation.
Striping
The second evaluation took place on Saturday, 13th May 2000. By this time, a sidewalk had been constructed on the southeastern side of Franklin Road and a new crosswalk across Savage Road (crosswalk 1), complete with ADA ramps, had been finished. Both crosswalks had been newly striped in a grating pattern. Specifically, the striping consisted of alternate 2-ft-wide bars of white paint and 2-ft-wide pieces of asphalt, arranged to form a grating pattern, the upper and lower boundaries of the grating being closed by a continuous white paint line of 9" thickness. The overall width of both crosswalks was 7 ft. Figure 5 shows a view of crosswalk 1 at this stage and Figure 6 does the same for crosswalk 2.
Figure 5. Crosswalk 1 at the time of the second evaluation.
Figure 6. Crosswalk 2 at the time of the second evaluation.
The in-pavement flashing warning light installation
The third evaluation took place on Saturday, September 23rd, 2000, one week after the completion of the installation of the in-pavement flashing warning lights. The in-pavement flashing warning lights (Model ZA230) were purchased from Traffic Safety Corporation of Sacramento, California and installed by the New Jersey Department of Transportation. Figure 7 shows a close-up of one of the in-pavement flashing warning light units as installed.
Figure 7. One of the in-pavement flashing warning lights, installed.
The unit was originally designed for use on airport runways. Consequently, it is of rugged construction and designed to be set low enough in the pavement (projecting less than 0.50 inches above grade) that it will not be damaged by snowploughs. The light source used in the unit is a 45 W tungsten halogen lamp. The light output of the unit is predominantly in two directions, through apertures set at 180° to each other. The unit is installed so that these directions are along the main axis of the road. In Denville, only the direction towards approaching traffic emits light (the other direction is sealed). The light distribution from the unit is determined by the clear lens in the aperture of the unit. The effect of the lens is to direct a high luminous intensity beam along the road in the direction of approaching drivers. The color of the flashing warning lights was white. Four ZA 230 units were installed on crosswalk 1 and six units on crosswalk 2, as shown in Figure 8. Figures 9 and 10 show the installations at crosswalks 1 and 2 respectively.
Figure 8. A plan showing the arrangement of the in-pavement flashing warning lights installed on the two crosswalks.
Figure 9. Crosswalk 1 with the in-pavement flashing warning lights installed and operating, at the time of the third evaluation.
Figure 10. Crosswalk 2 with the in-pavement flashing warning lights installed, at the time of the third evaluation.
The activation of the in-pavement flashing warning light units was by four microwave detectors (Model AD1400) also purchased from Traffic Safety Corporation. One of these detectors was mounted on a pole at either end of both crosswalks. Figure 11 shows the location of one of the AD1400 detectors relative to the pedestrian. These devices detect the presence of a pedestrian in the detection area. The pedestrian has to be in the detection area for a minimum time before the flashing warning lights are activated. This time delay from detection to activation is necessary so that pedestrians passing though the detection area but continuing along the sidewalk do not trigger the flashing warning lights. The sensitivity of the detector, the detection area and the delay time are all adjustable on site.
Figure 11. One of the four microwave detectors used to activate the in-pavement flashing warning lights. The detector is in the small box on the pole behind and above the pedestrian.
Evaluation timing
Evaluations were made at the site on five occasions, from approximately 11 a.m. to dusk. Table 1 lists the dates of the evaluations, the status of the crosswalks, and the prevailing weather conditions. With one exception, the evaluations were made on a Saturday. This day was chosen because it was assumed that number and nature of the pedestrians using the crosswalks would be similar on that day through the summer, whereas pedestrian use on other days of the week would vary depending on the school year (The use of Saturday as the evaluation day was abandoned for the last evaluation because experience at the site taught us that the number of genuine pedestrians using the crosswalks during each hour often was insufficient to get enough data to measure drivers' reactions reliably. Consequently, it was decided to use members of the evaluation team as faux pedestrians to increase the amount of data. This decision ensured that the nature of the pedestrians using the crosswalks would be similar any day of the week). All the evaluations were made during the summer or fall. No evaluations were made during the winter because activities at Gardner Field were very limited during those months and there were few pedestrians using the crosswalks at that time.
Table 1. Day, date, status of crosswalks and the weather conditions for each evaluation
|
Evaluation Number |
Evaluation Day and Date |
Status of Crosswalks |
Weather Conditions |
|
1 |
Saturday, 6/12/1999 |
Original |
Dry and sunny |
|
2 |
Saturday, 5/13/2000 |
Striped |
Dry and sunny |
|
3 |
Saturday, 10/23/2000 |
Striped and flashing |
Initially dry and overcast, rain in late afternoon |
|
4 |
Saturday, 6/12/2001 |
Striped and flashing |
Dry and sunny |
|
5 |
Thursday, 9/13/2001 |
Striped and flashing |
Dry and sunny |
Measurements
At each evaluation, the following information was collected:
· Traffic flow was counted in five directions, each direction being observed for ten minutes in every hour. The amount of traffic exiting Gardner Field was small so the totals for directions 5 and 6 were combined.
· After the striping of the crosswalks and after the installation of the in-pavement flashing warning lights, genuine pedestrians were interviewed about their opinions of the crosswalks
· Observations were made of the behavior of drivers and pedestrians and the operational characteristics of the in-pavement flashing warning lights.
In addition, a video record of the crosswalks throughout the evaluation period was made from a position on the North side of Savage Road. This video was subsequently analyzed to obtain the following information:
· Vehicle approach speeds from directions 1 and 2, in the absence of a pedestrian and when a pedestrian was waiting to cross or was actually crossing.
· How many vehicles passed across the crosswalk while the pedestrian or group of pedestrians was waiting to cross.
· The number of times a vehicle passed over the crosswalk while there was a pedestrian, or group of pedestrians, on the crosswalk, i.e., there was conflict between vehicle and pedestrian.
Finally, a video recording was made from the front passenger seat of a vehicle driving around Denville, during daytime when the roads were dry. This video was recorded as part of Evaluations 1, 2 and 4. This video was used in the measurement of the conspicuity of the crosswalks to drivers who were unfamiliar with Denville. Accident statistics were not collected as part of the evaluation because study of the accident data in the area of the crosswalk, from 1996 to 1998, indicated that the number of accidents occurring during the evaluation period was likely to be insufficient to reach a reliable conclusion.
RESULTS
Traffic flow
The number of vehicles per hour passing over the crosswalks in each direction was estimated by multiplying the number counted during a ten-minute interval by six. The mean number of vehicles per hour passing over the crosswalks in each direction between 12.00 and 18.00 hours, for each evaluation are given in Table 2. The number of vehicles passing over the crosswalks in each direction for each hour that measurements were made are recorded in Appendix A in Tables 11 to 15.
Table 2. Mean number of vehicles per hour in each direction, between the hours of 12.00 and 18.00 hours, for each evaluation
|
Evaluation Number |
Direction 1 |
Direction 2 |
Direction 3 |
Direction 4 |
Directions 5 and 6 |
|
1 |
230 |
220 |
107 |
238 |
30 |
|
2 |
230 |
279 |
151 |
273 |
37 |
|
3 |
147 |
189 |
83 |
181 |
11 |
|
4 |
208 |
204 |
103 |
212 |
24 |
|
5 |
254 |
275 |
91 |
219 |
24 |
From Table 2 it can be seen that mean traffic flow is greatest for directions 1, 2 and 4, less for direction 3 and least for directions 5 and 6 combined. This pattern is stable over all five evaluations, although the amount of traffic is influenced by the weather. The least traffic flow for all five directions occurs on evaluation 3, when the weather started cloudy and the turned to rain in the late afternoon. It is concluded that there are no major changes in the traffic flow pattern over the five evaluations.
Driver behavior - approach speeds
The mean speed at which drivers approached the crosswalks was measured from the video record for direction 1 only. This direction requires the vehicle to travel in a straight line. All other directions require the vehicle to slow down to change direction. The approach speed for each vehicle was calculated from the time taken for the vehicle to travel a distance of 126 ft along Savage Road to the first edge of Crosswalk 1. These approach speeds were measured only for a single approaching vehicle, or for the vehicle at the front of a series of vehicles, i.e., only for vehicles whose speed was not influenced by that of vehicles immediately ahead. Approach speeds were measured without any pedestrian on or near the crosswalk, and when a pedestrian was obviously waiting to cross or was actually on the crosswalk. The mean approach speeds for direction 1, with and without a pedestrian present, between the hours of 12.00 to 18.00, for each evaluation are given in Table 3. The mean approach speeds for the hours that measurements were made at each evaluation are recorded in Appendix B in Tables 16 to 20.
From Table 3 it can be seen that the mean approach speed when a pedestrian was present is always slower than when a pedestrian was absent. The effect of the various crosswalk-marking systems can be seen by examining the changes between the different evaluations. Initially (Evaluation 1), the difference in mean approach speed for a pedestrian absent and present is 4.5 mi/h. After reconstructing one crosswalk, installing a new crosswalk, and striping both (Evaluation 2), the difference in mean
Table 3. Mean approach speed, in miles per hour, to crosswalk 1, from direction 1, with and without a pedestrian present, for the hours between 12.00 and 18.00, for every evaluation.
|
Evaluation Number |
Mean approach speed without a pedestrian (mi/h) |
Mean approach speed with a pedestrian present (mi/h) |
Difference in mean approach speed |
|
1 |
29.5 |
25.0 |
4.5 |
|
2 |
34.1 |
24.2 |
9.9 |
|
3 |
27.7 |
21.6 |
6.1 |
|
4 |
26.0 |
23.0 |
3.0 |
|
5 |
28.7 |
27.5 |
1.2 |
approach speed increases to 9.9 mi/h, mainly because of an increase in approach speed when no pedestrian is evident. Immediately after installing the in-pavement flashing warning lights (Evaluation 3) the difference in mean approach speeds decreases to 6.6 mi/h, but the mean approach speed when a pedestrian is present is a minimum for all the evaluations (21.6 mi/h). Over the next year (Evaluations 4 and 5) the difference in mean approach speeds decreases as the mean approach speed when a pedestrian is present increases until it is almost the same as when no pedestrian is present.
Driver behavior - crossing while pedestrians waiting
The number of vehicles that passed over the crosswalk while a pedestrian or group of pedestrians was waiting was counted from the video record. To be included in this total, a vehicle had to cross the crosswalk while a pedestrian was standing at the edge of the crosswalk, clearly waiting to cross. The number of crossing events measured and mean number of vehicles crossing per event, between the hours of 12.00 to 18.00, for each evaluation, are given in Table 4.
Table 4. Number of pedestrian crossing events and the associated mean number of vehicles passing over the crosswalk while a pedestrian was waiting, per event, for the hours between 12.00 and 18.00, for every evaluation.
|
Evaluation Number |
Number of pedestrian crossing events |
Mean number of vehicles passing over the crosswalk, per event |
|
1 |
60 |
1.52 |
|
2 |
71 |
1.72 |
|
3 |
137 |
1.33 |
|
4 |
82 |
1.06 |
|
5 |
155 |
1.33 |
The number of crossing events measured and the number of vehicles crossing while a pedestrian was waiting, for the hours that measurements were made at each evaluation are recorded in Appendix C in Tables 21 to 25. From Table 4 it can be seen that the number of vehicles passing over the crosswalk while a pedestrian was waiting to cross, per crossing event, was less for evaluations 3 - 5 than for evaluations 1 - 2, i.e., after the installation of the in-pavement flashing warning lights.
Driver behavior - conflicts on the crosswalk
Another measure of driver behavior made from the video record was the number of times conflicts between pedestrians and drivers occurred over the use of the crosswalk. A conflict is defined as an occasion when a driver moves over the crosswalk while a pedestrian is on the crosswalk, the vehicle passing either in front or behind the pedestrian. The number of crossing events measured and mean number of conflicts per event, between the hours of 12.00 to 18.00, for each evaluation, are given in Table 5. The number of crossing events measured and the number of conflicts occurring, for the hours that measurements were made at each evaluation are recorded in Appendix C in Tables 26 to 30.
Table 5. Number of pedestrian crossing events and the associated mean number of conflicts per event, for the hours between 12.00 and 18.00, for every evaluation.
|
Evaluation Number |
Number of pedestrian crossing events |
Mean number of conflicts per event |
|
1 |
60 |
0.32 |
|
2 |
71 |
0.00 |
|
3 |
137 |
0.05 |
|
4 |
82 |
0.06 |
|
5 |
155 |
0.05 |
From Table 5 it can be seen that the mean number of conflicts per crossing event was dramatically reduced by striping the crosswalks so that they were clearly identified as crosswalks, a status that was doubtful at the time of the first evaluation. Specifically, before striping the probability that a vehicle would cross either behind or in front of you
while you were on the crosswalk was 32%. Once the crosswalks were properly striped this probability was reduced to zero. Over time, the probability increased slightly to about 5%. Adding the in-pavement flashing warning lights did not change this probability.
Pedestrians' opinions
As part of the second evaluation, after the new striping of the crosswalks, and during the fourth evaluation, approximately nine months after the installation of the in-pavement flashing warning lights, a questionnaire about the perceived safety of the crosswalks was given to pedestrians using the crosswalks. The questionnaire used is given in Appendix D.
The characteristics of the pedestrians who answered the questionnaire on each occasion are given in Table 6. Table 6 gives a picture of the everyday use of the crosswalk. Clearly, the crosswalk is not heavily used; it took several hours to collect twenty interviews from genuine pedestrians. This pattern is also consistent with the frequency of use of the crosswalk reported by the interviewees. When the crosswalk is used, the people using it are almost equally divided between teenagers / young adults and mature adults. The unexpectedly high proportion of teenage users probably reflects their lack of access to a car and the presence of Gardner Field on one side of the crosswalk. The most interesting response in Table 6 is the high percentage of interviewees who claim that they had themselves been involved in an accident or near-miss or had seen an accident or near-miss at the crosswalk. Presumably, these claims refer mainly to near-misses, because police accident statistics do not show a high level of accidents reported.
Table 6. Demographic details of the pedestrians interviewed after the striping of the crosswalks (Evaluation 2) and after the installation of the in-pavement flashing warning lights (Evaluation 4). Data is given on the number of pedestrians interviewed, their gender and age, the average number of times they used the crosswalk per week, and whether they had had or had seen an accident or near-miss involving a pedestrian at the crosswalk.