National Academies Press: OpenBook

Guide to Pedestrian Analysis (2022)

Chapter: Chapter 2 - Pedestrian Volume Counting

« Previous: Chapter 1 - Introduction
Page 5
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 5
Page 6
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 6
Page 7
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 7
Page 8
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 8
Page 9
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 9
Page 10
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 10
Page 11
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 11
Page 12
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 12
Page 13
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 13
Page 14
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 14
Page 15
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 15
Page 16
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 16
Page 17
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 17
Page 18
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 18
Page 19
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 19
Page 20
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 20
Page 21
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 21
Page 22
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 22
Page 23
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 23
Page 24
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 24
Page 25
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 25
Page 26
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 26
Page 27
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 27
Page 28
Suggested Citation:"Chapter 2 - Pedestrian Volume Counting." National Academies of Sciences, Engineering, and Medicine. 2022. Guide to Pedestrian Analysis. Washington, DC: The National Academies Press. doi: 10.17226/26518.
×
Page 28

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.

5   C H A P T E R 2 Pedestrian volume data are increasingly being used by transportation engineers and plan- ners. Applications range from prioritizing safety improvements on a city- or region-wide basis to measuring the usage of new pedestrian facilities and even to designing specific pedestrian improvements. Although most transportation agencies have long collected motor vehicle count data, and best practices for doing so are well established (1), the field of pedestrian data collection con- tinues to evolve. Challenges that practitioners experience when trying to collect pedestrian data include the lack of funding, staff time, technological tools, organizational interest, a defined agency need for pedestrian count data (2, 3), and information on best practices for collecting and using the data. To address practitioners’ need for information on how best to use their resources when collecting and applying pedestrian count data, a number of research efforts have been conducted in recent years at the national and state levels to gather the experience of transportation agencies leading the field in establishing best practices for developing count programs and overcoming barriers to pedestrian data collection. This chapter builds on recent work to document the current state of the practice. Collection of Pedestrian Volume Data This section provides a high-level overview of pedestrian volume data collection, divided into the following tasks: • Identifying count applications, • Identifying count durations and locations, • Identifying counting method(s), • Ensuring data quality, and • Identifying the need for supplemental data. More information is available in the following documents, which are referenced throughout Chapters 2 and 3: • The Traffic Monitoring Guide (1) offers guidance on collecting traffic data. It includes a chapter on traffic monitoring for nonmotorized (i.e., pedestrian and bicycle) traffic and defines a format for reporting nonmotorized counts. The guide discusses key differences between monitoring for motorized and nonmotorized traffic and outlines a process for developing permanent and short-term nonmotorized data collection programs. It also briefly explains each of the available bicyclist and pedestrian counting technologies and their potential application. Pedestrian Volume Counting The lack of pedestrian and bicycle volume data is a barrier to transportation agency efforts to plan more effective facilities and to improve safety for pedestrians and bicyclists.

6 Guide to Pedestrian Analysis • NCHRP Report 797: Guidebook on Pedestrian and Bicycle Volume Data Collection (2) pro- vides guidance on developing and applying nonmotorized count programs. The guidebook covers – Count applications, including case studies; – Planning and implementing a count program, including checklists and case studies; – Correcting raw count data to account for site- and product-specific counting errors; – Expanding short-term count data to estimate longer-duration volumes; and – Counting technologies, including their typical applications, strengths and limitations, relative costs, installation needs, and accuracy. A companion report, NCHRP Web-Only Document 229 (4), updates NCHRP Report 797’s information on various counting technologies with additional information developed by a follow-up research effort. • FHWA’s Exploring Pedestrian Counting Procedures reviews pedestrian counting method- ologies “to identify key issues and recommend creative strategies for developing accurate, timely, and feasible approaches for measuring pedestrian travel” (5, p. 12). It provides a literature review; findings from an interactive webinar with practitioners; and information on equip- ment for collecting pedestrian count data, strategic considerations for counting programs, data management, and pedestrian counting techniques and procedures. In comparison with the two documents described above, it offers more detailed information about data management procedures and best practices. The remainder of this chapter provides information about the key activities involved in col- lecting pedestrian volume data and sources for further information. While some guidance is oriented towards larger count programs incorporating a variety of locations and data types, it can be applied to any pedestrian counting effort. Getting Started As noted in this chapter’s introduction, lack of organizational interest or support in perform- ing pedestrian counting (or other types of pedestrian analyses) may need to be overcome. One possible approach that has been successful is to start small and to grow the counting program once the value of the initial count applications has been demonstrated. Transportation agencies lacking funding to perform counts on their own can consider partnering with service or advocacy organizations to organize volunteer counts. Another option is to partner with other organizations (e.g., a regional planning agency, a health-related organization) to collect or share data (2). Potential Applications Pedestrian volume data have a variety of useful applications. However, count data are only useful when they have a clear purpose and the intended use of the data has helped determine count type, duration, frequency, and location. NCHRP Report 797 (2) includes a chapter on nonmotorized count data applications and is an excellent resource for understanding potential motivations for collecting pedestrian data. Key applications from the guidebook are shown in Figure 2-1. Selecting Count Durations and Locations Count Durations There are two primary count types, continuous and short duration: • Continuous counts. Continuous counts at permanent sites provide information on temporal trends. Given that there are both one-time and ongoing costs associated with installing and operating an automated pedestrian counter, most jurisdictions that use permanent automated

Pedestrian Volume Counting 7   counters typically install them at a small number of their data collection sites (2). ese con- tinuous counts can be used to track usage of the agency’s most important pedestrian facilities. As described later in this chapter, continuous counts can also be used to adjust short-term counts to account for weather- and land use-related demand variations and to expand short- term counts to estimate longer-term volumes such as annual average daily pedestrians. • Short-duration counts. Short-duration counts can provide anywhere from an hour to a few weeks of data. ey are typically used to expand the geographic reach of a counting program, for example, by rotating a small set of automated counters around dierent count sites or by using volunteers to perform a large number of manual counts on the same day. Short-duration counts can also be used to capture pedestrian attributes not available from automated counts, such as pedestrian gender and age group (2). Short-term counts may also be obtained as part of a multimodal count eort; for instance, a two-hour intersection turning movement count may include motor vehicles, bicyclists, and pedestrians. e strongest data collection programs typically combine a variety of data types to provide spatial and temporal coverage. Limited resources usually mean that all locations where an agency might desire data cannot be counted; therefore, the agency needs to prioritize the purposes for which it collects data. A key consideration when short-term counts are conducted is that pedestrian volumes at a given location are typically more variable than motor vehicle volumes. On top of seasonal variations in temperature, precipitation, humidity, and the number of daylight hours, day-to- day variations in precipitation and temperature can substantially aect pedestrian activity (1). Additionally, because pedestrian volumes are normally low relative to motor vehicle trac, more day-to-day variability (on a percentage basis) is seen in pedestrian counts (2). As a result, the shorter the count duration, the less likely the raw counted volume will represent typical volumes at that location at that time of year. One approach to addressing the variability in pedestrian volume is to adjust short-duration counts by using information from continuous counts. is process is described later in this Measuring facility usage as part of routine system monitoring, including tracking changes in usage over time. Evaluating before-and-after volumes to assess the impact of facility improvements or changes. This activity may help justify further system improvements and build support for investments in pedestrian infrastructure. Monitoring travel patterns to help identify the extent to which factors such as land use patterns and transportation network characteristics influence local walking activity. Safety analysis through quantifying pedestrian exposure, as discussed in Chapter 3. Project prioritization by identifying essential networks or indicating areas where improvements may be needed on the basis of user behavior. Multimodal model development to forecast future demand over a large transportation network. Pedestrian Model Calibrated with Count Data Figure 2-1. Key applications from NCHRP Report 797 (2). Graphic courtesy of the City of Berkeley, CA (6).

8 Guide to Pedestrian Analysis chapter. The continuous count data provide insight into whether pedestrian activity on a given day or at a given time was higher or lower than normal, along with information about longer- term seasonal volume trends. While having one continuous count station can provide valuable information about variation in pedestrian activity, the Traffic Monitoring Guide recommends using three to five continuous count stations for each travel pattern (“factor group”) of interest (1). At a city level, these groups might represent a downtown area, a single-family residential neighborhood, recreational facilities, and so on. At a state level, these groups might represent different climatic zones or different-sized communities. The greater the number of well-placed continuous count stations, the greater the ability to estimate pedestrian volumes accurately over a range of land use and pedestrian activity patterns. In the absence of continuous count data, the preferred method is to use as long a count dura- tion as feasible. Guidance in the literature varies as to the length of the period for collecting short-duration counts. For example (5): • The National Bicycle and Pedestrian Demonstration Project suggests taking a series of 2-hour counts for up to three consecutive days or weeks, with fewer counts at high-activity locations (e.g., more than 100 people per hour in the middle of the day) and more counts at locations with low or irregular activity levels (7). • The Traffic Monitoring Guide advises that while 2-hour data are better than no data, the error obtained when factoring 2-hour counts may be high. It recommends using 12-hour counts to create a time-of-day profile (1). • Other researchers have demonstrated that counting for one full week (8, 9) or even 1 month (10) minimizes error when average annual volumes are estimated. Deciding how long to count is as important as deciding when to count. Errors in the volume estimate can be minimized by (2) • Counting at times with high activity levels (e.g., summer), • Counting during good weather to minimize the effect of seasonal variability, • Conducting several short counts during different time periods, and • Extrapolating by using a single day-of-year factor from a continuous count rather than by using standardized day-of-week and month-of-year factors. Count Locations NCHRP Report 797 (2) describes four approaches for selecting count locations: • Representative locations of the community as a whole; • Target locations associated with a particular project, facility, or issue (e.g., safety concerns); • Control locations unaffected by projects to be compared with similar locations affected by a project (target locations) to estimate how much of the observed change in pedestrian volume was caused by the project and how much was caused by background factors; and • Random locations within categories of desired characteristics or within a defined study area. Chapter 3 of NCHRP Report 797 provides further details on each of these approaches. For example, if representative locations are being selected, count locations should be located in different geographic parts of the community, surrounded by different types of land uses, and found on different types of facilities and should reflect the community’s socioeconomic char- acteristics as a whole (2). NCHRP Report 797 also emphasizes that “limiting count sites to loca- tions that are convenient, have the highest pedestrian or bicycle volumes, or are expected to have the greatest increases in walking and bicycling does not produce a representative sample” (2, p. 27).

Pedestrian Volume Counting 9   e North Carolina DOT developed a nonmotorized volume data program to complement its existing motorized trac monitoring system. e following steps were recommended for identifying nonmotorized count locations across the network (11). • Gather potential site locations. Contact local agencies and identify (a) where counts have been conducted in the past and (b) interest in program participation. Table 2-1 lists selection criteria that are critical or valuable for prioritizing sites. • Conduct a site visit. A site visit helps to rene and prioritize sites and identies whether the site is better suited for a continuous or a short-duration count. Use a checklist to gather information on the site visit. • Select continuous count stations. From information gathered in the previous steps, deter- mine which sites are best suited for continuous counters. Additional site-selection criteria may be used, such as sites with choke points, special events, a nearby school or university, or proximity to a recreational activity generator. • Select short-duration count sites. Short-duration sites should be dispersed to provide spatial and volume spread for the count program. ese sites may include locations that are infea- sible for continuous count stations. Selecting count locations depends largely on the purposes for which the counts are collected. If counts are being collected for safety improvements, other criteria, such as pedestrian crash history, should also be considered. Counting Methods Once the count application, location, and duration are identied, the next step is to determine the data collection method. ere are a variety of potential methods for collecting count data. e choice of method depends on count type (continuous or short-duration) and application. A few key questions help guide the selection of counting method: • Is a screenline count or an intersection count needed? Screenline counts count the number of pedestrians crossing an imaginary line along a segment, while intersection counts count pedestrian crossing volumes or total intersection users (Figure 2-2). Both count types may generate either total volumes or directional volumes. Screenline counts are simpler to collect, and most available automatic technologies are only applicable to screenline counts. Intersection Critical Criteria Valuable Criteria Ranking Site location Area type (urban, suburban, rural) Anticipated user type (mixed/everyday, recreation/weekend, commuter/weekday) Facility type (paved path, unpaved path, sidewalk) Type of count (pedestrian, pedestrian and bicycle) Ownership (municipal, county, state) Duration (short, continuous) Notes from virtual site audit Volume potential (high, medium, low) Existing count data (where available) or estimate of hourly volume Number of roadway lanes Sidewalks (one side, two sides, none) Agency submitting location Local contact information Notes from agency Table 2-1. Selection criteria for prioritizing sites. Figure 2-2. Screenline count versus intersection count (2).

10 Guide to Pedestrian Analysis counts can be useful for certain applications such as fine-tuning signal timing, assessing inter- section crash data, or evaluating intersection traffic operations. • Is a short-term or long-term count needed? If a short-term count is needed, it may be more effi- cient to collect data manually, either by having someone in the field record volumes, or by collect- ing video data and reducing it manually or through an automated process. If a long-term count is needed, manual counting will likely be infeasible and an automatic counter will be required. • Are there site constraints? Is the site in a location where equipment can be installed? Are there other users (such as bicyclists) who may be on the facility as well? Are pedestrians stop- ping, congregating, or walking erratically? These constraints will help guide the appropriate technology type if an automatic counter is being used. Section 3.3.2 of NCHRP Report 797 (2) provides a more-expansive list of questions to ask when selecting a counting method, while subsequent sections provide checklists for preparing and installing equipment. Several of the most common methods for collecting pedestrian counts are described in Toolbox 2-1. The information in Toolbox 2-1 is based on the Traffic Monitoring Guide (1), NCHRP Report 797 (2), and FHWA’s Exploring Pedestrian Counting Procedures (5). It is also important to consider life-cycle costs of the different equipment prior to choosing a technology. Life-cycle costs include procurement, installation, maintenance, and repair and replacement costs. Other considerations include data management and storage and vandalism of equipment. Correcting Systematic Count Errors As with all counting equipment, pedestrian counting equipment is rarely 100% accurate. Pedestrian counting can be particularly difficult in places where pedestrians walk in groups, side-by-side, holding hands, or in tightly packed crowds because such activity hides pedestrians from counting devices and makes it hard for equipment to distinguish individual pedestrians. NCHRP Report 797 (2) and its supporting web-only documents (3, 4) provide extensive testing and documentation of known systematic errors with different pedestrian counting technolo- gies. Specifically, NCHRP Web-Only Document 229 (4) documents how systematic errors can be addressed by applying correction factors or functions. Although no tested device produced 100% accurate counts, simple correction factors were developed for nearly all products and tech- nologies to provide reasonable estimates of actual pedestrian volume. Variations in accuracy between different products using the same sensor technology and between the same product used at different sites indicate that calibrating and validating a device after installation is particu- larly important (as discussed above). Device- and location-specific correction factors for each permanent counter installation, as well as device-specific correction factors for portable equip- ment used for short-term counts, are recommended. The North Carolina DOT provides a good example of how one jurisdiction creates location- and device-specific correction factors (11). For situations in which it is not possible to create a local correction factor, NCHRP Web-Only Document 229 (4) provides default correction factors for commonly used pedestrian counting equipment (Table 2-2). To apply these correction factors, select the sensor technology used to count pedestrians and multiply the raw count recorded by the device by the adjustment factor for the technology. For example, for a generic passive infrared counter, if 1,000 pedestrians are counted on a given day and the adjustment factor is 1.106, the corrected volume for that day would be 1,106 pedestrians. Similarly, for a thermal imaging camera, if 100 pedestrians are counted in an hour and the adjustment factor is 0.974, the corrected hourly volume would be 97 pedestrians.

Pedestrian Volume Counting 11   Toolbox 2-1. Common methods for collecting pedestrian counts. Counting Methods Source: Kittelson & Associates, Inc. Manual counts in the field can be collected by using paper sheets, traffic count boards, clicker counters, or smartphone apps. They are most appropriate for short-duration counts. Data collectors can capture pedestrian crossing volumes at an intersection or screenline counts along a roadway or path. Typical Usage Manual counts are typically used to obtain supplemental data such as age, gender, use of a mobility aid, or pedestrian signal compliance; when an intersection count is desired; or when a multimodal count (e.g., cars, trucks, bicycles, pedestrians) is desired. Counts can be performed by agency staff, consultants, or volunteers; in all cases, data collector training is essential to obtain good results. Advantages • Can be accurate with training • Can capture directional counts and user characteristics • Minimal equipment costs Disadvantages • High labor cost • Only for short-term counts • Data cannot be verified Ease of Installation • Requires data collector training Cost • Very high Manual counts from video are considered the most accurate method of collecting bicycle and pedestrian counts, since the video can be paused or rewound as necessary to accurately count large volumes or groups of pedestrians. As with manual field counts, user characteristics such as gender, age, and pedestrian signal compliance can also be collected. While manual counts are relatively inexpensive for short-duration counts, they are unsuitable for longer counts because of high labor costs when the data are being reduced. Typical Usage Manual counts from video have a similar application to manual counts in the field but are typically collected by a specialized data collection firm. A practitioner survey found that 44% of respondents who performed pedestrian counts used manual counts from video data as part of their pedestrian data collection program (2). Advantages • High accuracy • Can capture directional counts and user characteristics • Fewer personnel than in- field counts • Data can be verified Disadvantages • High labor cost Ease of Installation • Requires video camera set-up Cost • Very high Source: Kittelson & Associates, Inc. (continued on next page)

Counting Methods Source: Numina. Automated counts from video are a growing counting technology. Some vendors post-process video images to generate counts, at times accompanied by manual quality-control checks. Some vendors of traffic signal detection equipment use software to count pedestrians in real time from video or thermal camera images. This technology is most accurate when pedestrians and other roadway users are spatially separated. However, when modes are mixed, as they usually are at intersections, accuracy can be notably reduced (5). Typical Usage The organizations interviewed for this project who used automated counts from video were largely satisfied with the count accuracy, although lack of directional counts was cited as a drawback. Advantages • Portable • Good for crowded areas • Can be used for long-term counts • Data can be verified and potentially reviewed Disadvantages • Moderate to high equipment cost • Accuracy might not be able to be verified independently • May not be able to produce directional counts Ease of Installation • Requires video camera set-up Cost • High Source: Kittelson & Associates, Inc. Active infrared devices count objects that break an infrared beam sent between an emitter and a receiver placed on opposite sides of a pedestrian facility. Typical Usage Because both an emitter and a receiver are necessary, active infrared devices are most commonly used on shared-use paths in conjunction with a second counting method capable of counting only bicycles. Advantages • Portable • Easy installation • Can be used for long-term counts Disadvantages • Undercounting due to occlusion • Cannot distinguish bicycles from pedestrians • Directional counts not possible Ease of Installation • Requires a site where emitter and receiver can be installed facing each other Cost • Relatively low equipment cost and low hourly cost Source: Kittelson & Associates, Inc. Passive infrared devices detect warm bodies and therefore cannot distinguish bicyclists from pedestrians. Consequently, this technology is best used in a pedestrian-only environment or in combination with bicycle- counting technology (as shown here with the diamond-shaped loops buried in the pavement) to differentiate bicycles from pedestrians. Typical Usage Passive infrared counters can be used to collect data for several weeks or permanently. It is the most commonly used automated technology for counting pedestrians. Advantages • Portable and easy to install • Not affected by wet or foggy weather • Can be used for long-term counts Disadvantages • Undercounting due to occlusion • Directional counts not always possible • Cannot distinguish bicycles from pedestrians Ease of Installation • Requires proper installation training and potentially a permit; permanent installations may require installing a post to house the counter Cost • Low Toolbox 2-1. (Continued).

Pedestrian Volume Counting 13   Sensor Technology Adjustment Factor Thermal imaging camera 0.974* Passive infrared 1.106** Radio beam 1.125 *Factor is based on a single sensor at one site: use caution when applying. **Correction factor is based on a weighted average of results from three products from different vendors that had product-specific adjustments of 1.016, 1.157, and 1.369. Source: NCHRP Web-Only Document 229 (4). Table 2-2. Default correction factors for commonly used pedestrian counting equipment. Note that the default correction factors listed above do not account for bypass errors. ese errors occur when a pedestrian walks outside the device’s detection zone. Observation of pedes- trian behavior at the site is the only way to discover these types of inaccuracies. Quality Control Good data quality is critical to accurately informing decision making on new infrastructure investments and pedestrian safety. e required accuracy level may vary by purpose. While exact guidance on the required levels of data quality for dierent applications is not available, FHWA’s Exploring Pedestrian Counting Procedures (5) suggests that while safety applications require a high level of data quality, data quality within an order of magnitude of the actual value may be sucient for sketch planning and proposals. Quality control processes include validating and calibrating data collection equipment and establishing data format checks and data validity criteria for automated and manual count data. e following resources provide guidance on quality control checks. Counting Methods Source: Photo courtesy of Eco-Counter. Pressure and acoustic pads are installed underground and detect users passing directly over the sensor. Pressure pads detect weight changes, while acoustic pads detect sound wave passage. Typical Usage Pressure or acoustic pads are primarily used on unpaved trails, where they are easy to install. They can also be installed in sidewalks and used in locations where it is not possible to install an infrared counter. Advantages • Low maintenance cost • Low power consumption • Can count pedestrians on sidewalks Disadvantages • High installation cost • May not be able to differentiate groups of pedestrians • Requires users to pass directly over the sensor • Not feasible for locations where the ground freezes Ease of Installation • Requires pavement cutting if installed in sidewalks Cost • High Toolbox 2-1. (Continued).

14 Guide to Pedestrian Analysis Equipment The Traffic Monitoring Guide (1) recommends that organizations collecting count data perform the following tasks: • Test equipment and ensure it meets accuracy standards before field placement, • Calibrate equipment routinely, • Validate equipment performance periodically to ensure intended performance, • Conduct routine quality assurance tests, • Analyze and deliver data quickly so that data quality errors caught by users familiar with data patterns can be addressed quickly, and • Implement a feedback process for quick response to data quality concerns. When automated counters are used, equipment calibration and validation are key to ensur- ing robust data quality. Calibration includes performing a number of equipment tests to ensure correct functioning. The calibration process can identify both major (e.g., failed equipment) and minor (e.g., site setup) errors. These errors can significantly increase data collection costs and decrease data usability. The following elements are key to a robust traffic monitoring calibration program (1): • Implementing software tools to automate the calibration process; • Performing daily quality checks to ensure data are properly collected, processed, and stored; • Using monthly and yearly trends to determine validity; • Conducting field calibration; • Collecting manual counts to validate automated counts; and • Performing manual and electronic calibration of counting equipment annually. Data FHWA uses the following checks for all nonmotorized traffic data, including pedestrian data, as part of the Travel Monitoring Analysis System (TMAS) Version 2.8 (Steven Jessberger, personal communication, October 4, 2018): • Adjacent intervals. Check both sides of any zero interval unless the interval is the day’s first or last interval. Flag if the difference between any zero interval and an adjacent interval is greater than 50. • Consecutive zeros. For any count interval, check that there are no more than seven consecutive zero values. • Total daily count. When all count intervals and directions have been calculated, flag data if the total maximum daily count exceeds 5,000 or if the total minimum daily count is lower than 100. • Total hourly count. Flag data if the total hourly count exceeds 4,000. This includes the sum total of all count intervals and directions for the hour. • Identical counts. Flag intervals when there are more than three identical adjacent nonzero values, irrespective of the time interval that is being counted. • Year-over-year monthly average daily traffic. Flag data if variation in the monthly average daily traffic estimated for the same month in the previous year is greater than 20%. • Historical data. To evaluate the count on a given day, average the daily totals for the previous 6 weeks for a given day of the week at a given location. Flag the count if the variance is ±20% from the historical value for the same day of the week. If 6 weeks of prior data are not available, 2 weeks data at minimum must be used to run this check. In TMAS, these quality control flags are changeable, so that values reflective of travel patterns at a specific count location can be used instead of default values. These local quality control criteria would then be stored for future counts completed at the same location. There are eight aspects of data quality: accuracy, completeness, validity, timeliness, coverage, accessibility, how the data are being used, and data formats. —FHWA Traffic Monitoring Guide The TMAS is a national database maintained by FHWA for storing traffic volume data. The TMAS originally focused on collecting vehicle data from state DOTs but now includes non­ motorized data and will accept data from all agencies.

Pedestrian Volume Counting 15   Other Resources Additional guidance on quality control is available from the following sources: • National Highway Institute Traffic Monitoring Programs Course. This course outlines quality control principles for data collection, data processing, implementation, and documentation. Recommendations for ensuring quality include proper equipment training and maintenance, verifying that the equipment works properly, checking data for obvious errors, and performing detailed checks against historic data (12). • NCHRP Report 797. Section 3.3 of this report provides guidance on training staff; installing, validating, calibrating, and maintaining count equipment; and cleaning count data. Section 4.6 provides an example of checking data for anomalies (2). • “Quality Counts for Pedestrians and Bicyclists: Quality Assurance Procedures for Non- motorized Traffic Count Data.” This paper outlines three key quality assurance principles applicable to pedestrian count data: (a) quality assurance starts before data collection, (b) accept- able quality is determined by data use, and (c) measures can quantify data quality (13). Mea- sures of data quality include accuracy, validity, completeness, timeliness, coverage, and user accessibility. • Minnesota DOT Bicycle and Pedestrian Data Collection Manual. This manual outlines a three-step process for conducting quality control (14): – Visual inspection. Although it is a time-consuming process, visual inspection may help identify certain types of errors (e.g., repeating zeros) when counters are installed. – Detection of outliers. Outliers are data values that fall beyond a specified threshold (e.g., two or three standard deviations from the mean). Because this process may flag valid counts, the manual recommends conducting a web search to identify events that may have caused data spikes. – Assessment of validity of zero counts. Long strings of zero counts may occur in rural areas or in winter. Because pedestrian traffic has a seasonal nature, decision rules developed for motor vehicle traffic data cannot be applied: professional judgment is necessary. Manual Counting Checks Most available literature focuses on data quality checks for automated counts because these counts involve longer data collection periods and larger amounts of data. However, ensuring the accuracy of manual counts is also important because these counts are often used as ground truth when checking automated counts. Even so, manual counts collected at intersections with paper sheets or clickers can underestimate pedestrian volumes by 8% to 25% (15). Given that accuracy decreases after 2 hours of counting, the Traffic Monitoring Guide recommends that manual counters take periodic breaks (1), as observer inattention is another source of error with manual counts (5). Manual counts from video recordings are preferred because they make it easier to ensure accuracy. Summary Table While there are no standard procedures for automated checks of pedestrian count data, Table 2-3 summarizes current practice for performing data quality checks. Commonly encoun- tered checks include multiple days or hours with consecutive zeros, missing data, repeating counts, comparison with previous counts, and checks for outliers (5). Supplemental Data There are a number of other data sources that can be used to supplement pedestrian count data to provide more context and widen count data applications. Some sources use analytical

Source Upper Bound Data Gaps Direction Split Repeating Values Number of Consecutive Zeros FHWA TMAS Data flagged if total hourly count exceeds 4,000 Data flagged if adjacent zero hours exceed 50 Data flagged if three identical consecutive counts are encountered Data flagged if seven consecutive zeros encountered Total minimum daily count < 10 or total maximum daily count > 5,000 Colorado DOT Weekly check: Identify count sites with missing data days and flag sites with >5 days of missing data Annual check: A count is valid only if it has a full 24 hours of count data for each 24-hour period Weekly check: Flag any count site exhibiting a direction split greater than 70/30 Annual check: Same as weekly check Weekly check: During warm weather months, sites with more than two continuous days of hourly zero values flagged; this check is not applicable for cold-weather locations Annual check: Same as weekly check Weekly check: Flag counts with any daily total higher than three times the previous year’s ADT Annual check: Suspicious daily totals for each continuous count site are identified with the interquartile range formula: IQR = 2.5(Q3 − Q1) + Q3 Minnesota DOT (14) Data greater than 2 standard deviations above the mean flagged Check daily zero values in summer months North Carolina Data flagged if the upper bound exceeds IQR = Q3 + 3(Q3 − Q1) Splits greater than 3 standard deviations over average More than 3 days of zero counts BikePed Portal, Portland State University Flagged when hourly counts exceed 1,000 (low-volume sites), 2,000 (medium- volume sites), 4,000 (high-volume sites) Suspicious if ≥7 consecutive nonzero values; ≥6 consecutive nonzero values, volume >2; ≥5 consecutive nonzero values, volume > 5; ≥4 consecutive nonzero values, volume > 16; ≥3 consecutive nonzero values, volume > 100 Possibly suspicious at 12.5 hours; suspicious at 25 hours Turner and Lasley (13) IQR = 2.5(Q Q1) + Q3 Note: ADT = annual daily traffic; IQR = interquartile range; Q3 = third quartile of quarterly data; Q1 = first quartile of quarterly data. Source: Where not specifically noted otherwise, information in the table was derived from agency contacts by the research team. 3 − Table 2-3. Summary of data quality checks.

Pedestrian Volume Counting 17   methods to provide estimates of relative pedestrian activity in an area. Other sources, such as crash data, are used in combination with count data for specific applications such as safety analysis. Sampling Counts While 100% counts of pedestrians passing a location are useful in safety studies, when it comes to measuring trends, prioritizing projects, and making the case for new infrastructure, they do not provide detailed information about pedestrian activity through the network (16). Therefore, a sample of the total volume at multiple locations is sometimes measured to estimate overall mode split, route choice, and origin–destination patterns. A number of data sources can be used to develop sample counts: • Bluetooth and WiFi detection. Electronic devices such as mobile phones are equipped with Bluetooth and WiFi radios, each of which has a unique identification code that broadcasts a media access control (MAC) address. Bluetooth and WiFi readers can detect and store these MAC addresses. By using matching algorithms with data from multiple readers, travel times, travel speeds, and origin–destination matrices can be generated as the MAC addresses are detected at different points in the transportation network. Usual pedestrian detection rates from Bluetooth are between 2% and 12% (17). An emerging alternative to Bluetooth technology is WiFi, which has the advantage of a higher usage rate (16). A combined Bluetooth and WiFi system reported in the literature had an average pedestrian detection rate of 26% for the WiFi system and 2% for the Bluetooth system (17). Concerns about both technologies include low sampling rates, bias (i.e., only those people carrying mobile phones can potentially be detected), and privacy. • GPS data collection. GPS data collection has increased in popularity with the rapid proliferation of smartphones and apps (Figure 2-3). Bicyclists use apps such as Strava and MapMyRide to log their trips. However, pedestrians rarely use these apps. The data from these and other GPS- based applications have been used primarily in developing bicycle route choice models (18, 19). Other researchers have explored bicyclist speeds, travel times, and delays (20, 21). Extrapolat- ing these GPS traces to counts is challenging because of small sample sizes and bias (i.e., only those who have chosen to use the app can potentially be traced after they decide to log a trip). • Travel demand surveys. These surveys obtain information on the trip or activity patterns of a sample of households. These data are expanded to produce trip rates by mode for each analysis area. A traffic analysis zone is often used as the analysis area; however, block-size pedestrian analysis zones have also been used (22). Surveys are useful for producing aggregate trip rates; nevertheless, these estimates are generally unsuitable for estimating counts at a facility level due to small sample sizes. In addition, walk trips have traditionally been underrepresented in travel surveys (23). • Presence detection. Recent technological advances have led to the emergence of systems capable of detecting and differentiating users by mode. These systems are often camera-based and combine video images with artificial intelligence and machine learning algorithms to perform multimodal detection. These devices are typically used at signalized intersections to determine split times for motor vehicles. However, some technologies also count pedestrians and bicyclists. There has been limited evaluation of these technologies, but one study found that thermal cameras were not accurate for counting pedestrians and bicyclists in mixed traffic conditions (24). Even so, these technologies are continually evolving, and additional evalua- tion of the technologies’ ability to count pedestrians may be warranted. • Push-button actuations. Push buttons enable pedestrians to request service (i.e., a “Walk” phase) at signalized intersections (Figure 2-4). The latest traffic signal controllers are capable of recording high-resolution traffic data, including logging pedestrian push-button actuations. Some researchers have examined pedestrian actuation data and suggest that it can be used as a proxy for demand (25, 26). While push-button actuations cannot provide actual counts, they can be useful in determining the relative level of pedestrian activity at an intersection.

Source: Kittelson & Associates, Inc. Figure 2-4. Trafc signal push-button control. Figure 2-3. Screenshot of GPS data collection app.

Pedestrian Volume Counting 19   • Trip generation techniques. The Institute of Transportation Engineers’ Trip Generation Manual (27) has been used to predict the impact of multimodal trips in urban areas. However, multimodal behavior in urban contexts is not well documented, and motor vehicle impacts are often overestimated (28). NCHRP Report 770 (22) provides guidance on estimating pedes- trian trip generation. Land Use Land use characteristics affect pedestrian volumes and travel paths. For example, recreational trails are prone to different peaking characteristics (higher volumes on weekends and different weekday peak hours) than downtown sidewalks. On the basis of the surrounding land use at a continuous count site, a land use adjustment factor can be estimated to account for differences in volumes between a short-duration site and the continuous count site. Additionally, given their relatively short length, pedestrian trips on a facility are less likely to be through trips and more likely to have a nearby origin, destination, or both (2). A study used land use classifications in developing factor groups for estimating hour-to- week pedestrian count expansion factors. This approach was based on the assumption that the surrounding land use affects location-specific pedestrian activity. The land use classifica- tions in the study were central business district (CBD), school (excluding CBD locations), trail (excluding CBD and school locations), commercial (excluding trail, school, and CBD locations), and other (all remaining locations). The study found that the approach improved estimation error (29). Other researchers have used employment centers, residential areas, commercial areas, and locations near multiuse trails (30, 31); employment density (32, 33); and urbanism and climatic conditions (34) to classify pedestrian patterns. The impact of land use and urban form on travel behavior has been extensively researched. The attributes of the built environment—density, diversity, design, distance to transit, and destinations (known as the D’s)—have an impact on nonmotorized travel. Specific to pedestrian travel, researchers have found that areas with higher densities, compact pedestrian-oriented designs, and mixed use (35–37); proximity to transit (38, 39); and aesthetically pleasing walking environments (40) contribute to higher walking rates. Sociodemographic Factors Sociodemographic factors have an impact on traveler behavior. Differences in age, gender, income, vehicle ownership, education, and ethnicity have been observed with respect to walking rates. Sociodemographic data are often used in direct demand modeling, where regression models are used to model pedestrian volumes as a function of the surrounding environment and demographics. The sociodemographic factors used in these models vary on the basis of the context and application. In the United States, walking rates are higher at younger ages, and more women over 25 walk than men (22). The 2017 National Household Travel Survey (NHTS) found that walking trips were more frequently made for “social and recreational” and “shopping and errands” purposes than for other trip purposes (41). Other differences include the following (42): • Men are less likely than women to walk for recreation or to access transit but have similar walking commute rates and average trip distance. • Walking for commute purposes declines as income exceeds $30,000, while walking for recreation or exercise increases as income exceeds $30,000. • Walking rates are higher in households with lower levels of vehicle ownership and in house- holds where the number of available drivers exceeds the available vehicles. • Walking rates are higher for persons possessing higher levels of education. • Compared with other ethnicities, white and Asian populations have a higher rate of recreational trips and a lower rate of utilitarian trips.

20 Guide to Pedestrian Analysis Travel Behavior Data Travel surveys such as the NHTS (41) collect data on the personal travel patterns and char- acteristics of a sample of households and then expand results to provide higher-level trip esti- mates by mode, purpose, and other attributes. This information is generally useful for collecting mode share and origin–destination patterns, which can then be extrapolated for a larger area. However, these surveys cannot provide the fine granular data that count data can provide for a particular location or pedestrian facility. Transit Data Accessing transit through walking is one of the primary contributors to overall walking activity (Figure 2-5). According to the 2009 NHTS, 2.5% of all trips are specifically made for accessing transit, which represents 15.6% of all walk trips (22). Higher proportions of walking trips occur in areas with access to transit. Distance to the nearest transit stop has been used as a built environment measure when demand is being estimated. For example, one study used the number of regional rail transit stations and the number of bus stops within both 0.1 and 0.25 miles as predictors for estimating pedestrian crossing volumes at intersections (30). Stop-level boarding and alighting data collected by the transit agency can also provide an indication of pedestrian activity levels at a given location. Infrastructure Data A supportive pedestrian environment contributes to an increase in walking trips (28). Data on the presence of sidewalks and buffers, physical and effective sidewalk widths, pavement conditions, crosswalk dimensions, and other infrastructure can be used to estimate pedestrian demand in the form of predictor variables in a direct demand model. Additionally, these data can be used to evaluate pedestrian QOS (43). Current infrastructure data are not readily available in many locations in sufficient quality or detail. Technologies such as lidar are being increasingly used to collect infrastructure data and build inventories (44). Pedestrian Crash and Pedestrian–Vehicle Conflict Data An important application of count data is in safety analyses to determine risk or exposure. A number of measures have been proposed in the literature to quantify exposure, including Source: Kittelson & Associates, Inc. Figure 2-5. Walking to transit.

Pedestrian Volume Counting 21   volumes, sum of entering flows (motorized and nonmotorized) at an intersection, product of pedestrian and motor vehicle volume and its square root, estimated number of streets or travel lanes crossed, estimated total distance traveled, estimated total travel time, total number of trips, overall census population, and proportion of census population who report walking regularly (45). The ratio of pedestrian crashes or collisions to pedestrian volumes provides a measure of risk, which can be used to identify and prioritize locations for safety improvements. Motorized Traffic and Bicycle Volume Data Motorized traffic and bicycle volume data can be used in safety studies as predictors of pedes- trian crashes. An increase in vehicular volume has been shown to increase the probability of vehicle–pedestrian conflicts on a facility (46, 47), while another study found that the traffic volume on both the major and the minor streets was a significant predictor of pedestrian crashes at intersections (48). Traffic volumes can also be used as a measure to quantify exposure. Simi- larly, bicycle volume data may be useful in exploring pedestrian–bicycle collisions. Transportation Model Output Many regional transportation demand models either do not estimate pedestrian trips or are limited in their approach; however, a modeling effort in Portland, Oregon, demonstrated a framework for estimating pedestrian demand by incorporating sociodemographic and built environment characteristics (49). If pedestrian counts are available, they can be used in validating the outputs of travel demand models. However, regional-scale estimates from travel surveys reject partial walking trips, which is a limitation. Walk Score® (www.walkscore.com) has been used in walkability studies to evaluate the walking potential of a location on the basis of the shortest distance to a group of preselected locations, block length, and intersection density. A review of multiple studies found that walkability analysis using Walk Score was inconsistent but that Walk Score can act as a surrogate for built environ- ment density (50). Estimating Pedestrian Volume This section presents methodologies for estimating longer-term pedestrian volumes from short-duration counts and showcases best practices for volume estimation. Factors Influencing Pedestrian Volumes Many factors influence pedestrian volumes. As noted previously, pedestrian volumes are more variable than motor vehicle volumes because of a variety of factors, including weather (temperature, precipitation, humidity), other environmental conditions such as darkness, and surrounding land use (2). FHWA’s Exploring Pedestrian Counting Procedures (5) notes that “pedestrian traffic can be highly variable, especially in low traffic volume locations where hours of zero counts are common, but a track team out for a run can cause a sudden spike” (5, p. 56). Built environment factors, transportation system characteristics, socioeconomic characteristics, and pedestrian facility typology also influence pedestrian volumes (45). These factors can be used to select sites for conducting counts; additionally, if pedestrian counts are not available, they can also be used to estimate pedestrian volumes by using direct demand models, where it is important to con- sider the context-specific nature of the explanatory variables (51). Table 2-4 summarizes specific factors that significantly correlate with pedestrian activity. The spatial variability of pedestrian travel can vary substantially between adjacent cross- walks and sidewalks. This variability means that more data and greater care in creating models

22 Guide to Pedestrian Analysis Category Factor Relationship with Pedestrian Activitya Source Land use variables Nearby housing unit density + 32, 52, 53, 54 Nearby land use mix + 52, 54, 55, 56 Proximity to mixed-use buildings + 53 Proximity to multistory buildings + 53 Proximity to commercial buildings + 53, 56, 57 Presence of retail +/− 56, 58, 59 Proximity to parks + 53 Proximity to activity destinations + 32, 52 Proximity to a university + 32 Proximity to schools + 57 Cultural and entertainment space area + 56 Industrial area − 56 Proximity to vacant lots − 53 Nearby building setback distances − 55 Maximum slope − 32, 56 Distance from nearest water body − 60 Distance from CBD − 56, 60 Transit factors Access to transit stops/number of transit stops + 32, 54, 55, 56, 57, 58 Bus frequency + 56 Transportation system factors Sidewalk presence on nearby streets + 52, 53, 56 Nearby sidewalk connectivity + 53, 55 Sidewalk width + 56 Presence of bike lane + 56 Access to multiuse trails + 52 Nearby multiuse trail connectivity + 55 Nearby street network connectivity + 52, 55 Nearby intersection density + 52, 61 Nearby four-way intersections + 52 Nearby signalized intersection + 32, 56 Buffer between sidewalk and street on nearby streets + 52 Presence of street trees on nearby streets + 52 Presence of street lighting on nearby streets + 52, 53 Number of arterial roads nearby +/− 52, 56, 60 Number of collector roads nearby + 60 Nearby street block length +/− 53, 57, 61 Amount of principal arterials nearby − 56, 57, 60 Automobile speeds on nearby residential streets − 52 Automobile spaces in nearby parking area − 53 Difficulty of crossing nearby streets − 53 Table 2-4. Factors significantly correlated with pedestrian activity.

Pedestrian Volume Counting 23   Category Factor Relationship with Pedestrian Activitya Source Socioeconomic factors Nearby employment density +/− 32, 52, 53, 54, 56, 57, 58, 59, 61 Total employment + 56 Household automobile availability − 52, 55 Household income − 55, 56 Demographic factors Total population + 56 Nearby population density + 54, 56, 57, 58, 59, 61 Percentage of neighborhood residents that are nonwhite + 56, 60 Percentage of neighborhood residents with a college education + 56, 60 Student status + 55 Larger household of unrelated individuals + 55 Percentage of black residents − 56 Age − 55 Environmental factors Precipitation − 55, 56 Temperature − 56 a Plus sign represents a positive relationship (i.e., high housing density correlated with high pedestrian activity); minus sign represents an inverse relationship (i.e., large distance from CBD correlated with low pedestrian activity); plus sign/minus sign indicates the effect could be positive or negative, depending on the situation. Source: Turner et al. (56). Table 2-4. (Continued). and estimation methods are needed when estimating pedestrian volumes than when estimating motor vehicle volumes (5). Seasonal and Weather Adjustments Weather, season, time-of-day, and day-of-week are widely accepted factors that affect the expected volume of pedestrian traffic. For instance, some pedestrian facilities frequented by shoppers or recreational walkers may experience higher volumes on weekends than on weekdays, while other locations more closely associated with employment will have higher weekday pedestrian volumes. The National Bicycle and Pedestrian Documentation Project recommends counting pedestrians during weekday p.m. peak hours (5 to 7 p.m.) in mid-September (7). However, depending on the specific location and the weather at data collection time, those peak hours may not be representative of travel over the entire year. Given the known variability of pedestrian traffic due to weather and seasonal effects, adjusting pedestrian counts to account for those impacts is the best practice. To make such adjustments, data from continuous permanent counters are needed. These con- tinuous counters should be located so that the weather patterns and pedestrian travel patterns they record will be similar to the patterns expected to occur at the short-duration count stations whose counts will be adjusted with the continuous count data. As mentioned above, the Traffic Monitoring Guide recommends using three to five continuous counters per travel pattern (factor group) (1). A Washington State DOT guidebook (62) provides an example of adjusting pedestrian peak- hour counts and includes a table of pedestrian-specific adjustment factors for eastern Washington and the Puget lowland regions for the specific dates and peak hours when the Washington State DOT conducted its annual manual peak-hour counts. The adjustment factors were created from

24 Guide to Pedestrian Analysis pedestrian continuous counters in these specific regions of the state, in the same year, for a given travel pattern (in this case, noontime activity). These factors would be inappropriate to apply to the same regions in a different year, or to different regions in the same year, because weather conditions on the count day may vary substantially between regions and years. Simi- larly, if a count location has a commuter-oriented pedestrian travel pattern, with higher volumes during the morning or evening peak hours than at noon, one should not apply factors developed from continuous counters that show a peak in the middle of the day (i.e., noontime activity). Peak volumes occurring at midday are common for pedestrian traffic. NCHRP Report 797 (2) provides an example of adjusting short-term counts for weather condi- tions on the basis of continuous counter data. If a short-term count is collected on a rainy mid- week day and the continuous counter data indicate that pedestrian volumes that day were 60% of those observed on nonrainy midweek days at the same time of year, the short-term counts would be adjusted upward by 67% (1/0.6) to produce an estimate of the number of pedestrians that would have been counted at the location if the weather had not been rainy. Even when counts are adjusted with continuous count data, errors are expected to be large if only a single 2-hour count period is available. According to an extensive study of the error of peak-hour counts for estimating annual average daily nonmotorized traffic, “The estimated error based on a two-hour count alone is high. Overall there is about an 80% chance that the error will be within plus or minus 60%” (62, p. 28). Recommended count durations to minimize errors due to varying pedestrian volumes are discussed earlier in this chapter. If the count objective is to understand peak commute hours for the purposes of signal timing, it may be appropriate to count during evening hours, but practitioners should be aware that pedestrian counts may be higher at the noon hour and traffic signal timing may need to be adjusted accordingly (63). Land Use Adjustments As discussed previously, different types of land use and socioeconomic patterns can also be used to define factor groups when continuous count station locations are being identified. Doing so helps ensure that day-to-day and hour-to-hour variations observed in pedestrian volumes at the continuous count station reflect not only the weather at that time, but also the pedestrian demand patterns associated with adjacent land uses. For example, it would not be appropriate to use count data from a continuous counter located in the city’s CBD to adjust a short-term count collected in the vicinity of a school on summer break. As with weather-related factor groups, best practice is to provide three to five continuous pedestrian counters per defined land use or socioeconomic pattern (1). Estimating Annual Volume A common metric of motor vehicle traffic is annual average daily traffic (AADT). A similar metric can be computed for nonmotorized traffic. For pedestrian traffic, the Traffic Monitoring Guide (1) defines annual average daily pedestrian traffic (AADPT). For continuous count sites for which a full year of pedestrian counts is available, AADPT is simply the total volume of the full year divided by the number of days in the year. Although many jurisdictions would like to know the pedestrian traffic volume on roads and paths throughout their networks, it is infeasible to install permanent counters everywhere. Instead, short-duration counts are used to increase the geographic coverage of pedestrian count data. To minimize error when a short-term count is expanded to an annual volume, a minimum duration of 1 week is recommended for short-term counts (9, 62, 64).

Pedestrian Volume Counting 25   A standard approach to estimating AADPT recommended by the Traffic Monitoring Guide (1) is to create hourly, daily, and monthly adjustment factors for each year on the basis of data from continuous counting devices located in the same area as the short-duration counter being adjusted. Thus, if a 1-week count was made in April and the continuous counter shows that 11% of the annual pedestrian traffic occurs in April, the 1-week count data would first be expanded by (30/7) to estimate the total pedestrian volume in April, and then by (1/0.11) to estimate AADPT. Similarly, if a 2-hour count was made from 12 to 2 p.m. on a Wednesday in April, a time-of-day factor could be used to expand the 2-hour count to a daily count, a day- of-week factor could be used to convert the daily count to an average daily count, and a month- of-year factor could be used to convert the average daily count to an annual average daily count. Multiplying this result by 365 days per year gives AADPT. An alternative approach, described in Appendix D of NCHRP Report 797 (2), is to use a day- of-year expansion factor. For example, for a one-week count in April, if the continuous count data indicate the days counted represented 3% of the annual volume, the short-duration count would be adjusted by (1/0.03) to estimate AADPT. The day-of-year approach has been shown to produce approximately half the error of the standard approach when 1-day counts are adjusted. When a 1-week count is adjusted, the average error of the day-of-year approach is 12%, a level reached by the standard approach only after 2 to 3 weeks of counting (66, 64). Section 4.6 of NCHRP Report 797 (2) provides details and an example of computing temporal adjustment factors and adjusting short-duration counts. Chapter 4 of the Traffic Monitoring Guide (1) provides an example of using factors to adjust a short-duration (48-hour) count for variation by month and day for a given year in a specific region. Summary This chapter presents a number of resources to support the collection and estimation of pedes- trian volume data. Applications for pedestrian count data include monitoring facility usage, evaluating before-and-after volumes associated with pedestrian projects, monitoring traffic pat- terns, safety analysis, project prioritization, and multimodal model development. Counts can be collected at temporary, short-term sites and at permanent count stations; a counting program will ideally combine both types of counts to provide spatial and temporal coverage. The chapter describes common methods for collecting near-100% count data as well as sup- plemental data collection methods that sample pedestrian activity levels. Quality control of the data is an important consideration, regardless of how count data are collected. The chapter also described techniques for correcting systematic counting errors as well as techniques for adjusting counts made on a single day to estimate pedestrian volumes in other time periods (e.g., another day, an entire year) by adjusting for seasonal variations and weather effects. References 1. Federal Highway Administration. 2016. Traffic Monitoring Guide. Report FHWA-PL-17-003. U.S. Depart- ment of Transportation, Washington, DC. 2. Ryus, P., E. Ferguson, K. M. Laustsen, R. J. Schneider, F. R. Proulx, T. Hull, and L. Miranda-Moreno. 2014. NCHRP Report 797: Guidebook on Pedestrian and Bicycle Volume Data Collection. Transportation Research Board, Washington, DC. 3. Ryus, P., E. Ferguson, K. M. Laustsen, F. R. Proulx, R. J. Schneider, T. Hull, and L. Miranda-Moreno. 2015. NCHRP Web-Only Document 205: Methods and Technologies for Pedestrian and Bicycle Volume Data Collection. Transportation Research Board, Washington, DC. 4. Ryus, P., A. J. Butsick, F. R. Proulx, R. J. Schneider, and T. Hull. 2017. NCHRP Web-Only Document 229: Methods and Technologies for Pedestrian and Bicycle Volume Data Collection: Phase 2. Transportation Research Board, Washington, DC.

26 Guide to Pedestrian Analysis 5. Nordback, K., S. Kothuri, T. Petritsch, P. McLeod, E. Rose, and H. Twadell. 2016. Exploring Pedestrian Counting Procedures. Report FHWA-HPL-16-026. Federal Highway Administration, U.S. Department of Transporta- tion, Washington, DC. 6. Alta Planning + Design. 2010. Berkeley Pedestrian Master Plan. Final Draft. City of Berkeley, CA. 7. Alta Planning + Design. 2010. National Bicycle and Pedestrian Documentation Project: Instructions. http://bikepeddocumentation.org/application/files/3314/6671/8088/NBPD_Instructions_2010.pdf (as of April 7, 2020). 8. Nordback, K., W. Marshall, B. N. Janson, and E. Stolz. 2013. Estimating Annual Average Daily Bicyclists: Error and Accuracy. Transportation Research Record: Journal of the Transportation Research Board, No. 2339, pp. 90–97. 9. Hankey, S., G. Lindsey, and J. Marshall. 2014. Day-of-Year Scaling Factors and Design Considerations for Nonmotorized Traffic Monitoring Programs. Transportation Research Record: Journal of the Transportation Research Board, No. 2468, pp. 64–73. 10. El Esawey, M. 2014. Estimation of Annual Average Daily Bicycle Traffic with Adjustment Factors. Transpor- tation Research Record: Journal of the Transportation Research Board, No. 2443, pp. 106–114. 11. O’Brien, S., K. Jackson, S. Searcy, S. Warchol, C. Cunningham, M. Fuentes, M. Stull, D. Rodriguez, and E. Stolz. 2016. Bicycle and Pedestrian Data Collection. Phase I Final Report. NCDOT Project 2014-44. North Carolina Department of Transportation, Raleigh. 12. National Highway Institute. 2018. Traffic Monitoring Programs: Guidance and Procedures. http://www.nhi. fhwa.dot.gov/training/course_search.aspx?sf=0&course_no=151050 (as of April 7, 2020). 13. Turner, S., and P. Lasley. 2013. Quality Counts for Pedestrians and Bicyclists: Quality Assurance Proce- dures for Nonmotorized Traffic Count Data. Transportation Research Record: Journal of the Transportation Research Board, No. 2339, pp. 57–67. 14. Minge, E., C. Falero, G. Lindsey, M. Petesch, and T. Vorvick. 2017. Bicycle and Pedestrian Data Collection Manual. Minnesota Department of Transportation, St. Paul. 15. Diogenes, M. C., R. Greene-Roesel, L. S. Arnold, and D. R. Ragland. 2007. Pedestrian Counting Methods at Intersections: A Comparative Study. Transportation Research Record: Journal of the Transportation Research Board, No. 2002, pp. 26–30. 16. Lesani, A., and L. F. Miranda-Moreno. 2016. Development and Testing of a Real-Time WiFi-Bluetooth System for Pedestrian Network Monitoring and Data Extrapolation. Presented at 95th Annual Meeting of the Transportation Research Board, Washington, DC. 17. Malinovskiy, Y., N. Saunier, and Y. Wang. 2012. Analysis of Pedestrian Travel with Static Bluetooth Sensors. Transportation Research Record: Journal of the Transportation Research Board, No. 2299, pp. 137–149. 18. Broach, J., J. Dill, and J. Gliebe. 2012. Where Do Cyclists Ride? A Route Choice Model Developed with Revealed Preference GPS Data. Transportation Research Part A: Policy and Practice, Vol. 46, No. 10, pp. 1730–1740. 19. Hood, J., E. Sall, and B. Charlton. 2011. A GPS-Based Bicycle Route Choice Model for San Francisco, California. Transportation Letters, Vol. 3, No. 1, pp. 63–75. 20. Strauss, J., and L. F. Miranda-Moreno. 2017. Speed, Travel Time and Delay for Intersections and Road Segments in Montreal Network Using Cyclist Smartphone GPS Data.” Transportation Research Part D: Transport and Environment, Vol. 57, pp. 155–171. 21. Clarry, A., A. Imani, and E. Miller. 2018. Where We Ride Faster? Examining Cycling Speed Using Smartphone GPS Data. Presented at 97th Annual Meeting of the Transportation Research Board, Washington, DC. 22. Kuzmyak, J. R., J. Walters, M. Bradley, and K. M. Kockelman. 2014. NCHRP Report 770: Estimating Bicy- cling and Walking for Planning and Project Development: A Guidebook. Transportation Research Board, Washington, DC. 23. Clifton, K., and C. D. Muhs. 2012. Capturing and Representing Multimodal Trips in Travel Surveys. Review of the Practice. Transportation Research Record: Journal of the Transportation Research Board, No. 2285, pp. 74–83. 24. Kothuri, S., K. Nordback, A. Schrope, T. Phillips, and M. Figliozzi. 2017. Bicycle and Pedestrian Counts at Signalized Intersections Using Existing Infrastructure: Opportunities and Challenges. Transportation Research Record: Journal of the Transportation Research Board, No. 2644, pp. 11–18. 25. Day, C. M., H. Premachandra, and D. M. Bullock. 2011. Rate of Pedestrian Signal Phase Actuation as a Proxy Measurement of Pedestrian Demand. Presented at 91st Annual Meeting of the Transportation Research Board, Washington, DC. 26. Kothuri, S., T. Reynolds, C. M. Monsere, and P. J. V. Koonce. 2012. Preliminary Development of Methods to Automatically Gather Bicycle Counts and Pedestrian Delay at Signalized Intersections. Presented at 90th Annual Meeting of the Transportation Research Board, Washington, DC. 27. Institute of Transportation Engineers. 2017. Trip Generation Manual, 10th ed. Washington, DC. 28. Clifton, K. J., K. M. Currans, and C. D. Muhs. 2015. Adjusting ITE’s Trip Generation Handbook for Urban Context. Journal of Transport and Land Use, Vol. 8, No. 1, pp. 5–29.

Pedestrian Volume Counting 27   29. Griswold, J. B., A. Medury, R. J. Schneider, and O. Grembek. 2018. Comparison of Pedestrian Count Expan- sion Methods: Land Use Groups Versus Empirical Clusters. Transportation Research Record: Journal of the Transportation Research Board, No. 2672, pp. 87–97. 30. Schneider, R. J., L. S. Arnold, and D. R. Ragland. 2009. Methodology for Counting Pedestrians at Inter- sections: Use of Automated Counters to Extrapolate Weekly Volumes from Short Manual Counts. Transpor- tation Research Record: Journal of the Transportation Research Board, No. 2140, pp. 1–12. 31. Hocherman, I., A. S. Hakkert, and J. Bar-Ziv. 1988. Estimating the Daily Volume of Crossing Pedestrians from Short-Counts. Transportation Research Record: Journal of the Transportation Research Board, No. 1168, pp. 31–38. 32. Schneider, R. J., T. Henry, M. F. Mitman, L. Stonehill, and J. Koehler. 2012. Development and Application of the San Francisco Pedestrian Intersection Volume Model. Transportation Research Record: Journal of the Transportation Research Board, No. 2299, pp. 65–78. 33. Ryan, S., D. E. Sidelinger, S. Saitowitz, D. Browner, S. Vance, and L. McDermid. 2014. Designing and Implementing a Regional Active Transportation Monitoring Program through a County–MPO–University Collaboration. American Journal of Health Promotion, Vol. 28, pp. S104–S111. 34. Nordback, K., and M. Sellinger. 2014. Methods for Estimating Bicycling and Walking in Washington State. Report WA-RD 828.1. Washington State Department of Transportation, Olympia. 35. Lawrence Frank & Co., Inc.; Sacramento Council of Governments; and Mark Bradley Associates. 2009. I-PLACE3S Health and Climate Enhancements and Their Application in King County. King County, Seattle, WA. 36. Kockelman, K. M. 1996. Travel Behavior as a Function of Accessibility, Land Use Mixing, and Land Use Balance: Evidence from the San Francisco Bay Area. Thesis. University of California, Berkeley. 37. Kuzmyak, J. R., Fregonese Associates, and Fehr & Peers. 2010. Local Sustainability Planning Model Develop- ment. Final Report. Southern California Association of Governments, Los Angeles. 38. Parsons Brinckerhoff Quade & Douglas, Inc.; R. Cervero; Howard/Stein-Hudson Associates, Inc.; and J. Zupan. 1996. Mode of Access and Catchment Areas of Rail Transit. TCRP Project H-1. Unpublished research findings available from the Transit Cooperative Research Board, Transportation Research Board, Washington, DC. 39. Cambridge Systematics, Inc.; Parsons Brinckerhoff; Mark Bradley Research & Consulting; CCS Planning & Engineering, Inc.; Hausrath Economics Group; Hunt Analytics, Inc.; Lawton Consulting; and Corey, Canapary & Galanis. 2002. San Francisco Travel Demand Forecasting Model Development. Final Report. San Francisco County Transportation Authority, San Francisco, CA. 40. Cambridge Systematics, Inc.; and Deakin, Harvey, Skabardonis, Inc. 1995. The Effects of Land Use and Travel Demand Management Strategies on Commuting Behavior. Report DOT-T-95-06. U.S. Department of Trans- portation, Washington, DC. 41. McGuckin, N., and A. Fucci. 2018. Summary of Travel Trends: 2017 National Household Travel Survey. Report FHWA-PL-18-019. Federal Highway Administration, U.S. Department of Transportation, Washington, DC. 42. Agrawal, A., and P. Schimek. 2007. “Extent and Correlates of Walking in the USA.” Transportation Research Part D: Transport and Environment, Vol. 12, No. 8, pp. 548–563. 43. Transportation Research Board. 2016. Highway Capacity Manual: A Guide for Multimodal Mobility Analysis, 6th ed. Washington, DC. 44. Ai, C., and Q. Hou. 2019. Improving Pedestrian Infrastructure Inventory in Massachusetts Using Mobile LiDAR. Report No. 19-007. Massachusetts Department of Transportation, Boston. 45. Turner, S., I. Sener, M. Martin, S. Das, E. Shipp, R. Hampshire, K. Fitzpatrick, L. Molnar, R. Wijesundera, M. Colety, and S. Robinson. 2017. Synthesis of Methods for Estimating Pedestrian and Bicyclist Exposure to Risk at Areawide Levels and on Specific Transportation Facilities. Report FHWA-SA-17-041. Federal Highway Administration, U.S. Department of Transportation, Washington, DC. 46. Schneider, R. J., R. M. Ryznar, and A. J. Khattak. 2004. An Accident Waiting to Happen: A Spatial Approach to Proactive Pedestrian Planning. Accident Analysis & Prevention, Vol. 36, No. 2, pp. 193–211. 47. Zegeer, C. V., D. L. Carter, W. W. Hunter, J. R. Stewart, H. F. Huang, A. H. Do, and L. S. Sandt. 2006. Index for Assessing Pedestrian Safety at Intersections. Transportation Research Record: Journal of the Transportation Research Board, No. 1982, pp. 76–83. 48. Torbic, D. J., D. W. Harwood, C. D. Bokenkroger, R. Srinivasan, D. Carter, C. V. Zegeer, and C. Lyon. 2010. Pedestrian Safety Prediction Methodology for Urban Signalized Intersections. Transportation Research Record: Journal of the Transportation Research Board, No. 2198, pp. 65–74. 49. Clifton, K. J., P. A. Singleton, C. D. Muhs, and R. J. Schneider. 2015. Development of a Pedestrian Demand Estimation Tool. Report NITC-RR-677. National Institute for Transportation and Communities, Portland, OR. 50. Hall, M., and Y. Ram. 2018. Walk Score® and Its Potential Contribution to the Study of Active Transport and Walkability: A Critical and Systematic Review. Transportation Research Part D: Transport and Environment, Vol. 61, pp. 310–324.

28 Guide to Pedestrian Analysis 51. Munira, S., and I. Sener. 2017. Use of Direct-Demand Modeling in Estimating Nonmotorized Activity: A Meta- Analysis. Report UTC Safe-D 01-003. Texas A&M Transportation Institute, College Station. 52. Handy, S. 2005. Critical Assessment of the Literature on the Relationships Among Transportation, Land Use, and Physical Activity. Background paper for Special Report 282: Does the Built Environment Influence Physical Activity? Examining the Evidence. Transportation Research Board, Washington, DC. https://itspubs. ucdavis.edu/wp-content/themes/ucdavis/pubs/download_pdf.php?id=2536 (as of April 7, 2020). 53. Ewing, R., and R. Cervero. 2001. Travel and the Built Environment: A Synthesis. Transportation Research Record: Journal of the Transportation Research Board, No. 1780, pp. 87–114. 54. Pulugurtha, S. S., and S. R. Repaka. 2008. Assessment of Models to Measure Pedestrian Activity at Signal- ized Intersections. Transportation Research Record: Journal of the Transportation Research Board, No. 2073, pp. 39–48. 55. Shriver, K. 1997. Influence of Environmental Design on Pedestrian Travel Behavior in Four Austin Neighborhoods. Transportation Research Record, No. 1578, pp. 64–75. 56. Turner, S., I. Sener, M. Martin, L. D. White, S. Das, R. Hampshire, M. Colety, K. Fitzpatrick, and R. Wijesundera. 2018. Guide for Scalable Risk Assessment Methods for Pedestrians and Cyclists. Report FHWA-SA-18-032. Federal Highway Administration, U.S. Department of Transportation, Washington, DC. 57. Miranda-Moreno, L. F., P. Morency, and A. M. El-Geneidy. 2011. The Link Between Built Environment, Pedestrian Activity and Pedestrian–Vehicle Collision Occurrence at Signalized Intersections. Accident Analysis & Prevention, Vol. 43, No. 5, pp. 1624–1634. 58. Schneider, R. J., L. S. Arnold, and D. R. Ragland. 2009. Pilot Model for Estimating Pedestrian Intersection Crossing Volumes. Transportation Research Record: Journal of the Transportation Research Board, No. 2140, pp. 13–26. 59. Jones, M. G., S. Ryan, J. Donlon, L. Ledbetter, D. R. Ragland, and L. Arnold. 2010. Seamless Travel: Measuring Bicycle and Pedestrian Activity in San Diego County and Its Relationship to Land Use, Transportation, Safety, and Facility Type. Report UCB-ITS-PRR-2010-12. California PATH Program, University of California, Berkeley. 60. Hankey, S., G. Lindsey, X. Wang, J. Borah, K. Hoff, B. Utecht, and Z. Xu. 2012. Estimating Use of Non- Motorized Infrastructure: Models of Bicycle and Pedestrian Traffic in Minneapolis, MN.” Landscape and Urban Planning, Vol. 107, No. 3, pp. 307–316. 61. Krizek, K. 2003. Operationalizing Neighborhood Accessibility for Land Use-Travel Behavior Research and Regional Modeling. Journal of Planning Education and Research, Vol. 22, pp. 270–287. 62. Johnstone, D., K. Nordback, and M. Lowry. 2017. Collecting Network-Wide Bicycle and Pedestrian Data: A Guidebook for When and Where to Count. Washington State Department of Transportation, Olympia. 63. Kothuri, S., A. Kading, E. Smaglik, and C. Sobie. 2017. Improving Walkability Through Control Strategies at Signalized Intersections. Report NITC-RR-782. Transportation Research and Education Center, Portland State University, Portland, OR. 64. Nosal, T., L. F. Miranda-Moreno, and Z. Krstulic. 2014. Incorporating Weather: A Comparative Analysis of Average Annual Daily Bicyclist Estimation Methods. Transportation Research Record: Journal of the Trans- portation Research Board, No. 2468, pp. 100–110.

Next: Chapter 3 - Pedestrian Safety Analysis »
Guide to Pedestrian Analysis Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Roadway designs and signal phasing that address the safety of all road users are being implemented in many cities around the country. As part of this, accurate methods for estimating pedestrian volumes are needed to quantify exposure and, in turn, evaluate the benefits of pedestrian safety measures.

The TRB National Cooperative Highway Research Program's NCHRP Research Report 992: Guide to Pedestrian Analysis presents a state-of-the-art guide to conducting pedestrian traffic analysis on the basis of volume, safety, operations, and quality of service. In addition to the guide, the research provides new evaluation methods for use with the Highway Capacity Manual.

Supplemental to the report is NCHRP Web-Only Document 312: Enhancing Pedestrian Volume Estimation and Developing HCM Pedestrian Methodologies for Safe and Sustainable Communities; two computational engines for implementing the new and updated analysis methods developed by the project: Signalized Crossing Pedestrian Delay Computational Engine and Uncontrolled Crossing Pedestrian Delay and LOS Computational Engine; a Video; five presentations from a peer exchange workshop: Project Overview, Pedestrian Volume Counting, Pedestrian Operations Analysis, Pedestrian Quality of Service Analysis, Pedestrian Safety Analysis, and an Implementation Plan.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!