Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
103 A p p e n d i x A Case Study 1: Alameda County Use of Continuous Count Patterns to Compare Short Pedestrian Counts Alameda County, California, used approximately 1 month of automated count data to iden- tify patterns of pedestrian activity at 13 sidewalk locations in 2008. Having continuous pedes- trian activity pattern data helped address common challenges faced when comparing short (e.g., 2-hour) counts collected at different times of the week. These challenges include account- ing for differences in activity level volumes by time of day, differences in activity patterns by land use, and differences in volumes by weather condition. With the help of the UC Berkeley Safe Transportation Research and Education Center (SafeTREC), Alameda County identified relationships among activity levels occurring during different time periods. The percentages of pedestrian volume represented by each hour of the week at all 13 automated pedestrian counter locations in Alameda County were averaged to create a composite weekly pedestrian volume profile (see Figure A-1). For example, SafeTREC estimated that 1.07% of the weekly pedestrian volume occurred between 12 p.m. and 1 p.m. on Wednesday. If analysis of these data over time proved this volume ratio to be relatively consistent and pre- dictable based on observations at all locations, one could use the number of pedestrians counted between 12 p.m. and 1 p.m. on a Wednesday at another location to estimate that locationâs weekly pedestrian volume. If 100 pedestrians were counted during this hour at a particular location, the weekly volume estimate would be about 9,300 pedestrians (100/0.0107 = 9,346 ). Using this same weekly pattern, a count of 100 pedestrians at a different location on Saturday afternoon would give a comparable weekly volume estimate of about 12,000 pedestrians (100/0.0084 = 11,905). Similar activity profiles can be developed for seasons of the year, allowing the weekly estimate to be projected to an annual volume based on the time of the sample and seasonal adjustment factor. However, Alameda County and SafeTREC recognized that pedestrian patterns are not consis- tent at all locations. Patterns vary depending on land uses around a count location. For example, compared to other automated counter locations in Alameda County, the loca- tions in employment centers (defined as locations with more than 2,000 jobs within 0.25 miles) had a greater proportion of their weekly pedestrian volumes during mid-day hours on weekdays (see Figure A-2). Employees in these areas may go out to lunch or to business meetings during the middle of the day. Therefore, counts taken at locations in employment centers should be extrapolated based on the typical employment center pattern rather than the general county pattern. In an employment center, if 100 pedestrians were counted between 12 p.m. and 1 p.m. Case Studies
104 Guidebook on pedestrian and Bicycle Volume data Collection Source: Robert Schneider, UC Berkeley SafeTREC. 0.0% 0.2% 0.4% 0.6% 0.8% 1.0% 1.2% 1.4% 12 A M 4 A M 8 A M 12 P M 4 PM 8 PM 12 A M 4 A M 8 A M 12 P M 4 PM 8 PM 12 A M 4 A M 8 A M 12 P M 4 PM 8 PM 12 A M 4 A M 8 A M 12 P M 4 PM 8 PM 12 A M 4 A M 8 A M 12 P M 4 PM 8 PM 12 A M 4 A M 8 A M 12 P M 4 PM 8 PM 12 A M 4 A M 8 A M 12 P M 4 PM 8 PM ruoHrep e muloV nairtsedeP ylkee WfotnecreP Hour of Week M T W Th F Sa Su Figure A-1. Alameda County typical weekly pedestrian volume pattern. Source: Robert Schneider, UC Berkeley SafeTREC. Figure A-2. Alameda County employment center vs. typical pedestrian volume pattern.
Case Studies 105 on a Wednesday, the weekly volume estimate would be about 7,500 pedestrians (100/0.0133 = 7,519), a difference of about 1,800 pedestrians compared to the weekly estimate produced by the unadjusted factor using the general county pattern. Continuous data from automated counters also showed that pedestrian volume patterns are affected by weather conditions. For example, compared to the average count for a par- ticular hour of the week, pedestrian volumes were approximately 5% lower when it was cloudy (i.e., they were about 0.95 times as high as a typical count). Therefore, to make short counts taken by manual data collectors on cloudy days comparable to counts taken on sunny days, Alameda County and SafeTREC divided the count by 0.95. For example, a volume of 100 pedestrians on a cloudy day would be equivalent to an average-day volume of 105 pedes- trians (100/0.95 = 105). For additional information, see Schneider, Arnold, and Ragland (2009b). This example adjustment does not account for how people may respond to weather conditions at different times of day. In actuality, weather adjustments may be lower during commute times and higher at times of day when there are more discretionary walking and bicycling trips (e.g., people may walk to lunch on a warm, sunny day but eat in the building cafeteria on a cold, rainy day). Case Study 2: Arlington County, Virginia Using Continuous Count Data to Achieve Multiple Purposes Arlington County, Virginia, started its pedestrian and bicycle counting program in 2009 with a single automated counter on a popular multi-use trail. The program expanded to include counters at more than 30 locations by 2012. The initial purpose of the program was to provide a baseline of pedestrian and bicycle volume data that could be considered along with volumes from other modes. Continuous data from the counter have shown expected seasonal patterns in bicycle use as well as overall growth in bicycle activity. For example, most monthly bicycle counts in 2011 were higher than the same monthly counts in 2010 (see Figure A-3). Source: Arlington County, VA (2012). Figure A-3. Custis Trail bicycle volume pattern.
106 Guidebook on pedestrian and Bicycle Volume data Collection Although the initial purpose was to provide baseline data on pedestrian and bicycle volumes, the Arlington County counting program has served additional purposes. For example, the mea- sured activity patterns show that bicycles are used for various reasons. Many counters in Arlington County document regular weekday morning and evening peaks, suggesting that people use bicycles to commute to and from work. Counters also show a single, mid-day peak on weekends and holi- days, suggesting that bicycles may also be used regularly for recreation, shopping, and other social activities. In addition, Arlington Countyâs automated counts show the influence of weather on bicy- cling. Comparing a normal weekday with a rainy weekday showed that rain may reduce bicycle commuting to just 25% of normal levels (see Figure A-4). Snow also has significant effects on bicycling levels. In particular, when several feet of snow were not cleared from the Custis Trail for 2 weeks, regular bicycle commuting levels were reduced to zero (Figure A-5). Arlington County pedestrian and bicycle program staff were able to use these data to illustrate the effect of poor winter maintenance and to show that many bicyclists could benefit from maintaining the trail year-round. For additional information, see Arlington County (2012). Source: Arlington County, VA (2012). Figure A-4. Illustrative impact of rain on bicycle volume. Source: Arlington County, VA (2012). Figure A-5. Illustrative impact of uncleared snow on bicycle volume.
Case Studies 107 Case Study 3: San Francisco, California Pedestrian Volume Patterns Provide Data for a Community-Wide Demand Model The San Francisco Municipal Transportation Agency (SFMTA) and San Francisco County Transportation Authority (SFCTA) used continuous automated count patterns to estimate annual pedestrian volumes at 50 intersections throughout the city. These annual volumes were used as the basis for an intersection pedestrian volume model developed by SafeTREC and Fehr & Peers Transportation Consultants. Accurate annual estimates were critical for developing the best pos- sible model. Although the researchers could have applied the same expansion factors to 2-hour counts at all locations, they based their annual volume estimates on more than 20 different automated count patterns (see Figure A-6). The pedestrian volume patterns at the intersection crossings were assumed to match the overall pedestrian volume patterns on adjacent sidewalks where the automated counters were located (because automated technologies were not available to detect pedestrians crossing the street). Statistical modeling showed that annual pedestrian volumes at San Francisco intersections were positively associated with household density, job density, high-activity zones with parking Source: Fehr & Peers Transportation Consultants and UC Berkeley SafeTREC (2011). Figure A-6. Example pedestrian volume patterns near San Francisco intersections.
108 Guidebook on Pedestrian and Bicycle Volume Data Collection meters, proximity to a university campus, and traffic signals, and were negatively associated with steep slopes. The model equation was then used to estimate pedestrian volumes at all 8,100 inter- sections in San Francisco (see Figure A-7). The model results are being used by SFMTA and SFCTA to support the following policy goals: (1) reduce the absolute number of severe injuries and fatalities; (2) improve walking conditions in areas with elevated crash risk; and (3) implement effective safety measures. For additional information, see Schneider et al. (2012). Case Study 4: University of California, Berkeley Pedestrian Volume Patterns Provide Exposure Data for Safety Analysis SafeTREC used long-term pedestrian volume patterns to control for pedestrian exposure and estimate pedestrian crash risk at 22 intersections along the campus boundary during typical spring and fall semester weekdays. Continuous count patterns were collected at three locations on campus (counts from one of the locations are shown in Figure A-8). Two-hour manual counts at each of the 22 intersections were extrapolated to a 10-year volume estimate based on the pedestrian activity pattern from the closest automated counter. Estimated 10-year volumes were then compared with the number of Model esmated pedestrian crossings per year (millions) Source: Fehr & Peers Transportation Consultants and UC Berkeley SafeTREC (2011). Figure A-7. Estimated San Francisco intersection pedestrian volumes.
Case Studies 109 reported crashes at each intersection during a 10-year period. Crash rates were also calculated by hour of the day. Results showed that pedestrian crash risk was generally higher at intersections along the boundary roadways with the lower pedestrian volumes. In addition, pedestrian risk in the eve- ning (6 p.m. to midnight) was estimated to be more than three times higher than in the daytime (10 a.m. to 4 p.m.) (see Figure A-9). For additional information, see Schneider, Grembek, and Braughton (2013). Case Study 5: Washington State Department of Transportation Volunteers Collect Annual Pedestrian and Bicycle Counts The Washington State Department of Transportation (WSDOT) has worked with the Cascade Bicycle Club (CBC) since 2008 to implement an annual volunteer-based statewide pedestrian and bicycle count program. In 2012, nearly 400 volunteers collected manual counts at more than 200 locations in 38 different cities. This level of coverage would not be possible for the Source: UC Berkeley SafeTREC (2012). Figure A-8. Pedestrian volume patterns at one UC Berkeley campus entrance.
110 Guidebook on pedestrian and Bicycle Volume data Collection state without volunteer resources. Each community has a Local Count Coordinator (often from a local transportation agency) who is responsible for organizing the local counts and reporting to WSDOT. Each year, volunteers conduct 2-hour counts from 7 a.m. to 9 a.m. and from 4 p.m. to 6 p.m. on a Tuesday, Wednesday, or Thursday in late September. Most counts are at intersections, but some are along multi-use trail, roadway, and sidewalk segments. Volunteers are given counting instructions and data collection forms prior to the count datesâthis approach helps data col- lectors prepare and likely helps reduce counting errors. Although the counts are still tallied on paper forms and can be sent to WSDOT for database entry, WSDOT now offers a webpage where volunteers can enter their count data directly. In addition, CBC created an online system to help count volunteers register and select count times and locations. Results from 83 locations that have been counted consistently in the morning and 64 locations that have been counted consistently in the afternoon each year since 2009 have shown overall increases in walking and bicycling (see Figure A-10). In addition, the manual counting approach has made it possible to document helmet use as well as pedestrian and bicyclist gender in different communities throughout the state. For more information, see Washington State DOT (2012). Case Study 6: Columbus, Ohio Three Organizations Coordinate to Collect Trail Counts The City of Columbus, Ohio, partnered with the Mid-Ohio Regional Planning Commission (MORPC) and the Rails-to-Trails Conservancy (RTC) in 2010 to document trail use (pedestri- ans plus bicyclists) along several major corridors (see Figure A-11). Although none of the three organizations had an extensive budget for counting, their partnership made it possible to collect All 22 Boundary Roadway Intersections Time Period Reported Pedestrian Crashes Estimated Crossing Volume Crashes/10M Crossings 00:00-05:59 0 2,025,899 0 06:00-07:59 1 8,025,759 1.25 08:00-09:59 6 45,451,089 1.32 10:00-11:59 7 48,181,827 1.45 12:00-13:59 7 56,791,023 1.23 14:00-15:59 5 53,333,999 0.94 16:00-17:59 11 51,804,940 2.12 18:00-19:59 14 35,643,980 3.93 20:00-21:59 6 17,638,689 3.40 22:00-23:59 3 5,345,865 5.61 Total 60 324,243,069 1.85 Source: Schneider, Grembek, and Braughton (2013). Figure A-9. Pedestrian crash risk at campus boundary intersections by time of day.
Case Studies 111 Source: Cascade Bicycle Club (2013). Figure A-10. Washington State pedestrian and bicycle counts, 2009 to 2012. Source: Mid-Ohio Regional Planning Commission (2012). Figure A-11. Estimated annual trail volume by location.
112 Guidebook on pedestrian and Bicycle Volume data Collection a rich set of information about trail use patterns. The City of Columbus funded the study, pur- chasing three passive infrared counters and installing them permanently at three locations. RTC rotated several temporary counters to collect 2 monthsâ worth of data at seven other locations. MORPC organized and analyzed the data. The 2-month counts were extrapolated using seasonal volume patterns from two of the locations that had continuous 2-year data. The data were used to create a report with useful information about changes in trail activity patterns by time of day, day of week, and season of year (see Figures A-12 and A-13). According Source: Mid-Ohio Regional Planning Commission (2012). Figure A-12. Estimated annual trail volume by location, 2010â2012. Source: Mid-Ohio Regional Planning Commission (2012). Figure A-13. Estimated Olentangy Trail annual volume by month.
Case Studies 113 to MORPC, âThe counts are meant to serve as a baseline to document changes over time, while also assisting with grant applications, providing information to elected officials, and supporting/ justifying budget decisions. The trail counts inform the process of evaluating whether to widen selected trails.â Case Study 7: San Diego County Systematic Process Used to Select Permanent Count Sites The County of San Diego Health and Human Services Agency, San Diego Association of Gov- ernments, and San Diego State University have partnered to install automated pedestrian and bicycle counters throughout the region. Funding was provided by a Centers for Disease Control and Prevention Community Putting Prevention to Work (CPPW) grant. Once completed, this system is likely to become the largest set of permanent counters used for pedestrian and bicycle traffic monitoring in the United States. The first set of more than 30 automated count locations was selected using four criteria: 1. Locations along the existing or planned regional bicycle network; 2. Locations with a Smart Growth Opportunity Area (i.e., mixed-use, high-density infill devel- opment consistent with SANDAGâs Regional Comprehensive Plan); 3. Geographic variety (i.e., covering areas throughout the region); and 4. Demographic variety (i.e., covering a range of population and employment densities and median household incomes). A stratified sampling approach was used to achieve the fourth criterion. This approach used census data to define high, medium, and low categories of population density, employment den- sity, and median household income. The 27 sampling strata and their geographic distribution are shown in Figure A-14. For additional information, see Ryan (2013). Case Study 8: Colorado Department of Transportation Automated Counters Are Used to Identify Common Bicycle Volume Patterns The Colorado Department of Transportation used automated counters to identify specific types of pedestrian and bicycle activity patterns on multi-use trails. Counts were collected on more than 20 trails. Three distinct trail usage patterns, or factor groups, were identified using cluster analysis: ⢠Mountain Non-Commute. This pattern was typically observed in rural, mountainous areas (see Figure A-15). There was often a high level of weekend and monthly variation in volumes. ⢠Front-Range Non-Commute. This pattern was typically observed in the Front-Range (more urbanized) region and was associated with bicycling for recreational or non-commuting trip purposes. Some rural mountain locations with higher utilitarian bicycling were also included. This pattern tended to have high weekend variation and low monthly variation. ⢠Commute. This pattern was typically observed in the urban and suburban Front Range region and in urban Mountain communities (see Figure A-16). There were often distinct morning and afternoon peaks in activity. This pattern often had relatively low weekend and monthly variation. These patterns are used to extrapolate short-duration counts to represent longer time periods, such as annual trail user volumes. For additional information, see Nordback, Michael, and Janson (2013).
114 Guidebook on Pedestrian and Bicycle Volume Data Collection Source: Nordback, Michael, and Janson (2013). Figure A-14. San Diego automated counter locations. Source: Nordback, Michael, and Janson (2013). Note: Inverse daily factor is the percentage of the average daily volume observed on each specific day. Each legend item represents a different bicycle monitoring location. The bold line is the average of all locations. Figure A-15. Mountain non-commute trail usage pattern.
Case Studies 115 Case Study 9: Minneapolis Counts Are Used to Map Pedestrian and Bicycle Volumes Throughout the Community The City of Minneapolis, Minnesota, began collecting annual pedestrian and bicycle counts in 2007. With the assistance of Transit for Livable Communities (TLC), the City has collected counts each September at 23 consistent locations for pedestrians and 30 consistent locations for bicyclists. These consistent count locations have documented increases in walking and bicycling. However, the City also has more than 300 additional non-motorized count locations counted once every 3 years. Most counts are conducted from 4:00 p.m. to 6:00 p.m. or from 6:30 a.m. to 6:30 p.m., and models are used to extrapolate these counts to 24-hour volumes for a typical September weekday. Data from the 300 count locations are used to create maps of estimated daily pedestrian and bicyclist volumes on a typical September weekday (see Figure 3-5 in Chapter 3). For additional information, see Minneapolis Department of Public Works (2013). Case Study 10: Five North American Cities Classifying Bicycle Traffic Patterns in Five North American Cities Researchers from McGill University and the UC Berkeley Safe Transportation Research and Education Center analyzed continuous count data to identify similarities between bicycle ridership patterns across different North American communities. Long-term bicycle counts were gathered from 38 locations in five cities (Montreal, Ottawa, Portland, San Francisco, and Vancouver) and along the Green Route in Quebec. This study introduced a classification scheme for analyzing bicycle traffic patterns. Among other findings, the analysis showed that the bicycle volume patterns fell into one of the follow- ing classifications: utilitarian, mixed-utilitarian, mixed-recreational, and recreational (see Fig- ure A-17). Study locations classified into each of these categories were found to have consistent hourly and weekly traffic patterns across cities, despite differences between these cities in terms of factors such as weather, size, and urban form. Seasonal patterns across the four categories and in the different cities were also identified. The study presents expansion factors for each category by hour and day of the week. Monthly expansion factors are also presented for each city. Finally, traffic volume characteristics are presented for comparison purposes. Source: Colorado Department of Transportation (2013). Figure A-16. Commute trail usage pattern.
116 Guidebook on Pedestrian and Bicycle Volume Data Collection 1.The pictured profiles are the mean values of the facilities belonging to each classification. Source: Miranda-Moreno et al. (2013). Hourly Profiles1 Daily Profile1Type Weekday Weekend Pr im ar ily U l ita ria n (P U ) U litarian loca ons exhibit two dis nct weekday peaks, much like automobile commuter paerns, and have much higher ridership during the week than on the weekend. The weekend profile builds smoothly to a single PM peak. In general, they maintain the highest ridership in the winter. M ix ed -U l ita ria n( M U ) Mixed u litarian loca ons s ll exhibit two peaks at the hourly level on weekdays, though the level of ridership between the peaks may be slightly higher than at primarily u litarian loca ons. The difference between weekday and weekend ridership is much less pronounced, and may even be negligible. Weekend ridership builds gradually to a PM peak, similar to primarily u litarian loca ons. They may retain less ridership in the winter than PU loca ons. M ix ed Re cr ea o na l( M R) Mixed recrea onal loca ons tend to maintain a consistent level of daily ridership throughout the week. However, unlike mixed-u litarian, their hourly profiles do not exhibit two dis nct commu ng peaks. S ll, their early AM ridership during the workweek may be slightly higher than primarily recrea onal loca ons.The daily profile may exhibit slightly higher ridership on the weekend. Ridership at these loca ons is generally considerably lower than PU or MU loca ons in the winter. Pr im ar ily Re cr ea o na l( PR ) Primarily recrea onal loca ons are typically in parks or serve recrea onal areas. They exhibit considerably higher ridership on the weekend than during the week. The workweek hourly profile closely resembles the weekend profile, which increases steeply to and decreases steeply from a mid day plateau. A slight dip around noon may be present as well. The decrease in ridership due to winter is most significant at recrea onal loca ons. Figure A-17. Summary of classification of bike traffic patterns.
