Activity-Based Travel Demand Models: A Primer (2014)

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.

51.1 WHY DO WE MODEL TRAVEL? Transportation decision makers confront diffi - cult questions and must make informed choices. How will the national, regional, or even local transportation system perform 30 years into the future? What policies or investments could infl uence this performance? How will eco- nomic, demographic, or land use changes af- fect transportation system performance? Will travel demand management strategies or in- telligent transportation systems alleviate con- gestion? Will a new transit investment attract riders? Given a set of desired outcomes, deci- sion makers must identify capital investments and policies that will achieve these objectives. Travel models are created to support decision making by providing information about the impacts of alternative transportation and land use investments and policies, as well as demo- graphic and economic trends. Travel models produce quantitative information about travel demand and transportation system perfor- mance that can be used to evaluate alternatives and make informed decisions. 1.2 WHAT IS A TRAVEL MODEL? A travel model is an analysis tool that provides a systematic framework for representing how travel demand changes in response to differ- ent input assumptions. Travel models may take many different forms. Some travel models seek to comprehensively represent multiple, inter- related aspects of regional travel behavior, such as what activities people engage in, where and when these activities occur, and how people get to these activities. Other models are more limited in scope, addressing a smaller trans- portation market such as airport-related travel, travel within a corridor or a particular district of a city. The type of travel model that is ap- propriate to use is dependent on the particular questions being asked by decision makers. The following sections identify some broad types of models used in transportation planning, though it should be noted that these model types are not strictly defi ned and that there is a continuum of capability and detail across these model types. 1.2.1 Sketch-Planning Models Sketch-planning models are the simplest types of travel models. These tools are designed to produce rough estimates of travel demand 1 MOVING TO ACTIVITY-BASED TRAVEL DEMAND MODELS

6Part 1: ACTIVITY-BASED TRAVEL DEMAND MODELS: A PRIMER where order-of-magnitude information is all that is required. These models are typically simple and easy to implement, require less data, and often are implemented using common desktop software tools such as spreadsheets and geographic information systems (GISs). However, although these tools are less expen- sive to develop and apply, they may not provide the level of detail required to analyze certain types of policy and investments decisions, and may not provide detailed output infor mation. As a result, sketch-planning models may be appropriate for specific targeted analyses, but cannot inform large-scale longer-term policy and investment decision making. 1.2.2 Strategic-Planning Models Strategic-planning models are often narrow in scope but incorporate significant detail in spe- cific areas of analysis. These models often are used when there is a desire to analyze many scenarios quickly and implemented using basic software and hardware tools; these models are less expensive to develop and apply. Strategic- planning models are useful for testing a wide range of large-scale policy and investment alter natives but may be less appropriate for analyzing detailed project alternatives. 1.2.3 Trip-Based Models Trip-based travel models have evolved over many decades. As their name suggests, trip- based models use the individual person trip as the fundamental unit of analysis. Trip-based models are widely used in practice to support regional, subregional, and project-level trans- portation analysis and decision making. Trip- based models are often referred to as “4-step” models because they commonly include four primary components. The first trip genera- tion components estimate the numbers of trips produced by and attracted to each zone (these zones collectively represent the geography of the modeled area). The second trip distribution step connects where trips are produced and where they are attracted to. The third mode choice step determines the travel mode, such as automobile or transit, used for each trip, while the fourth assignment step predicts the specific network facilities or routes used for each trip. Table 1.1 displays key travel questions and answers for trip-based and activity-based models. 1.2.4 Activity-Based Models Activity-based models have become more widely used in practice. Activity-based models share some similarities to traditional 4-step models: activities are generated, destinations for the activities are identified, travel modes are determined, and the specific network facili- ties or routes used for each trip are predicted. However, activity-based models incorporate some significant advances over 4-step trip- based models, such as the explicit representa- tion of realistic constraints of time and space and the linkages among activities and travel, TABLE 1.1. KEY TRAVEL QUESTIONS AND ANSWERS Key Travel Questions Trip-Based Model Components Activity-Based Model Components What activities do people want to participate in? Trip generation Activity generation and scheduling Where are these activities? Trip distribution Tour and trip destination choice When are these activities? None Tour and trip time of day What travel mode is used? Trip mode choice Tour and trip mode choice What route is used? Network assignment Network assignment

7Chapter 1: MOVING TO ACTIVITY-BASED TRAVEL DEMAND MODELS for an individual person as well as across mul- tiple persons in a household. These linkages enable them to more realistically represent the effect of travel conditions on activity and travel choices. Activity-based models also have the ability to incorporate the influence of very detailed person-level and household-level at- tributes and the ability to produce detailed in- formation across a broader set of performance metrics. These capabilities are possible because activity-based models work at a disaggregate person-level rather than a more aggregate zone-level like most trip-based models. 1.3 HOW DO WE USE TRAVEL MODELS? Travel models are used to provide objec- tive assessm ents of the advantages and dis- advantages of different alternatives. These alternatives may include capital investments, policies, land use configurations, socio- economic and demographic assumptions, and many other factors. By running the travel model with different sets of input assumptions representing these alternatives, analysts can evaluate differences between alternatives using a broad range of metrics and can help answer decision makers’ key questions. 1.4 WHY USE AN ACTIVITY-BASED TRAVEL MODEL? Activity-based models can be used to evaluate alternative investments and policies that are difficult to test using traditional trip-based or sketch-planning models. For example, activity- based models often provide much more robust capabilities and sensitivities for evaluating pric- ing scenarios. Because activity-based models typically function at the level of individual persons and represent how these persons travel across the entire day, the model is more sensitive to pricing policies that may vary by time of day, which involve more complex tolling schemes. Another critical advantage of activity-based models is that they produce more detailed per- formance metrics, such as how travel benefits (or disbenefits) accrue to different populations, which can be used to support equity analyses. In addition, activity-based models can produce all of the trip-based model measures used to support regional planning, regional air quality, transit, and transportation demand manage- ment forecasting. Table 1.2 broadly summarizes and com- pares some of the key features of the continuous spectrum of model types from simple sketch- planning models to detailed activity-based models. Sketch-planning models often have lower levels of spatial and temporal resolution, while activity-based models often incorporate moderate to high levels of spatial resolution (such as parcels) and temporal resolution (such as half-hours). Activity-based models also incor- porate the highest levels of person and house- hold detail in this continuum of model types. Strategic-planning models and activity-based models can have moderate to high levels of policy sensitivities, although trip-based models TABLE 1.2. COMPARISON OF MODEL TYPES Model Type Spatial/ Temporal Detail Person/ Household Detail Policy Sensitivity Run Time Cost Sketch Planning Low Low Low Low Low Strategic Planning Low–Moderate Low–High Moderate–High Low Low Trip-Based Low–Moderate Moderate Moderate Moderate Moderate Activity-Based Moderate–High High Moderate–High Moderate Moderate

8Part 1: ACTIVITY-BASED TRAVEL DEMAND MODELS: A PRIMER and activity-based models generally have longer run times and greater development, application, and maintenance times and costs. 1.5 ACTIVITY-BASED TRAVEL MODELS 1.5.1 Activity-Based Travel Model Definition A fundamental premise of activity-based travel models is that travel demand derives from people’s needs and desires to participate in ac- tivities. In some cases these activities may occur within their homes, but in many cases these ac- tivities are located outside their homes, resulting in the need to travel. Activity-based models are based on behavioral theories about how people make decisions about activity participation in the presence of constraints, including decisions about where to participate in activities, when to participate in activities, and how to get to these activities. Because they represent decisions and the resulting behavior more realistically, activity-based models are often better at rep- resenting how investments, policies, or other changes will affect people’s travel behavior. Activity-based models are distinguished from trip-based models by a number of fea- tures. Activity-based models represent each person’s activity and travel choices across the entire day, considering the types of activities the individual needs to participate in and setting the priorities for scheduling these activities (such as prioritizing work activities over shopping ac- tivities). As any individual’s schedule becomes filled, the time available to participate in and travel to addi tional activities diminishes. 1.5.2 Deficiencies of Trip-Based Models 1.5.2.1 Independence Assumptions Transportation policy and investment questions have become more complex. Decision makers are no longer confronted only with questions about how and where to expand transporta- tion system capacity, but they also must con- sider questions about how to best manage the existing transportation system. Trip-based models are not able to provide information to address these policy questions because they assume that all trips are made independently. They do not recognize that the locations, travel modes, and timing of travel made by an indi- vidual are inter related. In addition, they lack details on individual travelers and their coordi- nation with other household members. For example, a region may wish to evalu- ate tolling alternatives that vary by time of day and price on a critical facility that is sig- nificantly congested during peak periods. Tolls are to be used to manage congestion by influ- encing travelers’ choices regarding when, and possibly how and where, they travel. Because trip-based models do not consider the entire tour, or series of linked trips made by an indi- vidual, they would not be sensitive to a toll on the return home from work. Similarly, because trip-based models usually do not include sen- sitivity to time of day and scheduling choices, they cannot show how a midday toll may re- sult in increased evening traffic or the impacts of many other policies with time-of-day char- acteristics. Because trip-based models rely on aggregations of persons and households, they are limited in their ability to represent how different people may respond to different toll changes, depending on their travel purpose, income, and other factors. Trip-based models typically have high numbers of nonhome-based trips, which do not include important informa- tion such as trip purpose, traveler income, or relation to other trips in the person’s day; this factor limits the sensitivity of trip-based models to many transportation policy and investment alternatives. In addition, trip-based models are often insensitive to how overall demand levels are influenced by the accessibility to opportu- nities for shopping, eating, and various social and recreational activities.

9Chapter 1: MOVING TO ACTIVITY-BASED TRAVEL DEMAND MODELS 1.5.2.2 Aggregation Bias Aggregation bias refers to the assumption that group characteristics are shared by all the indi- viduals who are members of that group. Trip- based models use aggregations of households that share the same attribute values to make forecasts, the idea being that all households of the same type behave similarly. For example, one such aggregation would be all households with two persons, one worker and one automo- bile, living within a particular transportation analysis zone. A trip-based model would pre- dict the number of daily trips of various types for these households using a common rate. There is tremendous diversity in how dif- ferent types of persons and households make travel decisions depending on factors such as income, transit accessibility, competition with other household members for vehicles, travel times by detailed time of day, and many other influences. The use of average values applied to aggregate populations across aggregate spa- tial zones and time periods distorts a model’s sensitivity to investment and policy alterna- tives. Although it may be theoretically pos- sible to incorporate additional detail in trip- based models through the use of additional market segmentation (such as including more household income categories), zones or time periods, it is practically challenging because the aggregate trip-based model’s reliance on two- dimensional origin–destination (O-D) or production– attraction matrices causes model run times, storage, and memory requirements to increase exponentially as segmentation in- creases. As a result, most trip-based models incorporate significant levels of aggregation, which compromises their sensitivities to dif- ferent alternatives and limits their ability to provide detailed information on the impacts of these alternatives, reducing their usefulness as decision-support tools. 1.5.3 Activity-Based Model Features 1.5.3.1 Individual Travelers Activity-based models provide a more intuitive, consistent, and behaviorally realistic represen- tation of travel than trip-based models. Rather than representing each trip as independent, for each individual traveler chains of trips (tours) are modeled as part of generating overall daily activity patterns. By functioning at the level of the individual traveler, activity-based models are able to represent greater variation across the population than aggregate trip-based models. The types of policies and investments that are of interest to decision makers change over time. A key advantage of activity-based models is that they can incorporate new explanatory variables and new sensitivities much more easily because they are typically implemented using a microsimulation framework, in which individ- ual person and household choices are evaluated. Microsimulating individuals and households imposes fewer limitations on the levels of tem- poral and spatial resolution used in the model and allows for the use of detailed individual- level information. For example, a region may wish to consider a new pricing alternative in which users pay only once on entering an area and can then leave and return without paying again. An activity-based model can be modified to represent this type of policy, which would be impossible in a trip-based model. Figure 1.1 shows a tour composed of three trips. 1.5.3.2 Interrelated Decision Making Activity-based models represent the inter- related aspects of activity and travel choices for all travel conducted by a person or household during a day, including purpose, location, tim- ing, and travel modes, which results in a more detailed representation of how travelers may respond to investment and policy alternatives, as well as land use and socioeconomic changes.

10 Part 1: ACTIVITY-BASED TRAVEL DEMAND MODELS: A PRIMER One of the primary means through which con- sistency is achieved is through the representa- tion of tours and trips. Activity-based models treat the tours and trips made by an individual as interrelated across the entire day. These rela- tionships are manifest in a number of ways. For example, activity-based models ex- plicitly represent how individuals move from one geographic location to another during the day—the destination of one trip becomes the origin for the subsequent trip. This geographic consistency realistically bounds where, how, and when travelers can travel. Consistency in the representation of time of day also distinguishes activity-based models from trip-based models. In activity-based models, activities and travel are usually scheduled within the context of the time constraints of a single day. Participation in different activities is determined among a variety of potential purposes, depending on individual- level attributes, such as worker status, as well as network travel times and accessibility. Figure 1.2 illustrates a scheduling example in which a mandatory work activity is sched- uled first, a shopping activity is scheduled next, and finally, an eating out activity is scheduled in the remaining available time window between the other activities. In an activity-based model, as more activities are scheduled, the amount of time available to participate in addi tional activi- ties become smaller. Arrival and departure times at destinations constrain when additional stops can be made, which also helps to ensure a coher- ent and consistent schedule for every traveler. Activity-based models also use information about tours and trips to impose plausible con- straints on the travel modes that are available to travelers. For example, it is highly unlikely that a traveler who has used transit to get to work is going to drive home alone, because he or she does not have a vehicle to use. The use of detailed traveler attributes when selecting destinations and the incorporation of realistic time–space constraints on destination choice also help to ensure consistency. Finally, the consideration of intra-household interactions, in which household members may allocate or coordinate activity participation in complex ways, results in greater internal consistency at both the person and household level. Figure 1.1. A tour with three trips. HOME WORK SHOP Traveler at home unl 8:30am, when departs for work (trip #1). Traveler arrives at work downtown at 9:00am. Stays unl 5:00pm, when departs for supermarket (trip #2). Traveler arrives at supermarket at 5:20pm, and shops unl 6:00pm, when departs for home (trip #3). 1 2 3

11 Chapter 1: MOVING TO ACTIVITY-BASED TRAVEL DEMAND MODELS 1.5.3.3 Detailed Information Activity-based models incorporate significantly more detailed input information and produce significantly more detailed outputs than trip- based models. By operating at the level of indi- vidual persons and households, activity-based models can use a wider range of important explanatory variables to predict travel pat- terns than trip-based models. Attempting to incorporate more detailed household or person characteristics in trip-based models usually re- quires many additional files (trip matrices) to represent different market segments, substan- tially lengthening model run times and increas- ing memory and storage requirements. Instead, activity-based models use lists of households, persons, tours, and trips, which is much more efficient, and includes more information. Con- sistent representation of trips made jointly by household members, which often comprise a significant portion of shared-ride trips, is only possible using activity-based models. Activity- based models also include explicit and detailed models of time-of-day choices, such as the times spent participating in activities, the arri- val times, and the departures times. The tem- poral information is especially critical given the travel demand and transportation system man- agement policy and investment choices faced by decision makers. Table 1.3 lists household and person attributes for activity-based models. Activity-based models also provide more detailed outputs, which allows analysts and decision makers to understand the impacts of alternatives on different communities. These 2014.11.18 C46 Primer FINAL for composition.docx 15 based models from trip-based models. In activity-based models, activities and travel are usually scheduled within the context of the time constraints of a single day. Participation in different activities is determined among a variety of potential purposes, depending on individual-level attributes, such as worker status, as well as network travel times and accessibility. Figure 1.2<FIG1.2> illustrates a scheduling example in which a mandatory work activity is scheduled first, a shopping activity is scheduled next, and finally, an eating out activity is scheduled in the remaining available time window between the other activities. In an activity- based model, as more activities are scheduled, the amount of time available to participate in additional activities become smaller. Arrival and departure times at destinations constrain when additional stops can be made, which also helps to ensure a coherent and consistent schedule for every traveler. [Insert Figure 1.2] [Caption] Figure 1.2. Example of scheduling priorities. 3am 6am 9am Noon 3pm 6pm 9pm Midnight 3am 3am 6am 9am Noon 3pm 6pm 9pm Midnight 3am Shop 3am 6am 9am Noon 3pm 6pm 9pm Midnight 3am Work Eat ShopWork Work 2) 1) 3) Figure 1.2. Example of scheduling priorities. TABLE 1.3. ACTIVITY-BASED MODEL HOUSEHOLD AND PERSON ATTRIBUTES Household Attributes Person Attributes • Number of persons • Housing tenure • Residential location • Number and age of family members • Household income • Number of vehicles owned • Number of workers • Number of students • Relationship to householder • Gender • Age • Grade in school • Hours worked per week • Worker status • Student status • Transit pass ownership • Subsidized parking at work

12 Part 1: ACTIVITY-BASED TRAVEL DEMAND MODELS: A PRIMER detailed outputs also allow insights into how changes such as the vehicle miles traveled or emissions can be attributed to individual house- holds. It is even possible for traveler benefits to be attributed to clusters of employment, which can be important for economic development analyses. Detailed output information also facilitates better interfaces with other analysis tools, such as traffic simulation models. How- ever, different levels of detail in the model outputs are associated with different levels of confidence, which is an important consider- ation when applying the model. 1.5.3.4 Integrated Travel Demand Model System Activity-based models, as well as trip-based models, are always embedded within an inte- grated model system in which there is an inter- action between the activity-based or trip-based models, which predict the demand for travel, and network models, which predict how this demand affects the performance of the trans- portation network supply. Most activity-based models are embedded within a basic integrated model system that incorporates a limited num- ber of essential components (Figure 1.3): • Population synthesis models create de- tailed, synthetic representations of popu- lations of individuals within households (agents) whose choices are simulated in activity-based models. This population is based on information produced by regional economic models, land use models, and demographic models. • Activity-based travel demand models pre- dict the long-term choices (such as work location and automobile ownership) and the daily activity patterns of a given synthetic population, including activity purposes, loca tions, timing, and modes of access. These estimates of travel demand can be used to help evaluate alternative transpor- tation, land use, and other scenarios. • Auxiliary models provide information about truck and commercial travel, as well as special purpose travel such as trips to and from airports or travel made by visitors. The travel demand represented in auxiliary models complements the personal travel generated by the activity-based model. • Network supply models are tightly linked with activity-based demand models. The flows of travel by time of day and mode predicted by activity-based travel demand models and auxiliary models are assigned to roadway, transit, and other networks to generate estimates of volumes and travel times. Measures of impedance output from network supply models are usually used as input to activity-based models and other integrated model components. 1.5.4 Development Considerations 1.5.4.1 Data The data required to develop and apply an activity-based model are not significantly dif- ferent from the data required to develop a trip-Figure 1.3. Basic integrated model components. 2014.11.18 C46 Primer FINAL for composition.docx 19 of volumes and travel times. Measures of impedance output from network supply models are usually used as input to activity-based models and other integrated model components.</BL> [Insert Figure 1.3] [Caption] Figure 1.3. Basic integrat d odel components. <H2>1.5.4 Development Considerations <H3>1.5.4.1 Data The data required to develop and apply an activity-based model are not significantly different from the data required to develop a trip-based model. The primary data used to develop and apply both activity-based and trip-based models include household travel survey information, economic and demographic information about the spatial distribution of employment and households, and representations of transportation networks. Household travel surveys contain Population Synthesis Model Activity-Based Model Auxiliary Model Network Supply Model

13 Chapter 1: MOVING TO ACTIVITY-BASED TRAVEL DEMAND MODELS based model. The primary data used to develop and apply both activity-based and trip-based models include household travel survey infor- mation, economic and demographic informa- tion about the spatial distribution of employ- ment and households, and representations of transportation networks. Household travel surveys contain detailed information about whether, where, how, and when individuals and households travel. The same household surveys used to develop trip-based models can be used to develop activity-based models, although such surveys are subjected to much more scrutiny in developing an activity-based model because it is important that all the sur- vey data be consistent internally across all the individuals in each household. Data describing the spatial distribution of different types of households, persons, and em- ployment by sector are required to develop and apply both activity-based and trip-based model systems and are largely consistent between the two types of models. If more detailed in- formation is available, such as employment by detailed sector or households by detailed size category, this can be used more easily within the activity-based model framework. Activity- based models do require the development of one additional type of input, a “synthetic pop- ulation” that represents a region’s travelers and their detailed attributes. However, this input can be prepared using readily available data and tools. Transportation network data requirements are very similar for activity-based and trip- based models. Most activity-based models used in practice are linked to traditional roadway and transit network components that are very similar to those used for trip-based models. However, because activity-based models in- clude more detailed representations of time of day, they often include networks with more time-period–specific information. 1.5.4.2 Staff and Consultant Requirements Activity-based models impose some different requirements related to staff knowledge and skills. Because they incorporate more detailed representations of travel choices, it is necessary that modeling staff have a good understanding of the activity-based modeling process and its statistical modeling methods. Often, the most effective means of developing this understand- ing is through on-the-job experience. Modeling staff also must have additional data manage- ment and statistical skills in order to be able to meaningfully summarize and understand the new detailed outputs produced by activity- based models. There is some additional burden associated with managing additional networks by time of day and with preparing synthetic population inputs. Most agencies have relied on consultants to develop, implement, and en- hance activity-based models, but often, they rely on their own staff to perform model runs and analysis. 1.5.4.3 Costs and Schedule The costs for developing activity-based models have significantly decreased to the point where recent activity-based model development efforts cost approximately the same as traditional trip- based model development efforts. The devel op- ment of the first activity-based models was more costly as a result of the initial effort to establish methodologies and create implementation soft- ware. Similarly, if an agency wants to imple- ment new or enhanced features or software, the costs can be higher; but, when existing method- ologies and software are used, the costs are no longer as great. As with trip-based models, the calibration of activity-based models can make costs and schedules uncertain, because differ- ent model applications may require different levels of effort. Some additional costs may be associated with computer hardware, because

14 Part 1: ACTIVITY-BASED TRAVEL DEMAND MODELS: A PRIMER activity-based model software is typically de- signed to employ distributed computing across multiple processors. 1.5.4.4 Model Run Times Activity-based model system run times are de- pendent primarily on the size of the population of the region being simulated, the number of zones and time periods for which the network supply models are run, and the amount of com- puting resources available. For the activity- based component of the model system, there is essentially a linear relationship between the size of the population and run times— simulating a population twice as big will take approxi- mately twice as long. The network model run times increase with the square of the number of zones, and linearly with the number of as- signment time periods. The relationship with run times and computing resources is more nuanced. Activity-based modeling software is designed to be distributed across multiple com- puter processors to reduce run times. As more processors are available run times are reduced, although the performance gain associated with each additional processor diminishes due to computational overhead associated with man- aging and exchanging data. Finally, model run times are also influenced by the complexity of the model design, the number of alternatives included, and feedback and the level of con- vergence required. Both trip-based model and activity-based model run times vary greatly, and in both cases, assigning the demand to the network is the most time-consuming part of model runs. 1.5.4.5 Stakeholder Acceptance The information produced by travel models is consumed by a variety of users for a broad range of purposes. Essential to the success of activity-based models is the acceptance by these stakeholders of the usefulness of the model out- puts. Such acceptance is predicated on the clear communication of the purposes and structure of the model, in order to address the concern frequently expressed about both trip-based and activity-based models being black boxes. However, acceptance can be firmly established only by demonstrating the model’s explanatory power and reasonable results when applied to actual alternative analyses, or through system- atic backcasting exercises. 1.5.5 Integration with Other Models While activity-based model systems represent a comprehensive approach to representing the activity and travel choices, they depend on rel- evant information that is often produced by other related models. For example, an activity- based model needs information about the loca- tion of future employment. Such information is sometimes produced by a separate land use model. A model system in which various dif- ferent components interact and exchange in- formation is often referred to as an integrated model. Integrated models provide the oppor- tunity to represent various diverse, dynamic, and interrelated aspects transportation system performance and land use. Although activity- based models are always embedded within a basic integrated model system comprising travel demand and supply components as de- scribed earlier, they can also be integrated with other models to create an extended integrated model system (Figure 1.4) that may include the following: • Regional economic models can provide overall estimates of regional employment or productivity by industrial sector based on the competitiveness of the region rela- tive to other regions and can also provide information on overall changes in regional households and population. This informa- tion provides regional controls for activity- based model inputs.

15 Chapter 1: MOVING TO ACTIVITY-BASED TRAVEL DEMAND MODELS • Land use models typically provide more de- tailed information about subregional land use development and redevelopment, the economy, employment, and population. There is a broad spectrum of approaches to modeling land use and the land use– transportation interaction, and the outputs from these models may be aggregated, disag- gregated, and transformed in order to pro- vide direct inputs to an activity-based model. • Demographic models show how the popu- lation is expected to change over time. Changes in demographics are manifest in dif- ferent transportation and land use choices, and activity-based models incorporate many demographic explanatory variables. • Air pollutant emissions models produce esti- mates of criteria pollutants and greenhouse gases based on network performance mea- sures derived from network supply models and vehicle performance assumptions. 1.5.6 Application Stories 1.5.6.1 Congestion Pricing Congestion pricing can encompass a wide vari- ety of different pricing and toll schemes that may be intended to manage demand, improve travel time reliability, reduce congestion, and increase usage of alternative modes. Activity- based models provide more flexibility than trip-based models in their ability to represent different alternatives. Activity-based models have been used in both New York and San Francisco to evaluate congestion pricing op- tions. In New York, an activity-based model was used to evaluate scenarios in which tolls were imposed for vehicles traveling to certain portions of Manhattan. In San Francisco, an activity-based model was used to evaluate area-based and cordon-based pricing scenarios. The activity-based models provided the ability to represent how different travelers respond differently to alternatives, and to provide a consistent representation of how the different alternatives impact the timing, destinations, modes, and even the generation of tours and trips. Because the models explicitly represent individual persons and households, the models could be used to evaluate a variety of alterna- tive schemes, including discounts offered to dif- ferent groups to address equity concerns. 1.5.6.2 Environment, Climate Change, and Air Quality A critical concern in many regions is how urban form influences trip making, and the envi ronmental impacts of alternative transpor- tation and land use configurations. In order to have an analysis tool that is more sensitive to these transportation–land use interactions, the Sacramento Area Council of Governments (SACOG) developed an activity-based model system that uses very fine-grained geography of Figure 1.4. Basic integrated model components.2014.11.18 C46 Primer FINAL for composition.docx 24 broad spectrum of approaches to modeling land use and the land use–transportation interaction, and the outputs from these models may be aggregated, disaggregated, and transformed in order to provide direct inputs to an activity-based model. • Demographic models show how the population is expected to change over time. Changes in demographics are manifest in different transportation and land use choices, and activity- based models incorporate many demographic explanatory variables. • Air pollutant emissions models produce estimates of criteria pollutants and greenhouse gases based on network performance measures derived from network supply models and vehicle performance assumptions.</BL> [Insert Figure 1.4] [Caption] Population Synthesis Model Activity-Based Model Auxiliary Model Network Supply Model AQ Emissions Model Demographic Model Regional Economic Model Land Use Model

16 Part 1: ACTIVITY-BASED TRAVEL DEMAND MODELS: A PRIMER individual parcels and has applied the model to a variety of innovative analyses. For ex- ample, to analyze the impacts of a large-scale residential development that was designed to produce more pedes trian, transit, and other short- distance trips, the activity-based model was used to compare the travel behavior of tens of thousands of residents in two different scenarios: one in which these residents lived in the proposed development and an alterna- tive in which these exact same residents lived in more typical suburban developments. The activity-based model demonstrated that the total vehicle miles traveled by these residents was significantly lower when they lived in the new development than when these same resi- dents lived in more typical developments. In addition to using the activity-based model to perform innovative transportation and land use scenarios, SACOG has used their activity- based model as the basis for air quality confor- mity analyses, as have other regional planning agencies, such as the Puget Sound Regional Council in Seattle. 1.5.6.3 Regional Transportation Plans Activity-based models have been used by MPOs to support the analysis of regional transportation plans. For example, the Metro- politan Transportation Commission, in the San Francisco Bay Area, recently and compre- hensively applied their activity-based model to assess the impacts of individual projects as well as plan scenarios. Approximately 100 indi- vidual model runs were performed in order to develop projects-level performance mea- sures. Outputs from the activity-based model were then used to support detailed analyses: a benefit–cost analysis as well as others. The benefit–cost analysis relied on detailed infor- mation produced by the activity-based model, such as the amount of transportation-related physical activity engaged in by each regional resident.