Trip Generation Rates for Transportation Impact Analyses of Infill Developments (2013)

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.

SUPPLEMENTAL TECHNICAL REPORT

3 5 Chapter 1 Introduction 6 Chapter 2 Overview of the Trip Generation Methodology for Infill Development 8 Chapter 3 Application of the Household Travel Survey Method 8 3.1 Background and Source of Surveys 9 3.2 Practitioner Need for Broad Applicability 9 3.3 Required Data for the Household Travel Survey Method 10 3.4 Linked-Trip Data 10 3.5 Geographic Units of Urban Area Data 12 Chapter 4 Example Adjustment Factors Using San Francisco Bay Area Travel Survey Data 12 4.1 Household Travel Survey Data 12 4.2 Defining General Urban and Urban Center Context 14 4.3 Weighting and Expansion of Survey Data 15 4.4 Estimated Mode Share by Land Use 15 4.5 Estimated Vehicle Occupancy by Land Use 16 4.6 Use of Local Travel Demand Model to Derive Adjustment Factors 17 Chapter 5 Selection of a Household Travel Survey as a Case Study 17 5.1 Criteria for Selecting a Metropolitan Area with a Suitable Household Travel Survey 17 5.2 Selection of a Metropolitan Area 18 5.3 Sufficiency of the Dataset 18 5.4 Next Steps in the Process 19 Chapter 6 Analysis of Household Travel Survey Data 19 6.1 GIS Analysis 19 6.2 Household Travel Survey Analysis 22 Chapter 7 Selection of Candidate Sites for Cordon Counts 22 7.1 Selecting Urban Infill Sites for Cordon Counts 23 7.2 Summary of the Data Collection Procedures 24 Appendix A Overview of Household Travel Surveys Assessed for Case Study 25 Appendix B Detailed Mode Share Tables for San Francisco Bay Area Example C o n t e n t s

4 27 Appendix C Example Output Tables for San Francisco Bay Area Infill Area Mode Share and Vehicle Occupancy Adjustment Factors 30 Appendix D Detailed Mode Share Tables for the Washington, D.C., Case Study 32 Appendix E Output Tables for Washington, D.C., Infill Area Mode Share and Vehicle Occupancy Adjustment Factors 35 Appendix F Prioritization of Candidate Sites for Cordon Counts 37 Appendix G Example Data Summaries for Candidate Sites 45 Notes and Citations to Supplemental Technical Report

5 This technical report supplements NCHRP Report 758: Trip Generation Rates for Transportation Impact Analyses of Infill Developments. It is a consolidation of interim reports prepared during the development of the recommended methodology for estimating vehicular trip generation of infill development. Specifically, this report describes the procedure the research team used to extract data from HTSs and the use of the data to derive factors used in the methodology. The objective of the research presented in NCHRP Report 758 is to develop an easily applied methodology to estimate automobile trip generation and mode shares of non-vehicular trips that can be used in the preparation of site-specific transportation impact analyses of infill devel- opment projects located within existing higher-density built-up areas. C H A P T E R 1 Introduction

6The following overview of the methodology is paraphrased from Chapter 3 of the report. The research team selected an approach for estimating the trip generation of infill development categorized as “ITE rate adjustment based on empirical data,” as described in the main body of the report. This approach met the research objective and, to varying degrees, all of the selection criteria. One of the predominant reasons this approach was selected is because it can be applied to the land uses in the ITE Trip Generation Manual1 and has few, if any, restrictions on land use catego- ries and applicable geography. The approach of employing empirical data provides the practitioner with flexibility in that there are no limitations or constraints in regards to land use classification or geography. Conceptually, the approach can be described with the following simplistic equation: = ×Auto_Trips Auto_Trips Adjustment_Factor(INFILL) (ITE) (INFILL) The recommended methodology applies adjustment fac- tors to data from the ITE Trip Generation Manual, resulting in a relatively straightforward conversion of data representa- tive of isolated automobile-dominated suburban land uses to data representative of dense urban areas served by extensive multimodal transportation systems. The selection process and subsequent development of the approach resulted in two ways to develop the adjustment factors employed by the approach: 1. Proxy site method – Adjustment factors are based on data collected at sites with similar characteristics and located in similar contexts as the proposed infill development site (the project being studied). The research team developed procedures for identifying proxy sites and obtaining the required data to develop the adjustment factors applied in the methodology. 2. Household travel survey method – Adjustment factors derived from empirical data found in the database of a regional HTS. This method extracts data representing the desired infill land use and context within physical areas at the scale of the TAZ. Extraction of data representing spe- cific land uses is based on the activities and trip purposes recorded by the travelers during the survey. As shown in Figure 2.1, the approach is made up of five primary steps: 1. Baseline ITE trip generation data are used to estimate the vehicular trip generation of the proposed infill development. 2. Baseline mode share and vehicle occupancy adjustment factors are used to convert baseline vehicle-trip estimates to baseline person trips. 3. An infill mode share adjustment factor representing the appropriate context is used to convert baseline person trips to infill person trips for those who travel by automobile. 4. An infill vehicle occupancy adjustment factor representing the appropriate context is used to convert infill person trips for those who travel by automobile to infill vehicle trips. 5. Infill vehicle trips are used in the evaluation of site traffic impacts. This supplemental technical report describes the proce- dures used to develop the infill mode share and vehicle occu- pancy adjustment factors using the household travel survey method described in the main report. C H A P T E R 2 Overview of the Trip Generation Methodology for Infill Development

7 Note: TIAs = transportation impact analyses. Figure 2.1. Approach for estimating vehicle trip generation.

8The following description of the use of the household travel survey method is paraphrased from Chapter 3 and 4 of the report. Infill adjustment factors may be derived for sites pro- posed within metropolitan areas that have current HTS data. This method of deriving mode share and auto occupancy is limited to the land use categories that can be deduced from HTS linked-trip data—essentially only the general categories (e.g., retail, office, multifamily housing, etc.) because the data from the surveys do not always distinguish between land use subcategories (i.e., grocery store versus home improvement center). However, HTS data can provide adjustment factors for all context types and, more importantly, can identify dif- ferences in the adjustment factors within each context type due to geographic location and varying socio-demographic characteristics within a region. This method will result in adjustment factors for general land use categories within any context type either (a) averaged across the metropolitan region, or (b) specific to any TAZ located in the region. Although this method can be used to generate the adjustment factors used in traffic impact analyses of infill development, it can also be used for broader types of analyses, including: • Creation of a region- or area-wide database of mode share and vehicle occupancies by TAZ (representing the context within the TAZ) for adjusting ITE trip generation rates for sites in different locations in the region to ensure consis- tency in infill development traffic impact studies within the region or area. • Scenario analyses comparing the transportation benefits or impacts of shifting growth in development between urban infill and suburban or greenfield locations. • Studies of large-scale activity centers requiring an under- standing of how the center’s mode share is influenced by its location within the region, proximity to transit, and other built environment characteristics. • Development of local or regional trip generation rates and mode shares covering a range of contexts for inclusion in agency traffic impact analyses guidelines. 3.1 Background and Source of Surveys Household travel surveys provide information at the regional level on the relationships between the characteris- tics of personal travel and the demographics of the traveler. They are used to identify travel patterns and to provide the necessary data to support the development, calibration, and validation of regional travel forecasting models. Analysis of the data at different scales can be used for more detailed sub- regional studies and research. Surveys include demographic characteristics of households, people, and vehicles, as well as detailed information on the daily activities of individuals in a household and their mode of travel for all purposes. Data are collected from a sample of households in the study area and expanded to provide regional estimates of trips and miles by travel mode, trip purpose, and other household characteristics. Household travel surveys are typically conducted by MPOs—agencies responsible for maintaining regional travel data as well as developing and maintaining travel demand fore- casting models. Household travel surveys have evolved over the past 50 years, but the state of the practice today is the travel-/ activity-based survey, which is considered the best source of household and person-trip generation data by mode for the regions that have conducted these surveys. This type of sur- vey focuses on each activity throughout the day and records trip details to and from each activity. (Data are gathered on the basis of trip-end activities.) The methodology proposed in this report is based on the use of travel-/activity-based household travel surveys. There are two national-scale travel surveys: the National Household Travel Survey (NHTS) and the Nationwide Personal Transportation Survey (NPTS). While these national-scale C H A P T E R 3 Application of the Household Travel Survey Method

9 surveys provide some useful information, they are not repre- sentative at the scale of the region, city, or urbanized area. In fact, according to the Travel Survey Manual:2 “It has long been determined by most metropolitan regions that data collected in one region has little relevance to another region. While there is no doubt that there will be local contextual issues that may make transfer of data difficult or in appropriate at times, the major reason for this perception is that each house- hold travel survey is usually sufficiently different in design and execution from any other survey, resulting in comparisons from region to region that are completely obscured by method- ological and implementation differences.” This statement, referring to a lack of survey standardiza- tion, in part informs the recommendations of this proposed methodology. Many of the metropolitan areas of the United States have conducted relatively recent household travel sur- veys, most coinciding with the 2000 census. Some regions, such as the San Francisco Bay Area, conduct surveys every 10 years. The availability of recent surveys (no older than year 2000) is an important factor in the recommendation and proposed organization of the methodology. Appendix A of this technical report contains summary information on the household travel surveys that were reviewed as part of this study. 3.2 Practitioner Need for Broad Applicability Maximizing the value of the proposed methodology to practitioners who prepare urban traffic impact studies requires that the method be applicable over a wide range of metropolitan areas. A method based on data aggregated at the national scale (weighted average of adjustment factors from multiple metropolitan areas) is simple to use and can be applied by the practitioner regardless of its location in the United States. This method is similar to the use of trip gen- eration rates published by ITE. The simplicity of using ITE trip generation rates is one of the reasons for its popularity and widespread use. However, data aggregated at the national scale cannot credibly reflect the unique characteristics of individual urban areas. In contrast to adjustment factors derived from national aver- ages are factors derived from local data from the practitioner’s study area. Local scales may range from a relatively compact dis- trict surrounding a transit station (such as a handful of TAZs) to the entire metropolitan area composed of multiple cities and including the study area. The benefit of adjustment factors derived from local survey data is a better representation of the actual conditions the practitioner is studying. The downside is a smaller set of trip records and loss of precision inherent in smaller samples. In recommending a scale of data aggregation, the investi- gators considered the needs of the practitioner, the precision of the survey results (i.e., number of trip records), applica- bility over a broad range of urban areas, ease of use by the practitioner, availability of data, and the effort and budget required to prepare the factors so that the methodology is immediately usable by practitioners. There is an inherent trade-off between the higher precision gained from a large number of trip records aggregated from multiple metropoli- tan areas and the lower precision of a smaller amount of data representing the practitioner’s actual study area or, at least, metropolitan area. 3.3 Required Data for the Household Travel Survey Method The data commonly available from a HTS to use the household travel survey method can be divided into four categories: 1. Household data – Characteristics of the household and its location. 2. Person data – Demographic, socioeconomic, and employ- ment information for one or more members of the household. 3. Vehicle data – Type, ownership, and usage of private vehi- cles available to household members. 4. Travel and activity data – Detailed travel, activities, and origins/destinations of the daily trips by one or more household members. The minimum required data for estimating infill trip gen- eration adjustments are listed in Table 3.1, which describes the required variables and how they are used in deriving the adjustment factors. These variables are generally standard in travel surveys and should be available, in one form or another, from all the major metropolitan areas. The linked-trip data contain many other variables and helpful information for cross-referencing household, person, vehicle, and activity data. Typically, a household travel survey records individual segments of each trip separately every time the traveler stops for a specific purpose on the way to an ultimate destination, including when changing modes. For example, driving from home to the train station, taking the train, and walking to the workplace are three segments of a single trip, each using a dif- ferent mode of travel. These are called “unlinked trips.” How- ever, the travel described here is actually one home-based work trip using rail transit as the primary mode of travel. Driving to the train station and walking to the workplace are secondary. The consolidation to a single trip purpose by a single mode describes a linked trip. Household travel survey records of unlinked trips are manip- ulated to produce linked trips. The linked-trip data contain

10 multiple variables from the four categories of data: household, person, vehicle, and travel/activity. Linked-trip data are made up of individual trip records, each of which represents one person’s travel for an activity by the primary mode of travel. Each trip record is identified by a general trip purpose [e.g., home-based work (HBW), home-based shopping (HBS), non–home-based (NHB), start and end time of travel, mode of travel, passengers (if by auto), mode of access to primary travel mode, origin and destination activities and place, and numerous other data]. Based on a review of the data, the research team determined that the linked-trip data contained the appropriate information for deriving mode split and vehicle occupancy for various land uses and time periods. Linked-trip data were selected as the best source of data because their trip records were cross-referenced to the variables needed to calculate the adjustment factors. 3.4 Linked-Trip Data By their nature, travel/activity surveys report individual segments of each trip separately every time the traveler stops for a specific purpose on the way to an ultimate destination, including when changing modes. Each segment is an inter- mediate part (with an intermediate mode) of an entire trip sequence making up one trip from the origin to the destina- tion for a primary purpose (e.g., home-based work trip). For example, driving from home to the train station, taking the train, and walking to the workplace are three segments of a single trip with a home-based work trip purpose and with rail transit as the primary mode of travel and walking as the mode of access to/from transit. The access to transit mode is secondary and ignored for most purposes. MTC’s 2000 Bay Area Travel Survey (2000 BATS), which was used as the primary source of data to develop and evaluate the proposed methodology, includes records of unlinked trips that can be aggregated to produce linked trips. The linked-trip data contain multiple variables from the four categories of data; household, person, vehicle, and travel/activity. Linked- trip data from the 2000 BATS were made up of individual trip records, each of which represented one person’s travel for an activity by the primary mode of travel. Each trip record is iden- tified by a general trip purpose (e.g., HBW, HBS, NHB), start and end time of travel, mode of travel, passengers (if by auto), mode of access to primary travel mode, origin and destination activities and place, and numerous other data. During the initial assessment of the 2000 BATS data, the investigators determined that the linked-trip data contained the appropriate information for deriving mode split and vehi- cle occupancy for the land uses and time periods required for the study. 3.5 Geographic Units of Urban Area Data The unit of geography used for household travel data is typ- ically some form of zone system. Most regions are divided into multiple zone systems of varying scales, including the stan- dard census geographic units of census tracts, block groups, and blocks. In the 2000 BATS, the following geographic units (number of units is included in the parenthesis) were determined to be available: • Census block (76,250), • Census block group, • Census tract (1,405), • Census transportation planning product TAZs (4,070), • PUMA – public-use microdata area (54), • Super-PUMA districts (9), HTS Variable Definition Origin activity/destination activity or origin land use/destination land use Provides the activity purpose or land use of the origin and the destination of the trip. Used to associate the trip with a particular land use. General purpose Provides home-based and non–home- based trip information. Used to cross- check data and to populate adjustment factors when using travel demand forecasting model data. Primary mode of travel Provides the primary mode of travel for individual trip records (ignoring mode of access). Used to develop adjustment factor mode split. Origin TAZ/destination TAZ Provides the zone of the origin and the destination of the trip. First used to identify trip records within TAZs designated as general urban/urban center, then used to classify the trip as inbound or outbound in the extraction of peak hour records. If available, an address can be used to determine the origin or destination zone. Day of trip Identifies the day the trip occurred. Used to classify trips as weekday or weekend. Start time/end time Starting and ending time of the trip. Used to classify trips in either a.m. or p.m. peak period. Number in vehicle Provides the number of people in a vehicle. Used to determine vehicle occupancy. Table 3.1. Linked-trip data variables in deriving adjustment factors.

11 • Super-districts (34), and • MTC TAZs (1,454). To develop and evaluate the proposed methodology, MTC’s 1,454 zone TAZ system was selected as the geographic unit for the following reasons: • TAZs are the smallest scale of district in urban areas, with the exception of census block group or blocks, providing a reasonable resolution for focusing on trip records meeting transit proximity criteria. • Trip origins and destinations are identified by TAZ. • Census data, as summarized by MTC, corresponds to TAZs. • Existing and future land use and demographic information (for use in travel demand models) is aggregated by TAZ and useful in the proposed methodology for evaluating future site conditions.

12 This chapter provides a detailed example of how mode split and vehicle occupancy adjustment factors are derived from household travel survey data. As described in the previous chapter, the investigators selected the household travel sur- vey data from the 2000 BATS. This activity-based survey was conducted with over 15,000 households in the nine-county Bay Area with a year-2000 population of 6,800,000 in nearly 2,500,000 households. The data are readily available, well- documented, and summarized by MTC for cross-checking against the resulting adjustment factors. Wherever possible, the investigators identify the limita- tions of the data or the process. This example is based on using data aggregated at the TAZ level for contexts classified as general urban and urban center within the Bay Area. Addi- tional discussion on the classification of context zones is pro- vided in the main body of the report. 4.1 Household Travel Survey Data MTC’s household travel survey data are contained in four primary datasets: person, household, vehicle, and travel/ activity. Additionally, MTC releases unlinked and linked- trip data. Linked-trip data were selected as the best source of data because their trip records were cross-referenced to the variables needed to calculate the adjustment factors. The linked-trip dataset contains many other variables and help- ful information for cross-reference to the household, person, vehicle, and activity datasets. 4.2 Defining General Urban and Urban Center Context A comprehensive literature review of quantifying urban area designations is included in research conducted in Califor- nia (Trip-Generation Rates for Urban Infill Land Uses in Cali- fornia, Phase 2: Data Collection, Final Report3). The California research identified a definition of “urban” based on the density of residential dwelling units, employment, or a combination of both. The investigators used criteria similar to the Califor- nia research to designate TAZs in the San Francisco Bay Area as general urban/urban center. As shown in the following, to set the upper limit of general urban/urban center density, the team used the lower limit of density in known urban cores. Lower limit of urban core density = upper limit of general urban/urban center density: Employees/gross acres greater than 70 and/or households/gross acres greater than 40 In this example, the team used the known urban cores of San Francisco’s and Oakland’s central business districts. Year- 2000 census data were used to determine residential density and MTC’s year-2000 employment database (used in travel demand forecasting) was used to determine employment density. The lower limit of general urban/urban center was set as the upper limit of the Bay Area’s suburban density. Suburban densities range greatly, so the investigators considered MTC’s and Florida Department of Transportation’s quantitative criteria for suburban areas and the criteria documented in Trip-Generation Rates for Urban Infill Land Uses in California, Phase 2: Data Collection, Final Report to establish the lower limit of general urban/urban center. The following summa- rizes these thresholds: Upper limit of suburban density = lower limit of general urban/urban center density: Employees/gross acres less than 35 and/or households/gross acres less than 10 These thresholds, individually and in combination, were used to isolate the Bay Area’s TAZs that meet either or both criteria. This resulted in a relatively small number of TAZs C H A P T E R 4 Example Adjustment Factors Using San Francisco Bay Area Travel Survey Data

13 (176) out of the 1,454 in the Bay Area. About 32,800 individ- ual trip records have either an origin or destination in these 176 TAZs. These zones were the initial pool for a series of steps that incrementally reduced the pool and the eventual size of the sample trip records. The sequence of steps of que- rying and isolating trip records is: 1. Use trip purpose at origin and destination to isolate trips by the four proposed study land uses (residential, restaurant, retail, and office); 2. Identify TAZs within ½-mile walk of rail/ferry or ¼ mile of high-frequency bus routes, and isolate trip records within these zones; 3. Isolate weekday trip records from weekend trip records; 4. Isolate a.m. and p.m. peak periods from daily trip records; and 5. Isolate inbound and outbound trip records. The sequence of the key steps is briefly described in the following sections. Detailed discussion on isolation and a summary of data by key variables are not included. 4.2.1 Variables Representing Study Land Use Categories The four land use categories proposed for initial develop- ment of adjustment are residential, restaurant, retail, and office. Because the available data are activity-based and not place-based, the type of land use at the origin or destination of the trip needs to be inferred from the trip records. The linked-trip data contain a variable for the activity purpose at both ends of the linked trip. While not in great detail, the activity variable has 17 activities the participant can choose from. From these the investigators selected those activities that best represent an activity at a specific land use. Clearly missing from household travel survey data are trips made by delivery or service people, with the exception of survey par- ticipants who are employed in these fields. This is a potential source of error in estimating actual trips but not a signifi- cant error in calculating mode split or vehicle occupancy. The activities used to determine land uses associated with trips are discussed in the following paragraphs. Residential-related trips were selected from the trip records with an origin purpose or a destination purpose classified as “home.” It appears that the majority of the trip records are from the residents themselves and do not capture non–resident- related trips. Restaurant-related trips were selected from the trip records with an origin purpose or a destination purpose classified as “meals.” The “meal” activity encompasses “at home, take-out, restaurant, coffee, and snack.” Based on this definition, it is not possible to eliminate the meals eaten at the home that do not generate an external trip from the home. The trip records include both employee and patron trips. Retail or shopping-center–related trips were selected from the trip records with an origin purpose or a destination pur- pose classified as “shopping away from the home” or “per- sonal services/bank/government.” The “personal services/ bank/government” purpose includes barber, beauty shop, dry cleaning, banking, and government services. The trip records include only patron trips. Office-building–related trips were selected from the trip records with an origin purpose or a destination purpose clas- sified as “work or work related.” These trip records include all work trips and do not classify the origin or destination as office. Under the MTC variables, there is no reasonable way to separate work trips from those specific as to an office building except by reviewing the participant’s comments. However, on review, this was not deemed to be an impractical way to isolate office building related trips. While this is a potential source of error and inaccuracy, it was used because the resulting mode split and vehicle occupancy represent work-related trips for all land uses, including office buildings. Further, a comparison of the resulting mode splits and vehicle occupancies does not look unreasonable compared to generalized work mode splits for the entire region or within ½ mile of rail stations as published by MTC. Therefore, while use of the “work or work-related” trip purpose captures trip records to non-office locations, the data appear to reasonably represent work trips to office land uses. The inability to distinguish between different workplaces for work-related trips may not be a universal problem with household travel survey data. An examination of the 2007 Chicago area household travel survey includes a variable iden- tifying the origin or destination as a restaurant but does not specifically include office buildings. This issue will require fur- ther exploration to determine its extent and whether the issue results in significant error. 4.2.2 Transit Proximity For the 2000 BATS data set, only the general urban and urban center land uses within ½ mile of a rail (or ferry) station and within ¼ mile of a high-frequency (maximum 15-min headways for 6 or more hours of the day) bus route were con- sidered infill development. Mode split and vehicle occupancy adjustment factors are, therefore, determined from TAZs that meet these proximity criteria. For the San Francisco Bay Area example, GIS was used to map rail and ferry stations and to identify a ½-mile buffer around the station. A similar analysis was prepared for high-frequency bus lines. 4.2.3 Determining TAZ Proximity to Transit A TAZ was considered to be in proximity to the transit facility if the TAZ was entirely within the buffer ring or if the

14 edge of the buffer ring covered more the 1⁄3 of the TAZs area. In metropolitan areas where GIS layers of transit routes and stops are not available, the exercise of selecting transit proxi- mate TAZs is performed manually. This exercise demonstrated that it is impractical to use very close proximities to transit (i.e., 400–600 ft) if the typical size of a TAZ is greater than ¼ mile. Although the resolution of transit proximity could be improved using census blocks (or geo-coded origins and destinations), the significant reduction in trip records (or the likelihood of blocks without any trip records) may offset the benefit of increased resolution. When acquiring transit system data, it is important that the transit route, stop, and frequency information match the year of the household survey data or that it be demonstrated that the transit system has not under- gone change since the survey was conducted. 4.3 Weighting and Expansion of Survey Data Household survey data are typically adjusted through a process of weighting and expansion. According to the MTC:4 Sample weighting is a technical necessity to account and cor- rect for geographic and demographic biases in a survey. Sample expansion, on the other hand, is the process used to factor up survey records to represent aggregate demographic and travel characteristics. The weighting factors used in this analysis [the 2000 BATS] are essentially combined weighting and expansion factors. The objective of applying weighting factors to samples is to draw valid conclusions about the entire study popula- tion based on the survey results of a relatively small sample. Weighting is determined by comparing the sample variables to the actual values from a known credible source like the census. 4.3.1 Weighting and Expansion Method Used in San Francisco Bay Area Example The 2000 BATS data were weighted and expanded based on Census 2000 data. Weighting and expansion factors for each trip record are based on the PUMA of household character- istics, including household size, vehicles available, tenure, and race/ethnicity. The system of PUMAs in the San Francisco Bay Area makes up 54 districts encompassing the region’s 1,454 TAZs. For this proposed methodology, weighting and expansion factors may be applied to trip records in this method so that the resulting mode splits and vehicle occupancies are more representative of the areas where the data were collected. This can be particularly important in study areas that are rich in the types of household demographics that influence transit use. 4.3.2 Creating New Weighting and Expansion Factors Since the trip records used to derive mode split and vehicle occupancy in the example used in this methodology are from a subset of the San Francisco Bay Area (e.g., general urban and urban center in proximity to rail stations and high-frequency bus routes), the weighting and expansion factors developed for the entire region in the 2000 BATS survey are not rep- resentative of the geographical subset. New weighting and expansion factors would need to be developed using the same weighting and expansion procedure used by MTC. Validating the new factors requires comparing the subset demographic and mode share data to 2000 census data for the selected gen- eral urban and urban center areas before and after weight- ing and expansion. Weighting districts for the revised factors may continue to be done using PUMA districts since smaller scales of district (e.g., TAZs or census tracts) may prove to be too small. The need to develop new weighting and expansion factors is a challenging aspect of this methodology. It is a complex analytical effort with the potential for error by practitio- ners unfamiliar with the procedure. In applying the house- hold travel survey method, the practitioner should carefully review documented procedures and analysis of the weighting and expansion method the MPO applied to the HTS sampled data. Reviewing the procedures and comparing the survey’s sampled data to the weighted and expanded data can inform the practitioner of the expected change if the procedure were applied to the TAZs being used to extract adjustment factors. 4.3.3 Comparison of Data by Level of Disaggregation As stated previously, the number of usable trip records decreases as the records representing the total population are disaggregated into finer resolution to distinguish and isolate trips by (1) context, (2) land use, (3) day of week, (4) prox- imity to transit, (5) peak period, and (6) directionality. In general, a decrease in the number of trip records corresponds to a decrease in the precision of the survey findings. The 2000 BATS data provide over 236,000 total linked-trip records representing the total population. At each successive disaggregation, the number of records can diminish sub- stantially. For example, by isolating the trip records to those within general urban/urban center contexts, the trip records decrease by 85% to about 33,000. The smallest number of trips in any of the distillations is less than 50 records for one direction of 1 peak hour for the retail land use category. The average number of peak hour trip records for a given land use and transit proximity scenario is about 1,750. Figure 4.1 graphs the number of trip records that result from disaggre-

15 gating the data to represent context, land use type, weekday peak time periods, and proximity to transit. 4.4 Estimated Mode Share by Land Use The trip records for each land use category were aggre- gated by mode to determine mode share percentages for bike, walk, bus, rail, car, and other. The percentages in the last two columns of Table 4.1 are the factors included in the trip generation conversions. 4.5 Estimated Vehicle Occupancy by Land Use Table 4.2 presents an example (p.m. peak hour) summary of average vehicle occupancy adjustment factors by land use category. Occupancies based on the survey data are used to Trip Records Figure 4.1. Trip records by time period, proximity to transit, and land use. Table 4.1. Example mode share adjustment factors for San Francisco Bay Area (p.m. peak period). Land Use Context Criteria Type of Transit (Maximum 15- min Headway) Proximity Mode Share (All Trips) Transit Walk/Bike Residential General urban/ urban center (30–70 emp/gross acre) (10–40 DUs/gross acre) Bus Rail: ½-mile Bus: ¼-mile 17.5% 13.3% Rail 16.2% 13.7% Restaurant Bus 14.3% 16.6% Rail 15.5% 19.8% Retail Bus 10.7% 15.3% Rail 11.7% 16.3% Office/work Bus 22.4% 8.6% Rail 20.6% 9.4% Notes: DUs = dwelling units. Based on 2000 BATS data from 93 TAZs around rail stations and 170 TAZs around high-frequency bus stops. Rail and bus modes must have 15-min headways or shorter for at least 6 hours of the day to meet transit proximity criteria.

16 convert between person trips and vehicle trips in the pro- posed methodology. Appendix B of this technical report dis- plays the mode splits and vehicle occupancy by land use for (1) the entire Bay Area, (2) weekday trips in areas designated as general urban/urban center, (3) a.m. and p.m. peak hour trips by proximity to high-frequency bus service, and (4) a.m. and p.m. peak hour trips by proximity to rail service. Appen- dix C of this technical report presents the mode share and average vehicle occupancy adjustment factors by land use for daily, a.m. peak, and p.m. peak hours. 4.6 Use of Local Travel Demand Model to Derive Adjustment Factors If a regional or local travel demand forecasting model is available to the practitioner, and the model has a mode split submodel, data from the model can be used to derive mode share adjustment factors reflecting the existing or future planning horizon of the study area. The model-projected adjustment factors can only be used to estimate traffic gen- eration of the forecast year of the model. The A columns of Table 4.3 are predetermined using house- hold travel survey data and represent the percentage of total person trips by each trip purpose. The B and C columns, com- pleted by the practitioner, contain transit and nonmotorized trip purpose data for the study area TAZs and analysis year as output from a regional or local travel demand model. If the model being used does not provide mode share for walk or bike, then the method described can be used to obtain those mode splits. The D columns, the mode share adjustment factors, are calculated by multiplying the percent of trips by purpose (A) times the corresponding mode share percentages (B or C) and summing the totals into the corresponding cell of the D column. Table 4.2. Example vehicle occupancy adjustment factors for San Francisco Bay Area (p.m. peak period). Land Use Context Vehicle Occupancy Residential General urban/ urban center (30–70 emp/gross acre) (10–40 DUs/gross acre) 1.62 Restaurant 2.11 Retail 1.50 Office/work 1.27 Table 4.3. Template for determining mode share adjustment factors using local travel demand model forecasts by TAZ, San Francisco Bay Area (p.m. peak hour in general urban/urban center contexts). Land Use A B C D Typical Percent of Total Trips Analysis Year TAZ Projected Percent of Trips By Transit Analysis Year TAZ Projected Percent By Nonmotorized Modes Esmated Analysis Year Mode Share (%) HBW [1] HBO [2] NHB [3] HBW [1] HBO [2] NHB [3] HBW [1] HBO [2] NHB [3] Transit [1] Walk/Bi ke [2] Residential 46.2% 53.8% 0% ___% ___% ___% ___% ___% ___% Restaurant 0% 40.8% 59.2% Retail 0% 44.1% 55.9% Office/work 68.5% 0% 31.5% Example equaons for determining mode share by land use type: Transit mode share (from row with desired land use) = (cells A[1]*B[1] + A[2]*B[2] + A[3]*B[3]) = D[1] Walk/bike mode share (from row with desired land use) = (cells A[1]*C[1] + A[2]*C[2] + A[3]*C[3]) = D[2]

17 This chapter summarizes the development of a case study to extract adjustment factor data from a household travel survey. This information was originally submitted to NCHRP in a technical memorandum5 for this study. As suggested by the research panel, the following metropolitan areas were the focus of this evaluation: • Atlanta, GA. • Dallas, TX. • Salt Lake City, UT. • Denver, CO. • Washington, D.C. 5.1 Criteria for Selecting a Metropolitan Area with a Suitable Household Travel Survey The research team contacted staff responsible for coordi- nating transportation activities at the selected MPOs to assess the availability and usability of data necessary to meet the requirements of the case study. The research team interviewed MPO staff regarding their most recent household surveys and data included in their geographic information systems (GISs). In addition to the resulting interview findings, the project team reviewed other important characteristics to determine the suitability of the candidate metropolitan areas for the case study, including confirmation of the following criteria: • Existence of urban light or heavy rail transit serving gen- eral urban and/or urban core areas; • A metropolitan household travel survey was conducted when the rail system was in operation and contains the minimum required data variables, as described in previous submissions; • Metropolitan household travel survey data would be avail- able from the MPO for analysis by the research team, and MPO staff were available for questions and clarifications; • MPO or another agency would make available to the research team GIS data files with sufficient land use data to determine current household and employment data by TAZ, and the availability of current rail and bus transit route, stop, and schedules (preferably in a GIS); and • Metropolitan area contains a quantity of infill develop- ments in close proximity to qualifying rail stations or bus stops that can be successfully isolated for an accurate modal person-trip cordon count. 5.2 Selection of a Metropolitan Area Based on information provided by MPO staff, the previ- ously mentioned criteria, project requirements, and data col- lection considerations, the research team selected Washington, D.C. (the Metropolitan Washington Council of Governments) as the source of data for the case study. One of the princi- pal reasons for selecting Washington, D.C., was the age of its household survey (completed in 2008). This is an important consideration given that the validation process would have to rely on 2011 traffic count data. Older data would be difficult to substantiate given the difficulty of verifying that land uses and the location/availability of transit (or other variables strongly correlated to trip generation) have remained constant over extended periods of time. The research team also ranked Denver a strong candidate because its ongoing household travel survey is expected to have more records than Washington, D.C.; however, com- plete data would not have been available in time. The research team eliminated the remaining possible metro areas (Atlanta, Dallas, and Salt Lake City) from further consideration because the available data were dated or the survey’s dataset contained a small number of records. Summary findings for each of the candidate metropolitan areas are provided in Appendix A of this technical report. The research team identified Washington, D.C., as one of the most viable candidates because it has an established transit C H A P T E R 5 Selection of a Household Travel Survey as a Case Study

18 system that operates within urban and suburban areas, and local governments promote infill and transit-oriented develop- ments around transit stations. The Washington Metropolitan Area Transit Authority operates and maintains Metrorail and Metrobus, the major rail and bus transit systems in the area. 5.3 Sufficiency of the Dataset One potential challenge with the Washington, D.C., data was whether there would be sufficient records at the level of detail required by the methodology. Although the research team did not expect that this would negate Washington, D.C., as a candidate, it was not possible to make a final assessment until the required linked-trip records were extracted from the total dataset. The linked-trip records were required to meet multiple criteria, including the provision of trip purpose, primary mode share, geographic identifiers, and the ability to discern origin and destination land use type (i.e., single family, multi family, office, retail, and restaurant). The most recent survey for Wash- ington, D.C., was completed in 2008 and includes 11,000 house- holds (approximately 88,000 trips). Following is a summary of the Washington, D.C., house- hold travel survey data received. For reference, the number of trips from the initial demonstration project that used the 2000 BATS is provided in parentheses: • 87,926 (236,573 in 2000 BATS dataset) total trips. • 63,107 (176,083) trips with origin purpose or destination purpose of “home.” • 27,210 (72,275) trips with origin purpose or destination purpose of “work.” • 31,462 (67,295) trips with origin purpose or destination purpose of “shop.” • 8,088 (36,827) trips with origin activity or destination activity of “eat a meal outside of home or work.” 5.4 Next Steps in the Process The next three steps in the process were to extract the data outlined in the following. Details of the extraction processes undertaken are provided in Chapter 6. • The research team requested detailed GIS information for the Washington, D.C., metropolitan area aggregated at the TAZ level to analyze and identify zones that meet the general urban/urban center criteria. Once the GIS data were obtained and analyzed, the research team developed a map of the zones classified as general urban/urban center. Subsequently, the research team extracted the appropriate records from the dataset and assessed the viability of the resulting data for use in the proposed methodology. • Viable data were used to produce a series of mode share matrices by land use type, time of day, and direction of travel. These matrices were used to populate the variables in the proposed methodology’s equations. The research team then documented the viability assessment in a brief report that also documents limitations and potential sources of error in the data. • Once the research team confirmed the viability of the data, they systematically sought sites within the Washington, D.C., metropolitan area for use in the validation case study. The search process used GIS data, Google Maps, and the knowledge of the research team’s staff (Kimley-Horn and Associates in Northern Virginia). The research team then prepared a validation site-selection work plan document- ing selection criteria and data collection procedures.

19 This chapter describes how the research team distilled household travel survey data for the Washington, D.C., case study selected in Chapter 5 to segregate trips and mode share to and from TAZs that met the criteria for the four primary land use categories being studied (residential, restaurant, retail, and office). Further, this chapter describes the process used to identify candidate sites within these TAZs that could be used to collect cordon traffic counts to validate the meth- odology proposed in this research project. 6.1 GIS Analysis The research team obtained GIS information from several sources for the purpose of identifying TAZs that met the cri- teria described in the previous section. Following are the pri- mary GIS layers used during the analysis, as well as the source of the data: • Traffic analysis zones – 3,669 records were included in the database provided by the Metropolitan Washington Coun- cil of Governments. • Metro lines – Five records were included in the database obtained from the website http://data.dc.gov/. • Metro stations – 86 records were included in the database obtained from the website http://data.dc.gov/. • Bus lines – 178 records were included in the database obtained from the website http://data.dc.gov/. • Bus stops – 12,091 records were included in the database obtained from the website http://data.dc.gov/. Using information provided in the TAZ layer, the research team first identified TAZs that met the general urban/urban center criteria detailed in Chapter 4. Accordingly, the TAZ was required to have employment per gross acre exceeding 70 or households per gross acre exceeding 40 to be identified as general urban/urban center. The Metro station GIS layer was used to identify TAZs in which transit service covers greater than 33% of the TAZs physical area based on a ½-mile service radius extending out from each of the stations. As described in Chapter 4, “high-frequency bus stops” are defined as being stops served by lines with 15-min headways that are maintained for at least 6 hours of each weekday or are located within transit corridors with multiple lines trav- eling in the same direction that effectively meet the 15-min headway criterion. During the next step, the research team identified TAZs in which greater than 33% of the zone’s area was accessible based on a ¼-mile service radius extending out from high-frequency bus stops. The research team’s analysis of the San Francisco Bay Area travel survey data was comparatively simple given that MTC provides GIS data including bus transit headways and sched- ules by stop. Based on a review of available GIS files from the candidate metropolitan areas, recent bus transit headway data in GIS format may not typically be available. As a result, these data had to be obtained and coded into GIS separately. Figure 6.1 demonstrates how the GIS analysis was carried out. 6.2 Household Travel Survey Analysis The investigators then reviewed the linked-trip data of the individual TAZs meeting both the general urban/urban cen- ter and rail transit proximity criteria described earlier. Sub- sequently, the data contained in the linked-trip records were further disaggregated by applying the following screens: • Trip purpose or activity at origin and destination to isolate trips representing each of the four proposed study land uses (residential, restaurant, retail, and office). • a.m. and p.m. peak periods. • Inbound and outbound records for origin and destination. • Vehicle occupancy for trips by automobile mode. This same process was applied to the general urban/urban center TAZs that met the high-frequency bus stop criteria. C H A P T E R 6 Analysis of Household Travel Survey Data

20 The Washington, D.C., HTS is activity-based and not place- based, so the type of land use at the origin or destination of the trip needs to be inferred from the trip purpose description provided in the trip records. Survey participants selected from 13 different predetermined trip purposes (or activities). From these, the investigators chose those activities that best repre- sented the activity for a specific land use. Given the manner in which the household travel survey was conducted, trips made by delivery or service people (with the exception of survey participants who are employed in these fields) were not cap- tured. This is a potential source of undercounting in estimat- ing actual trips for each of the study’s land uses. The following is a brief overview regarding trip purposes and potential sources of error for each of the study’s four land uses, as also documented in the application of the methodol- ogy to the 2000 BATS: • Residential trips are identified as those with an origin trip purpose or a destination trip purpose classified as “home.” As one would expect, the vast majority of the trip records are reported by residents, and nonresident trips are not well represented. As such, the records likely under-report trips related to deliveries, services, and guest trips unless recorded as an activity by another survey participant. • Restaurant trips are identified as those with an origin activ- ity or a destination activity classified as “eat a meal outside of home or work.” The trip records only include patron trips and do not capture employee, delivery, or service trips unless recorded by another survey participant. This may affect calculation of the mode split or vehicle occupancy since restaurant patrons may have significantly different travel characteristics to and from restaurants than employ- ees of the restaurant. Employee-related trips are captured in the “work” trip purpose. • Retail or shopping center trips are identified as those with an origin trip purpose or a destination trip purpose classi- fied as “shop.” Like restaurants, the trip records for shop- ping only include patron trips and do not capture employee, delivery, or service trips unless recorded by another survey participant. This may affect calculation of the mode split Figure 6.1. Example GIS analysis of TAZs.

21 and vehicle occupancy. Employee-related trips are captured in the “work” trip purpose. • Office building trips are identified as those with an origin trip purpose or a destination trip purpose classified as “work.” Based on this descriptor, it is not possible to positively deter- mine whether a work trip has an origin or destination that is an office building. During the Phase 1 assessment of the 2000 BATS data, it was determined that this potential source of error may be acceptable because the MTC’s detailed docu- mentation of the household survey data suggested that mode split and vehicle occupancy for office trips were similar to those of aggregated non-office work trips. The investigators are not certain this finding is applicable to other metropoli- tan areas and are attempting to find sources that will validate the premise in the Washington, D.C., area. 6.2.1 Extracting Mode Split and Vehicle Occupancy from Household Travel Survey Data The linked-trip records for each land use category were dis- aggregated by mode to determine mode split percentages for transit, auto (driver and passenger), walk, bike, and other. This resulted in the derivation of adjustment factors by land use in proximity to rail stations and high-frequency bus routes for daily, a.m. peak, and p.m. peak hours. Detailed mode share tables are included in Appendix D of this technical report. This information was then used to populate the mode split adjustment and vehicle occupancy tables, shown in Appen- dix E of this technical report. The tables in the appendix also include the vehicle occupancy adjustments. Using the proposed methodology, mode split data were used to determine the share of person trips generated by a land use category with travel by automobile, and the vehicle occupancy derived from the survey data was used to con- vert “person vehicle trips” to “vehicle trips.” It should be noted that vehicle occupancy data from the household travel survey could only be obtained when the participant was the driver of the vehicle. Vehicle occupancy when the participant was a pas- senger was not collected as part of this HTS. The mode split linked-trip records were disaggregated to inbound and out- bound trips for the four land use categories that met the rail criteria. This same process was conducted on the trip records that met the high-frequency bus stop criteria.

22 This chapter describes the process for collecting the empiri- cal data for validating the Washington, D.C., case study. The process of selecting the candidate sites is summarized in the following: • Finalize cordon count procedures. The research team finalized data collection procedures initially developed earlier in the study. The procedures were designed to col- lect peak-hour automobile trip data for establishing trip generation rates, and secondarily to collect person-trip data for establishing mode share. • Conduct cordon counts at urban infill sites and subur- ban control sites. Assess candidate sites against meeting the selection criteria, and assess suburban sites selected for use as control sites. Data were collected for representative sites making up the four primary land use categories (residential, restaurant, retail, and office) identified for study. • Review and compare cordon count data. Check the count data for errors or unreasonable results. Compare the vehicle cordon counts to an estimate of site traffic calculated using ITE data for similar land uses. Use the cordon counts and the site’s independent variable (e.g., square feet of floor area, dwelling units) to derive a trip generation rate for the validation site and compare the validation site’s trip genera- tion rates with the trip generation rates published in the ITE Trip Generation Manual for the same land use classification. 7.1 Selecting Urban Infill Sites for Cordon Counts The research team identified the TAZs in the Washington, D.C., metropolitan area that met the criteria established for defining general urban/urban center contexts. Candidate sites within these TAZs were screened based on their proxim- ity to “high-frequency bus stops” or rail stations. The sites remaining after screening for context and transit proximity were further evaluated using aerial photography and subse- quent field investigations to confirm that they possessed the required qualitative characteristics and that the sites could be cost-effectively surveyed. Field investigations recorded observable information regarding land uses, data collection requirements and challenges, and other local conditions. Finally, the research team reviewed the details of each site and prioritized the candidate locations based on: • The ability of the site to meet site-selection criteria. • That the surrounding context met the definition of general urban/urban center. • That the site met the criteria for proximity to transit, and the transit system met the criteria for quality of service. • That the necessary data could be collected from the site cost-effectively. Using the approach described, the research team identified 30 candidate sites but determined that further information was needed before expending limited resources on collecting data—information that could only be confirmed visually in the field. Senior members of the research team experienced in data collection visited the candidate sites on weekdays during the month of November 2011. Site-specific data collection plans were prepared for the high-priority locations that satisfied the scrutiny of the team’s engineers in the field. As a result of the site visits, the research team rejected five of the candidate sites primarily related to the practicality of collecting data (three residential, one office, and one restaurant). The research team replaced the five rejected candidate sites with new residential, office, and retail sites (one residential, two office, and two retail sites). The replacement sites were subjected to the selection criteria and field investigations to determine their viability as candidates for use in validation. The final step before collecting data required by the research team was to contact each site’s owner or its representative to confirm its participation in the study and to determine required land use attributes and independent variables. Owner inquiries C H A P T E R 7 Selection of Candidate Sites for Cordon Counts

23 were, as required, supplemented with public database searches to determine required land use attributes (i.e., gross square feet, number of employees, residential units). 7.1.1 Selecting Suburban Control Sites for Cordon Counts In addition to the urban infill sites selected for data collec- tion, a representative suburban site was identified for each of the four land uses to serve as the control population (to be compared with ITE trip generation data, which are primarily based on suburban locations). The suburban sites were also subject to meeting selection criteria, field investigations, and preparation of site-specific data collection plans. 7.1.2 Final Validation Sites Based on the research panel’s input, 14 sites were ultimately selected from the candidate pool. Ten of the 14 locations rep- resented urban infill developments (four residential, three office, two retail, and one restaurant site). The remaining four sites represented developments in suburban contexts for use as control sites. Appendix F of this technical report includes a summary of candidate sites and their relative prioritization. 7.2 Summary of the Data Collection Procedures 7.2.1 Data Collection Dates and Time Periods The cordon counts of automobiles entering and exiting the sites were manually conducted during peak periods at each of the 14 study sites. For the purposes of this research study, a.m. and p.m. peak periods for the residential, office, and retail sites were identified as occurring between 7 a.m. and 9 a.m. and 4 p.m. and 6 p.m., respectively. These time periods contain the typical morning and afternoon hour representing the “peak hour of the adjacent street”—a common time period used in traffic impact analyses for the aforementioned land uses. As the restau- rant location generated an insignificant level of traffic during the traditional morning peak hour, counts were conducted dur- ing the lunch peak period (11:30 a.m. to 1:30 p.m.) as well as the traditional afternoon peak period (4:00 p.m. to 6:00 p.m.). The counts were conducted mid-week ( Tuesday, Wednesday, and Thursday) between November 8, 2011, and November 10, 2011, or between November 15, 2011, and November 17, 2011. 7.2.2 Data Collection Methodology Research team personnel stationed at each of the study site’s parking lot or garage access driveways recorded inbound and outbound vehicle trips and vehicle occupancy during the peak periods in 15-min increments. Since the intent of the cordon counts is to capture all of the vehicle trips generated by the site, the survey personnel recorded vehicle trips that were observed parking off-site (e.g., the adjacent street) and for which occu- pants subsequently entered/exited the study site. The selec- tion process attempted to avoid sites where visitors to the site chose to park on-street even when there was available and free off-street parking at the site itself. During the preliminary site investigation visits, candidate sites that were observed to have a significant number of trips originating from vehicles parked on-street were not selected for the data collection phase. Fur- ther, potential study sites were excluded from consideration if attractive or viable off-site parking options, such as nearby parking garages or lots, were available within reasonable prox- imity to the site. Off-site parking potentially introduces errors into the accuracy of the cordon counts. 7.2.3 Additional Data Collection In addition to the vehicle trips, survey personnel recorded other trip information that could be used as either local data in the proposed trip generation methodology or could be useful in validating the methodology. These data include: • Person trips. Survey personnel recorded the number of people entering and exiting the site’s building(s) regard- less of their mode of access. This information was recorded only if all of the building access points were visible to sur- vey personnel and recording these trips did not interfere with the accuracy of the vehicle counts. • Mode of access. Survey personnel recorded the mode of travel for each of the persons observed entering the site’s building when recording person trips. Mode of access (vehi- cle, walk, bike, or transit) was recorded only if the recorded person’s mode of transportation was clearly observed by the survey personnel. • General observations of site conditions. Survey person- nel observed and recorded the level of bicycle and transit activity in the streets surrounding the site, the surveyor’s judgment of the walkability of the adjacent streets (in terms of comfort, safety, and directness), and other conditions unique to each site that might influence the site’s trip gen- eration characteristics. Based on the general observations recorded by both the sur- vey personnel and supervising research team members during the course of the data collection, there was general agreement that there was a greater use of more nonmotorized modes of travel in the vicinity of the selected urban infill sites than the selected suburban control sites (primarily pedestrian and bike). Pedestrians were frequently observed walking to and from the general direction of Metro stations, while suburban sites, in contrast, were observed to have little or no pedestrian and bicycle activity. Appendix G of this technical report includes a sample of the results of the data collection surveys.

24 Atlanta (Atlanta Regional Commission) • Most recent survey completed in 2002. • Dataset includes approximately 8,000 households and 151,000 trips. • Survey data are not geo-coded, and transit information is not readily available in GIS. • TAZ information is available in GIS. • Currently in the process of preparing a new household sur- vey, anticipate completion in 2012. • Candidate metro area not recommended based on age of data and limited number of records. In addition, lack of GIS coded data would increase challenges. Dallas (North Central Texas Council of Governments) • Most recent survey completed in 2009 as part of the NHTS. • Dataset includes approximately 5,900 households and 49,000 trips. • TAZ and transit information in GIS, including bus routes. • It is unknown when household travel survey will be updated. • Candidate metro area not recommended based on expec- tation that the number of records available would not meet needs of the methodology. Denver (Denver Regional Council of Governments) • Survey conducted in 2010, and currently in process of developing weighting and expansion factors, which should be complete by October 2011. • Dataset includes approximately 12,000 households, 100,000 trips. • TAZ and transit information in GIS, including bus routes. • Public use files not yet available; however, may be able to obtain trip data without personal information—discussed with MPO. • Candidate metro area not recommended based on unavail- ability of weighting and expansion factors. However, if there are insufficient trip records from the Washington, D.C., survey, Denver could be substituted as the case study metro area. Salt Lake City (Wasatch Front Regional Council) • Last survey conducted in 1993. • Research team waited for response on size of dataset and number of trip records. • MPO in the process of selecting consultant to conduct statewide household travel survey with anticipated com- pletion in 2012 or 2013. • Candidate metro area not recommended based on age of survey data. Washington, D.C. (Metropolitan Washington Council of Governments) • Most recent survey completed in 2008. • Dataset includes approximately 11,000 households and 88,000 trips. • TAZ and transit information in GIS, including bus routes. • Candidate metro area selected by research team based on availability of recent data, immediate availability of public use files, and available GIS information. S U P P L E M E N T A L T E C H N I C A L R E P O R T A P P E N D I X A Overview of Household Travel Surveys Assessed for Case Study

25 S U P P L E M E N T A L T E C H N I C A L R E P O R T A P P E N D I X B Detailed Mode Share Tables for San Francisco Bay Area Example

26

27 S U P P L E M E N T A L T E C H N I C A L R E P O R T A P P E N D I X C Example Output Tables for San Francisco Bay Area Infill Area Mode Share and Vehicle Occupancy Adjustment Factors Example Daily Output Tables for San Francisco Bay Area Infill Area Mode Split and Vehicle Occupancy Adjustments to ITE Trip Generation Rates/Equations Land Use Context Criteria Type of Transit (Max 15-min Headway) Proximity Mode Share (All Trips) Transit Walk/Bike Residential General urban/ urban center (30–70 emp/gross acre) (10–40 DUs/gross acre) Bus Rail: ½-mile Bus: ¼-mile 13.3% 16.7% Rail 13.0% 17.1% Restaurant Bus 8.2% 27.7% Rail 8.2% 29.7% Retail Bus 6.0% 19.1% Rail 6.4% 19.9% Office/work Bus 16.7% 13.4% Rail 15.3% 14.6% Based on 2000 BATS data from 93 TAZs around rail stations and 170 TAZs around high-frequency bus stops. Rail and bus modes must have 15-min headways or shorter for at least 6 hours of the day to meet transit proximity criteria. Table A. Example summary of mode share adjustment factors by land use and proximity to transit for the San Francisco Bay Area (daily). Land Use Context Average Vehicle Occupancy Residential General urban/urban center (30–70 emp/gross acre) (10–40 DUs/gross acre) 1.66 Restaurant 1.91 Retail 1.56 Office/work 1.31 Table B. Example summary of average vehicle occupancy adjustment factors by land use for the San Francisco Bay Area (daily).

28 Example a.m. Peak Hour Output Tables for San Francisco Bay Area Infill Area Mode Share and Vehicle Occupancy Adjustments to ITE Trip Generation Rates/Equations Land Use Context Criteria Type of Transit (Max 15-min Headway) Proximity Mode Share (All Trips) Transit Walk/Bike Residential General urban/ urban center (30–70 emp/gross acre) (10–40 DUs/gross acre) Bus Rail: ½ mile Bus: ¼ mile 20.2% 13.4% Rail 19.3% 13.2% Restaurant Bus 26.7% 20.8% Rail 25.3% 20.6% Retail Bus 12.7% 11.4% Rail 13.1% 12.3% Office/work Bus 23.5% 8.4% Rail 20.2% 13.4% Based on 2000 BATS data from 93 TAZs around rail stations and 170 TAZs around high-frequency bus stops. Rail and bus modes must have 15-min headways or shorter for at least 6 hours of the day to meet transit proximity criteria. Table A. Example summary of mode share adjustment factors by land use and proximity to transit for the San Francisco Bay Area (a.m. peak hour). Land Use Context Average Vehicle Occupancy Residential General urban/ urban center (30–70 emp/gross acre) (10–40 DUs/gross acre) 1.62 Restaurant 1.37 Retail 1.49 Office/work 1.36 Table B. Example summary of average vehicle occupancy adjustment factors by land use for the San Francisco Bay Area (a.m. peak hour). Land Use Context Criteria Type of Transit (Max 15-min Headway) Proximity Mode Share (All Trips) Transit Walk/Bike Residential General urban/ urban center (30–70 emp/gross acre) (10–40 DUs/gross acre) Bus Rail: ½-mile Bus: ¼-mile 17.5% 13.3% Rail 16.2% 13.7% Restaurant Bus 14.3% 16.6% Rail 15.5% 19.8% Retail Bus 10.7% 15.3% Rail 11.7% 16.3% Office/work Bus 22.4% 8.6% Rail 17.5% 13.3% Based on 2000 BATS data from 93 TAZs around rail stations and 170 TAZs around high-frequency bus stops. Rail and bus modes must have 15-min headways or shorter for at least 6 hours of the day to meet transit proximity criteria. Table A. Example summary of mode share adjustment factors by land use and proximity to transit for the San Francisco Bay Area (p.m. peak hour). Example p.m. Peak Hour Output Tables for San Francisco Bay Area Infill Area Mode Share and Vehicle Occupancy Adjustments to ITE Trip Generation Rates/Equations

29 Land Use Context Average Vehicle Occupancy Residential General urban/ urban center (30–70 emp/gross acre) (10–40 DUs/gross acre) 1.62 Restaurant 2.11 Retail 1.50 Office/work 1.27 Table B. Example summary of average vehicle occupancy adjustment factors by land use for the San Francisco Bay Area (p.m. peak hour).

30 S U P P L E M E N T A L T E C H N I C A L R E P O R T A P P E N D I X D Detailed Mode Share Tables for the Washington, D.C., Case Study

31 Household Travel Survey Linked Trip Analysis Mode Split Summary by Scenario DC Area Eat Out DC Area Residential DC Area Shopping DC Area Work Transit 250 3.1% Transit 4438 7.0% Transit 844 3.0% Transit 3665 13.7% Auto Driver 3970 49.3% Auto Driver 38514 61.0% Auto Driver 17684 62.6% Auto Driver 18553 69.4% Auto Passenger 2242 27.9% Auto Passenger 13003 20.6% Auto Passenger 6068 21.5% Auto Passenger 1352 5.1% Walk 1464 18.2% Walk 4078 6.5% Walk 3434 12.2% Walk 2634 9.8% Bike 38 0.5% Bike 382 0.6% Bike 92 0.3% Bike 216 0.8% Other 83 1.0% Other 2692 4.3% Other 141 0.5% Other 322 1.2% Total Trip Records 8047 Total Trip Records 63107 Total Trip Records 28263 Total Trip Records 26742 Vehicle Occupancy 1.66 Vehicle Occupancy 1.35 Vehicle Occupancy 1.40 Vehicle Occupancy 1.13 Urban Rail Eat Out Urban Rail Residential Urban Rail Shopping Urban Rail Work Transit 120 8.8% Transit 1856 22.5% Transit 372 9.7% Transit 1515 28.5% Auto Driver 391 28.6% Auto Driver 3735 45.4% Auto Driver 1571 40.8% Auto Driver 2285 43.0% Auto Passenger 218 16.0% Auto Passenger 1068 13.0% Auto Passenger 458 11.9% Auto Passenger 236 4.4% Walk 584 42.8% Walk 1290 15.7% Walk 1364 35.4% Walk 1069 20.1% Bike 15 1.1% Bike 137 1.7% Bike 38 1.0% Bike 98 1.8% Other 37 2.7% Other 146 1.8% Other 45 1.2% Other 108 2.0% Total Trip Records 1365 Total Trip Records 8232 Total Trip Records 3848 Total Trip Records 5311 Vehicle Occupancy 1.66 Vehicle Occupancy 1.30 Vehicle Occupancy 1.34 Vehicle Occupancy 1.15 Urban Rail Eat Out Urban Rail Residential Urban Rail Shopping Urban Rail Work Transit 6 12.2% Transit 653 32.5% Transit 44 19.7% Transit 569 38.8% Auto Driver 17 34.7% Auto Driver 848 42.2% Auto Driver 89 39.9% Auto Driver 641 43.7% Auto Passenger 7 14.3% Auto Passenger 213 10.6% Auto Passenger 10 4.5% Auto Passenger 59 4.0% Walk 19 38.8% Walk 219 10.9% Walk 78 35.0% Walk 141 9.6% Bike 0 0.0% Bike 40 2.0% Bike 1 0.4% Bike 34 2.3% Other 0 0.0% Other 38 1.9% Other 1 0.4% Other 22 1.5% Total Trip Records 49 Total Trip Records 2011 Total Trip Records 223 Total Trip Records 1466 Vehicle Occupancy 1.35 Vehicle Occupancy 1.30 Vehicle Occupancy 1.16 Vehicle Occupancy 1.15 Urban Rail Eat Out Urban Rail Residential Urban Rail Shopping Urban Rail Work Transit 23 16.1% Transit 506 27.7% Transit 107 16.5% Transit 478 35.6% Auto Driver 48 33.6% Auto Driver 776 42.4% Auto Driver 292 45.0% Auto Driver 610 45.4% Auto Passenger 36 25.2% Auto Passenger 243 13.3% Auto Passenger 95 14.6% Auto Passenger 67 5.0% Walk 32 22.4% Walk 257 14.0% Walk 142 21.9% Walk 140 10.4% Bike 0 0.0% Bike 32 1.7% Bike 6 0.9% Bike 28 2.1% Other 4 2.8% Other 16 0.9% Other 7 1.1% Other 21 1.6% Total Trip Records 143 Total Trip Records 1830 Total Trip Records 649 Total Trip Records 1344 Vehicle Occupancy 1.69 Vehicle Occupancy 1.34 Vehicle Occupancy 1.36 Vehicle Occupancy 1.17 Urban Bus Eat Out Urban Bus Residential Urban Bus Shopping Urban Bus Work Transit 138 7.5% Transit 2288 18.8% Transit 441 8.1% Transit 1815 24.4% Auto Driver 646 34.9% Auto Driver 6177 50.8% Auto Driver 2591 47.5% Auto Driver 3809 51.2% Auto Passenger 337 18.2% Auto Passenger 1620 13.3% Auto Passenger 720 13.2% Auto Passenger 342 4.6% Walk 674 36.4% Walk 1634 13.4% Walk 1605 29.4% Walk 1222 16.4% Bike 17 0.9% Bike 174 1.4% Bike 43 0.8% Bike 125 1.7% Other 40 2.2% Other 266 2.2% Other 56 1.0% Other 128 1.7% Total Trip Records 1852 Total Trip Records 12159 Total Trip Records 5456 Total Trip Records 7441 Vehicle Occupancy 1.62 Vehicle Occupancy 1.28 Vehicle Occupancy 1.33 Vehicle Occupancy 1.13 Urban Bus Eat Out Urban Bus Residential Urban Bus Shopping Urban Bus Work Transit 8 10.4% Transit 806 27.3% Transit 50 15.4% Transit 698 33.4% Auto Driver 33 42.9% Auto Driver 1398 47.3% Auto Driver 154 47.5% Auto Driver 1074 51.5% Auto Passenger 13 16.9% Auto Passenger 330 11.2% Auto Passenger 22 6.8% Auto Passenger 82 3.9% Walk 22 28.6% Walk 282 9.5% Walk 94 29.0% Walk 161 7.7% Bike 1 1.3% Bike 53 1.8% Bike 2 0.6% Bike 44 2.1% Other 0 0.0% Other 87 2.9% Other 2 0.6% Other 28 1.3% Total Trip Records 77 Total Trip Records 2956 Total Trip Records 324 Total Trip Records 2087 Vehicle Occupancy 1.36 Vehicle Occupancy 1.27 Vehicle Occupancy 1.20 Vehicle Occupancy 1.13 Urban Bus Eat Out Urban Bus Residential Urban Bus Shopping Urban Bus Work Transit 29 13.8% Transit 631 24.0% Transit 126 13.5% Transit 585 31.0% Auto Driver 89 42.4% Auto Driver 1255 47.8% Auto Driver 481 51.4% Auto Driver 992 52.5% Auto Passenger 51 24.3% Auto Passenger 358 13.6% Auto Passenger 142 15.2% Auto Passenger 93 4.9% Walk 37 17.6% Walk 312 11.9% Walk 172 18.4% Walk 161 8.5% Bike 0 0.0% Bike 40 1.5% Bike 6 0.6% Bike 36 1.9% Other 4 1.9% Other 32 1.2% Other 8 0.9% Other 23 1.2% Total Trip Records 210 Total Trip Records 2628 Total Trip Records 935 Total Trip Records 1890 Vehicle Occupancy 1.71 Vehicle Occupancy 1.32 Vehicle Occupancy 1.36 Vehicle Occupancy 1.16 Pr ox to Hi gh Fr eq .B us St op Ge ne ra lU rb an Da ily Pr ox to Hi gh Fr eq .B us St op Ge ne ra lU rb an AM Pe ak Pr ox to Hi gh Fr eq .B us St op Ge ne ra lU rb an PM Pe ak Ge ne ra lU rb an AM Pe ak Pr ox im ity to Ra ilS ta o n Ge ne ra lU rb an PM Pe ak W as hi ng to n DC Ar ea Pr ox im ity to Ra ilS ta tio n Ge ne ra lU rb an Da ily Pr ox im ity to Ra ilS ta o n

32 S U P P L E M E N T A L T E C H N I C A L R E P O R T A P P E N D I X E Output Tables for Washington, D.C., Infill Area Mode Share and Vehicle Occupancy Adjustment Factors Washington, D.C., Household Travel Survey Linked-Trip Analysis Output Tables for Infill Area Mode Split and Vehicle Occupancy Adjustments to ITE Trip Generation Rates/Equations Land Use Context/Area Type Transit Mode Available (<15-min Headway) Transit Proximity Mode Share Percent by Transit Percent by Nonmotorized Rail <½ mile 28.5% 22.0% Residential General urban/urban center Bus <¼ mile 18.8% 14.9% Rail <½ mile 22.5% 17.3% Restaurant Bus <¼ mile 7.5% 37.3% Rail <½ mile 8.8% 43.9% Retail Bus <¼ mile 8.1% 30.2% Rail <½ mile 9.7% 36.4% Office/work Bus <¼ mile 24.4% 18.1% Table A. Summary of mode share adjustment factors by land use and proximity to transit for the Washington, D.C., case study area (daily). Land Use Context/Area Type A (Bus) A (Rail) Veh. Occ. Veh. Occ. Residential General urban/urban center 1.28 1.30 Restaurant 1.62 1.66 Retail 1.33 1.34 Office/work 1.13 1.15 Table B. Summary of average vehicle occupancy adjustment factors by land use for the Washington, D.C., case study area (daily).

33 Washington, D.C., Household Travel Survey Linked-Trip Analysis Output Tables for Infill Area Mode Split and Vehicle Occupancy Adjustments to ITE Trip Generation Rates/Equations Land Use Context/Area Type Transit Mode Available (<15-min Headway) Transit Proximity Mode Share Percent by Transit Percent by Nonmotorized Rail <½ mile 38.8% 11.9% Residential General urban/urban center Bus <¼ mile 27.3% 11.3% Rail <½ mile 32.5% 12.9% Restaurant Bus <¼ mile 10.4% 29.9% Rail <½ mile 12.2% 38.8% Retail Bus <¼ mile 15.4% 29.6% Rail <½ mile 19.7% 35.4% Office/work Bus <¼ mile 33.4% 9.8% Table A. Summary of mode share adjustment factors by land use and proximity to transit for the Washington, D.C., case study area (a.m. peak hour). Land Use Context/Area Type A (Bus) A (Rail) Veh. Occ. Veh. Occ. Residential General urban/urban center 1.27 1.30 Restaurant 1.36 1.35 Retail 1.20 1.16 Office/work 1.13 1.15 Table B. Summary of average vehicle occupancy adjustment factors by land use for the Washington, D.C., case study area (a.m. peak hour). Land Use Context/Area Type Transit Mode Available (<15-min Headway) Transit Proximity Mode Share Percent by Transit Percent by Nonmotorized Rail <½ mile 35.6% 12.5% Residential General urban/urban center Bus <¼ mile 24.0% 13.4% Rail <½ mile 27.7% 15.8% Restaurant Bus <¼ mile 13.8% 17.6% Rail <½ mile 16.1% 22.4% Retail Bus <¼ mile 13.5% 19.0% Rail <½ mile 16.5% 22.8% Office/work Bus <¼ mile 31.0% 10.4% Table A. Summary of mode share adjustment factors by land use and proximity to transit for the Washington, D.C., case study area (a.m. peak hour). Washington, D.C., Household Travel Survey Linked-Trip Analysis Output Tables for Infill Area Mode Split and Vehicle Occupancy Adjustments to ITE Trip Generation Rates/Equations

34 Land Use Context/Area Type A (Bus) A (Rail) Veh. Occ. Veh. Occ. Residential General urban/urban center 1.32 1.34 Restaurant 1.71 1.69 Retail 1.36 1.36 Office/work 1.16 1.17 Table B. Summary of average vehicle occupancy adjustment factors by land use for the Washington, D.C., case study area (p.m. peak hour).

35 S U P P L E M E N T A L T E C H N I C A L R E P O R T A P P E N D I X F Prioritization of Candidate Sites for Cordon Counts

37 S U P P L E M E N T A L T E C H N I C A L R E P O R T A P P E N D I X G Example Data Summaries for Candidate Sites

38

39

41

42

45 1. Institute of Transportation Engineers. Trip Generation Manual, An Informational Report, 9th Edition. Institute of Transportation Engineers. Washington, D.C., 2012. 2. Tierney, K.; Decker, S.; Proussaloglou, K.; Rossi, T.; Ruiter, E.; McGuckin, N.; Tierney, K. (Editor), Travel Survey Manual: How to Do a Survey, U.S. Department of Transportation and U.S. Environ- mental Protection Agency, June 1996. 3. Kimley-Horn and Associates, Inc. Trip-Generation Rates for Urban Infill Land Uses in California, Phase 2: Data Collec- tion, Final Report. California Department of Transportation. June 2009. 4. Metropolitan Planning Commission, Sample Weighing and Expan- sion Working Paper 2 – 1990 MTC Travel Survey, National Trans- portation Library, June 1993. 5. NCHRP Project 8-66: Revised Phase 2 Methodology Case Study Approach, National Cooperative Highway Research Program, May 5, 2011. Notes and Citations to Supplemental Technical Report

Abbreviations and acronyms used without definitions in TRB publications: A4A Airlines for America AAAE American Association of Airport Executives AASHO American Association of State Highway Officials AASHTO American Association of State Highway and Transportation Officials ACI–NA Airports Council International–North America ACRP Airport Cooperative Research Program ADA Americans with Disabilities Act APTA American Public Transportation Association ASCE American Society of Civil Engineers ASME American Society of Mechanical Engineers ASTM American Society for Testing and Materials ATA American Trucking Associations CTAA Community Transportation Association of America CTBSSP Commercial Truck and Bus Safety Synthesis Program DHS Department of Homeland Security DOE Department of Energy EPA Environmental Protection Agency FAA Federal Aviation Administration FHWA Federal Highway Administration FMCSA Federal Motor Carrier Safety Administration FRA Federal Railroad Administration FTA Federal Transit Administration HMCRP Hazardous Materials Cooperative Research Program IEEE Institute of Electrical and Electronics Engineers ISTEA Intermodal Surface Transportation Efficiency Act of 1991 ITE Institute of Transportation Engineers MAP-21 Moving Ahead for Progress in the 21st Century Act (2012) NASA National Aeronautics and Space Administration NASAO National Association of State Aviation Officials NCFRP National Cooperative Freight Research Program NCHRP National Cooperative Highway Research Program NHTSA National Highway Traffic Safety Administration NTSB National Transportation Safety Board PHMSA Pipeline and Hazardous Materials Safety Administration RITA Research and Innovative Technology Administration SAE Society of Automotive Engineers SAFETEA-LU Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (2005) TCRP Transit Cooperative Research Program TEA-21 Transportation Equity Act for the 21st Century (1998) TRB Transportation Research Board TSA Transportation Security Administration U.S.DOT United States Department of Transportation