National Academies Press: OpenBook

Integrated Transportation and Land Use Models (2018)

Chapter: Chapter 2 - Principles for Integration of Land Use and Transport Models

« Previous: Chapter 1 - Introduction
Page 13
Suggested Citation:"Chapter 2 - Principles for Integration of Land Use and Transport Models." National Academies of Sciences, Engineering, and Medicine. 2018. Integrated Transportation and Land Use Models. Washington, DC: The National Academies Press. doi: 10.17226/25194.
×
Page 13
Page 14
Suggested Citation:"Chapter 2 - Principles for Integration of Land Use and Transport Models." National Academies of Sciences, Engineering, and Medicine. 2018. Integrated Transportation and Land Use Models. Washington, DC: The National Academies Press. doi: 10.17226/25194.
×
Page 14
Page 15
Suggested Citation:"Chapter 2 - Principles for Integration of Land Use and Transport Models." National Academies of Sciences, Engineering, and Medicine. 2018. Integrated Transportation and Land Use Models. Washington, DC: The National Academies Press. doi: 10.17226/25194.
×
Page 15
Page 16
Suggested Citation:"Chapter 2 - Principles for Integration of Land Use and Transport Models." National Academies of Sciences, Engineering, and Medicine. 2018. Integrated Transportation and Land Use Models. Washington, DC: The National Academies Press. doi: 10.17226/25194.
×
Page 16
Page 17
Suggested Citation:"Chapter 2 - Principles for Integration of Land Use and Transport Models." National Academies of Sciences, Engineering, and Medicine. 2018. Integrated Transportation and Land Use Models. Washington, DC: The National Academies Press. doi: 10.17226/25194.
×
Page 17
Page 18
Suggested Citation:"Chapter 2 - Principles for Integration of Land Use and Transport Models." National Academies of Sciences, Engineering, and Medicine. 2018. Integrated Transportation and Land Use Models. Washington, DC: The National Academies Press. doi: 10.17226/25194.
×
Page 18
Page 19
Suggested Citation:"Chapter 2 - Principles for Integration of Land Use and Transport Models." National Academies of Sciences, Engineering, and Medicine. 2018. Integrated Transportation and Land Use Models. Washington, DC: The National Academies Press. doi: 10.17226/25194.
×
Page 19
Page 20
Suggested Citation:"Chapter 2 - Principles for Integration of Land Use and Transport Models." National Academies of Sciences, Engineering, and Medicine. 2018. Integrated Transportation and Land Use Models. Washington, DC: The National Academies Press. doi: 10.17226/25194.
×
Page 20

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.

13 Since Ira Lowry developed his Model of Metropolis in the early 1960s, the integration of land use and transport has essentially remained unchanged. The land use model provides the location of population and employment used in the transport model to generate travel demand. Conversely, the transport model provides zone-to-zone travel times used in the land use model to calculate accessibilities that affect location choice of population and employment. Locations with higher accessibility are assumed to be—with other conditions remaining the same—more desirable than locations with lower accessibility. This two-way interaction between land use and transport has been visualized in the land use/transport feedback cycle shown in Figure 2-1. Although details have improved over the last 50 years, the land use/transport integration methods largely remained unchanged in practical applications. Some integrated land use/transport models have implemented one-way feedback only. Sometimes, updated travel times are fed into the land use model, but the land use model results are not used in the transport model. This commonly is implemented when there is no consensus on the validity of land use model results. In other cases, updated socioeconomic data are fed from the land use model to the transport model, but accessibilities are not fed back to the transport model. Such model design has been chosen when the transport model has very long runtimes or when changes in accessibilities over time are expected to be so minor that it is not worth the effort to fully integrate the models. Most models described in this report, however, take advantage of the full integration and feed data between land use and transport models in both directions. 2.1 Accessibilities Accessibilities describe, for every origin zone, how many destinations can be reached how easily. Destinations of interest commonly relate to residences and employment, but accessibilities can also be calculated to schools, parks, retail facilities, and others. If the number of destinations is growing in a vicinity, accessibility increases. Similarly, if it becomes easier to travel to those destinations (maybe because a new road has reduced travel times), accessibility increases as well. Accessibilities can be calculated using travel time for any mode of interest, such as auto, transit, walking, and air. In land use models, accessibilities are applied to influence location choice decisions, because both households and firms tend to prefer locations with higher accessibilities. The simplest form of accessibility is a count of destinations within a certain travel time. For a given location, one might ask how many jobs can be accessed by car within 30 minutes. Using GIS software, this accessibility is easy to calculate. However, this method does not distinguish between employment that is 5 minutes away and employment that is 25 minutes away. Every job within 30 minutes is given the same weight. Figure 2-2 shows an example of the number of jobs that can be reached within 30 minutes by transit for the Washington, DC, metropolitan area. C H A P T E R 2 Principles for Integration of Land Use and Transport Models

14 Integrated Transportation and Land Use Models Figure 2-1. Land use/transport feedback cycle (Wegener and Fürst 1999). Figure 2-2. Number of jobs accessible within a 30-minute transit ride for the Washington, DC, region (Source: US EPA, Office of Policy, Smart Location Database, https://www.epa.gov/smartgrowth/smart- location-mapping).

Principles for Integration of Land Use and Transport Models 15 Hansen (1959) proved empirically a high correlation between accessibility and urban devel- opment. Although his study focused on residential development, the principal idea of his approach is valid for businesses as well. First, he distributed the total metropolitan growth in the Washington, DC, area to zones according to their share of vacant developable land. This “probable development” most likely differs from the actual development. Hansen calculated the ratio of actual development over probable development and showed that this ratio highly correlated with accessibility. Zones with better accessibility had a higher share of residential development than originally anticipated, and zones with lower accessibility had a smaller share of development. Hansen defined accessibility of zone i as being directly proportional to the size of activities in all zones and inversely proportional to the distance between those zones and zone i. Equation 2-1A w f ci j ij J i∑ ( )= α where Ai = Accessibility in zone i wj = Number of activities in zone j α = Parameter to emphasize or reduce the weight of denser areas f(cij) = Function describing the impedance to travel from zone i to zone j Often, the function f(cij) is expressed as an exponential function: ( ) ( )= βexp Equation 2-2f c tij iji where β = Parameter to set relevance of impedance (β < 0) tij = Impedance from zone i to j, commonly defined as travel time or travel distance The accessibility defined above is called potential accessibility or, in honor of the inventor, Hansen accessibility. Figure 2-3 shows the potential accessibility of population by car travel time for the Washington-Baltimore region. The accessibility map reflects the population density in the area and accounts for faster travel times along major highway routes, such as I-95 between the two cities and I-270 leaving Washington in a northwestern direction toward Frederick, MD. The potential accessibility was enhanced by Wilson (1967) in a similar but behaviorally richer approach to estimate accessibility: exp Equation 2-3A X Y O D ci i j i j ij J i i i i i∑ ( )= β with exp 1 X Y D ci j j ij J i i i∑( )( )= β − and ∑( )( )= β −exp Equation 2-41Y X O cj i i ij i i i i

16 Integrated Transportation and Land Use Models where Oi = Number of activities with origin i Dj = Number of activities in destination j Xi and Yj = Terms describing the interaction between zone i and zone j A variant of this is accessibility is the logsum term, which can be seen as a distance-digressive averaging, but also has an economic interpretation as the expected utility of (living at) location i under certain conditions (Ben-Akiva and Lerman 1985, Train 2009). The main benefit over simple accessibility measures is that the logsum may include travel time by various modes (such as auto travel time and transit travel time), weighted by their corresponding utility for a given origin-destination pair: 1 ln exp Equation 2-5,L w ci j i j J i i i∑ ( )= µ −µ α where Li = Logsum of zone i µ = Logsum parameter Accessibilities can be defined specifically for the task at hand. For example, to calculate walk- and-bike accessibilities, the logsum function can be limited to those two modes. Instead of travel time, generalized costs that include both travel time and travel costs may be used. If accessibility to schools is needed, schools by size can be used as the destination variable wj. Today, a great variety of different accessibility indicators are used in land use models, trip generation models, and spatial analyses. A comprehensive list of accessibilities can be found in Geurs and van Wee (2004) and Schürmann, Spiekermann, and Wegener (1997). Figure 2-3. Potential accessibility in the Washington/ Baltimore region.

Principles for Integration of Land Use and Transport Models 17 2.2 Frequency of Interaction Integrated land use/transport models commonly are used to predict the effect of policies and scenarios. The ideal temporal integration is visualized in Figure 2-4. The model starts with socioeconomic data for a base year, say 2000, which are fed into the transport model. Updated impedances in form or travel time, distances, and/or costs are fed from the transport model to the land use model. The land use model is used to update socioeconomic data from one year to the next, and the updated socioeconomic data is sent to the transport model to model the following year, and so on. This full integration is desirable, yet rarely achieved. Model runtimes usually prohibit such a tight temporal integration. Alternatively, the transport model or the land use model or both are run for selected years only. For example, the UrbanSim land use model commonly models every year, the IRPUD model commonly models every 3rd year, and, for PECAS implementations, it is common to model every 10th year. The less frequently the land use model is run, the more abrupt changes in population and employment will be. Similarly, the frequency of the transport model to some degree depends on the runtime of the transport model. The IRPUD model models the transport system every 3 years, the PECAS implementation in Ohio runs the transport model every 5 years, and the UrbanSim application in the San Francisco Bay Area runs the transport model every 10 years. For sketch planning models, it is uncommon to represent this iterative feedback between land use and feedback. There is no generally accepted solution of how often the land use and transport models need to run for reasonable model results. The user has to strike a balance between changes in model output that are reasonably smooth and model runtimes. The transport model in the Atlanta Metropolitan Area, for example, runs for 24 hours. With such runtimes, it is not feasible to run the model every simulation year. On the other hand, travel times, distances, and costs are unlikely to change dramatically from one year to the next, making it very reasonable to model selected years only. Users need to remember, however, that more draconian scenarios will require more frequent interaction between land use and transport to reasonably respond to the scenario. For example, if a scenario tested increasing transportation costs dramatically, population would react by changing travel behavior and location choice, which would change travel times and costs. Several iterations between the land use model and the transport model are needed to fully assess the scenario. Land use and transport are not assumed to be in perfect equilibrium. One could feed back data between the land use and transport model several times within the same simulation period. TM 2000 LUM 2000 2001 TM 2001 LUM 2001 2002 TM 2002 LUM 2039 2040 TM 2040 SED 2000 Zonal data: Population and employment Travel time, distance and cost matrices SED: Socio-Economic Data TM: Transport Model LUM: Land Use Model Figure 2-4. Temporal integration of land use and transport models.

18 Integrated Transportation and Land Use Models In reality, however, land use changes slowly. Although the transport system is near an equilib- rium, because travelers can change destinations, modes, time of day, and paths within minutes, the land use system tends to “lag behind” the transport system. If travel times change, house- holds do not move immediately because the costs of relocation are high (both financially and emotionally). Some models even introduce an additional delay, such as the demand for housing in time period t is added to the market not before the time period t + 1 (Wegener 1998). This delay accounts for the time it takes to recognize demand, plan a housing project, get the building permit, and construct the project. 2.3 Levels of Integration There are different levels of model integration, ranging from separate models with manual data transfer to tightly integrated models that operate as a single software. Model integration faces some scientific as well as technical challenges (Argent 2004; Belete, Voinov, and Holst 2014; van Delden et al. 2011). The main scientific challenges relate to dealing with different domains, paradigms, assumptions, scales, and spatial and temporal resolutions used by different models. Technical challenges include implementing the software integration of the models, providing dynamic feedback loops, managing data exchange and storage, and developing relevant user interfaces (Lam et al. 2004). Integration approaches do not only refer to land use and transport models, but to any other model that the user may want to connect to the modeling suite. Common examples include economic models that predict the growth of the economy, health models that assess the effect on human health, emissions models that quantify gaseous emissions (such as CO2, NOx, or PM) and noise, or freight models that predict goods transport. The integration is especially difficult if the models are developed independently without any built-in methods for linking to other models (Shahumyan and Moeckel 2017). Moreover, the models often are developed in different programming languages and software environments and may have various licensing restrictions. The task becomes even more complicated with proprietary models or when the access to a model’s source code is limited. Shahumyan and Moeckel (2017) defined key requirements for the model integration: • Ability to develop models independently. • A modular approach supporting reusability of components and adding new components. • Minimal or no change in source codes of each model. • Capacity to link models developed in different programming languages. • Ability to deal with different licensing requirements. • Avoidance of all manual data transfer (because it slows down model runs and is error-prone). • User-friendly graphical interface. • Compatibility with GIS for easy data visualization and spatial analysis. • Adequate runtime. • Minimal costs and efficient timing for implementation. The general definition of model coupling implies that originally independent model processes interact. The implementation of such interaction, however, can vary, depending on the project goals, available resources, the models’ specifics, requirements, and limitations. For the cou- pling of environmental models, Brandmeyer and Karimi (2000) developed a five-level coupling hierarchy, of manual data transfer, loose coupling, shared coupling, joined coupling, and tool coupling.

Principles for Integration of Land Use and Transport Models 19 The manual data transfer method, the most basic level of model coupling, includes manual extraction, transfer, and conversion of output produced by one model for use as an input by other models. Although this approach requires minimal initial cost and time to use, it is not convenient when multiple runs and frequent data exchange are required (Brandmeyer and Karimi 2000). Manual data transfer is error-prone and it may be impossible to determine at a later point if incorrect data were transferred. The data exchange between models is automated in loose coupling. Models, although, still work independently and the user interacts with each model separately (Wong et al. 2009). Loose coupling also has a low initial cost, requires minimal changes to existing codes, and allows the models to still be developed independently. However, if data structures change in any of the linked models, data conversion mechanisms between the affected models require particular attention. In shared coupling, the models either share a user interface or the data storage. For the first approach, a single user-friendly interface hides the internal coupling method, making it less confusing (Berry, Buckley, and McGarigal 1997). In data coupling, the models are kept separate, but share the data storage (van Walsum and Veldhuizen 2011). Shared user-interface coupling supports proprietary models and reduces the time the user needs to interact with the model. However, a user-interface update often is required when models are updated. Data coupling makes data maintenance simpler. However, the overall model interface and performance depend on the Database Management System used (Brandmeyer and Karimi 2000). Joined coupling employs both the common user interface and data storage and may use two structurally different approaches: embedded coupling, where one model contains another (Liu et al. 2014); and integrated coupling, where each model is a peer of every other model (Sudicky et al., 2003). Although joined coupling reduces the development costs and promotes code reusability, it requires access to the models’ source codes and a single operating system. For tool coupling, the models are coupled using a modeling framework (Babendreier and Castleton 2005, Moore and Tindall 2005). This supports multiple developers working on models simultaneously and can be used with both legacy and new models. The order of running the models and the data feedback frequencies should drive the model coupling choice. The sequential coupling scheme provides the weakest form of the integration, where the first model runs the required time step/period and provides the output to the second model, which only runs after getting the results of the first model (van Walsum and Veldhuizen 2011). This scheme is often used for manual data transfer or loose model coupling. A drawback of such an approach is that the stability between the two linked models is determined by the model that gets updated first, which can lead to inconsistencies for the second model. In con- trast, the fully coupled scheme supports the full feedback between models within each time step. However, full coupling usually requires code modification to organize such feedback. Moreover, it may result in iterations within iterations and increase its computational load, essentially reducing the overall efficiency (van Walsum and Veldhuizen 2011). Each of the described approaches has advantages and disadvantages (Brandmeyer and Karimi 2000, Droppo et al. 2010) and the selection of the method mainly depends on the model require- ments, research goals, and available resources. The level of model integration should depend on the direction of information exchange and frequency of data flows. Model direction refers to the sender model and receiver model of information. For example, an economic model is used to provide regional control totals of population and employment growth for the entire study area. Although the land use model allocates this growth to individual

20 Integrated Transportation and Land Use Models zones, the overall growth is provided exogenously by a national economic input/output model. In theory, the performance of this study area could be fed back into the economic model, as, for example, tighter land use restrictions could push some growth to neighboring regions. In reality, however, the effects of scenarios for a local area on the national economy are minimal. Given that economic growth is used as a one-way flow of information, the integration between the economic model and the land use model could be kept off line (loose coupling) and solved with a single file transfer covering a 40-year growth forecast. Another aspect of model integration is the frequency of interaction. For example, mobile emissions are calculated every time after the transportation model runs. This is a one-way flow of information: transportation generates emissions and emissions (commonly) do not affect travel behavior. However, although information flows one way, information exchange is fre- quent enough that the transportation model and the mobile emissions model warrant closer integration. Frequent data flows deserve closer integration to ease information flow, even if the flow is only one way. The tightest integration should be pursued for models that exchange information bi- directionally and frequently. This level of integration often applies to the integration of land use and transportation models. These models exchange information in both directions: the loca- tions of households and employment define the origins and destinations in the transportation model, and travel times are converted into accessibilities that affect household relocation deci- sions. Given the frequency of this bi-directional flow, these two models deserve the most atten- tion in terms of data exchange methods to ensure information exchange with little translation loss and limited effect on model runtime. But, more integration is not always better. Using the appropriate level of integration improves model stability and runtimes without compromising important links between models.

Next: Chapter 3 - Screening Survey »
Integrated Transportation and Land Use Models Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB's National Cooperative Highway Research Program (NCHRP) Synthesis 520: Integrated Transportation and Land Use Models presents information on how select agencies are using sketch planning models and advanced behavioral models to support decision making. The synthesis describes the performance of these models and the basic principles of land use/transport integration.

  1. ×

    Welcome to OpenBook!

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

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

    No Thanks Take a Tour »
  2. ×

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

    « Back Next »
  3. ×

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

    « Back Next »
  4. ×

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

    « Back Next »
  5. ×

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

    « Back Next »
  6. ×

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

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

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

    « Back Next »
Stay Connected!