Integrated Transportation and Land Use Models (2018)

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.

1 Integrated land use/transport models are used to represent the land use/transport feed­ back cycle in a model. Everything else being equal, more accessible neighborhoods are more desirable, in terms of location choice, than less accessible neighborhoods. Location choice and transportation are interdependent. For example, if more households and/or firms decide to locate in the suburbs, the origins and destinations of trips change, which affects levels of congestion. That land use and transport are interdependent is not new, but many trans­ portation models ignore this link to land use changes. Integrated land use/transport models improve the reasonability of results compared to stand­alone transport models and also offer the opportunity to analyze additional scenarios. Possible effects from alternative zoning scenarios, denser development near transit stations, or affordable housing programs can be tested regarding their influence on population and employment distributions and their effects on resulting travel demand. This report presents the results of investigation into how these models have performed and how an integrated land use/transport model has affected decision making in the region. For this report, the research team has identified different types of land use models that can be integrated with transport—sketch planning models and advanced behavioral models (specifically, microsimulation discrete choice models and spatial input­output models) (see Figure S­1). These are discussed below. Sketch planning models are rule­based and excellent for long­term visioning, because they tend to run faster and allow users to assess the development capacity of considered land use scenarios. Such models do not model human behavior explicitly but develop rules for development interactions. Examples of sketch planning models include CommunityViz, Envision Tomorrow, and UPlan. Behavioral models, on the other hand, try to represent human behavior based on behav­ ioral theory at the expense of (commonly) longer runtimes and larger data requirements. Such advanced models can be further categorized into microsimulation discrete choice models and spatial input­output models. Both behavioral model categories use advanced theory to explain human behavior (the approaches differ in model implementation). Micro­ simulation discrete choice models are based on behavioral discrete choice theory, while spatial input­output models are derived from economic theories. Survey results indicate that the most used microsimulation discrete choice model in the United States and Canada is UrbanSim; other examples (primarily used in academia) include ILUTE, IRPUD, and Silo. The most widely applied spatial input­output models in the United States and Canada are CUBE Land, MEPLAN, PECAS, and TRANUS. This report describes the basic principles of land use/transport integration. In the most common approach, the land use model provides zonal population and employment to the S U M M A R Y Integrated Transportation and Land Use Models

2 Integrated Transportation and Land Use Models transport model, while the transport model feeds back zonal accessibilities that affect loca­ tion choice in the land use model. Some models use distance to transit stations or highway interchanges as additional location factors; others use mode choice logsums to account for accessibilities by different modes. Although the transport model represents travel demand and traffic volumes for an average day in a given year, the land use model needs to model the change in land use from one period to the next. Therefore, the land use and transport models need to run alternately. Many land use models run in single­year increments; other land use models run in 5­ or 10­year increments. Sketch planning land use models usually run a base year and one future year, and intermediate years may be interpolated. Some land use models have been tightly integrated with transport models, which means they operate within one programming environment, access one common database, and provide a single user interface. Most integrated land use/transport models, however, are loosely coupled—individual stand­alone models run sequentially (that is, one model starts, reads in all data, processes data, writes out results, is closed, and the next model starts, reads in data, processes data, etc.). Screening Survey A screening survey was conducted on line to understand which integrated land use/ transport models are most commonly used in the United States and Canada. Results from 63 completed surveys showed 37% of all agencies do not operate a land use model and do not plan to do so in the near future. Almost 60%, however, responded that they either use an integrated land use/transport model or are planning to implement one in the near future. Although this survey is likely to be biased toward agencies that have thought about integrated land use/transport modeling, the results indicate that such models are widely used and relevant for transport and planning agencies in the United States and Canada. The most frequently used models according to the survey are CUBE Land (spatial input­ output) and CommunityViz (sketch planning), followed by UrbanSim (microsimulation discrete choice), PECAS (spatial input­output), and Envision Tomorrow (sketch planning) (see Figure S­2.) Although this survey is not representative of models used in the United States and Canada, the survey showed a large diversity in models applied. This report is intended to help decisionmakers to decide which model best fits their particular needs. The screening survey also showed that these integrated land use/transport models are predominately used for road and transit infrastructure scenarios, or traditional transportation scenarios. Of the survey respondents, 5 agencies have applied zoning scenarios, but another 18 agencies plan to use their modeling suite for this purpose in the future. Figure S­3 provides additional screening survey information. Sketch Planning Models Rule-Based Models Behavioral Models Microsimulation Discrete Choice Models Spatial Input- Output Models Figure S-1. Classification of land use models.

Summary 3 Figure S-2. Land use models used in practice according to the screening survey. 3 3 4 4 5 8 14 15 22 11 17 18 23 25 Equity analysis Freight scenarios Pricing studies Environmental impact analysis Zoning scenarios Transit infrastructure scenarios Road infrastructure scenarios Current scenario Planned scenario Figure S-3. Most frequently tested scenarios with integrated land use/transport models. In-depth Interviews In­depth interviews were conducted with seven U.S. agencies that use integrated land use/ transport models (see Figure S­4). The surveys showed that many agencies operate sophis­ ticated models that have influenced decisionmakers. The interviews showed that model results are treated as “an additional voice at the table.” Although final judgment continues to be made by decisionmakers, model results are used as additional evidence as to whether or not a proposed policy has the intended effects, and whether or not the magnitude of the effect warrants the effort of implementing a policy. Agencies that use such models have reported that confidence in model results generally was increased because the land use/ transport feedback cycle was represented explicitly.

4 Integrated Transportation and Land Use Models Based on these in­depth interviews, the researchers identified nine key criteria to use when selecting a land use model for integration with a transport model: 1. Open-source software. Agencies mentioned this as a key criterion to ensure that staff members can review code to fully understand the model formulation. Some agencies also wanted to be able to modify the code. 2. The level of model developer dependence. Each agency interviewed had a contract with the model developer or someone who had worked closely with the model developer. There was general agreement that having access to the model developer, or someone else who knows the model inside­out, is a key factor to successfully implementing the model. 3. Modeling expertise within the agency. The more qualified the staff was in modeling, the less outside help for model implementation and model maintenance was required. 4. Programming expertise. Although implementing a land use model does not necessarily require staff members to understand the programming code, interviewees mentioned that being able to read and, if necessary, modify the code in house was helpful. The programming or scripting language staff members are familiar with could therefore influence the choice of a model. 5. Runtime of the software. If many scenarios are to be analyzed, the runtime of the integrated model becomes a key factor. If only one future scenario will be modeled to represent the land use/transport feedback cycle without any interest in modeling alter­ native scenarios, runtime is less of an issue. However, even if only a few scenarios will be modeled, runtime is a key parameter for calibrating a model. Given that many land use models run fairly fast nowadays (such as finishing all years to be modeled in an overnight run), the bottleneck for many agencies would be transport models that run too long to be efficiently integrated with a land use model. 6. Advice from an expert panel. Expert Panels may consist of technical experts, domain experts, and stakeholders in the region, such as land use planners, real estate experts, highway planners, transit agencies, environmental experts, academics or utilities providers. 7. Support from upper management. In most agencies that integrated land use/transport models, upper management helped set goals for what should be analyzed, but decisions on model selection and implementation were left to the experts of the modeling team. 8. Transparency within the agency, with partner agencies, and with the general public. Several interviewees reported sharing input data, model design, and model results with other modeling experts and stakeholders in their region, which generally increased con­ fidence in model results and acceptance of decisions supported by model results. Figure S-4. Locations of agencies that were the subjects of in-depth interviews.

Summary 5 9. A long-term vision for integrated land use/transport modeling within the agency. Having an internal long­term plan for model implementation, development, funding and staffing approved by upper management and the modeling staff has helped agencies to develop more sophisticated and more easily defendable integrated land use/transport models. The interviews revealed additional research needs, in particular on data standards and the development of open­source modeling frameworks. The shapefile, a common data standard for geo­referenced data, is in the form of points, lines, or polygons. There is no comparable standard in land use or transport modeling. Every software uses its own implementation of networks, zone systems, matrices, synthetic populations, and so forth and these are commonly not transferable from one modeling system to another. The OMX system is an exception— it is an open­source concept to store matrices that makes it easy to share trip tables or travel times between modeling systems. Similar concepts need to be developed for networks, synthetic populations, and other modeling elements, so that agencies can share input and output data more easily. Likewise, open­source modeling frameworks need to be developed for integrated land use/transport models. ActivitySim is an alternative open­source frame­ work for activity­based models, as is VisionEval for strategic planning tools. The Consortium for Scenario Planning, an initiative of the Lincoln Institute of Land Policy, supports the implementation of sketch planning models. Comparable concepts need to be developed for advanced land use models. Implementing regional model user groups was reported as an important step to keep modelers abreast of the latest developments. Similarly, establishing an expert panel of people who can advise on the implementation process and application of integrated land use/transport models was repeatedly mentioned by the interviewed agencies as a way to avoid pitfalls and add credibility to model results. Agile development, a computer science term, refers to building simple models first and improving the model gradually, based on current modeling needs. Several agencies emphasized the value of this approach. Because it allows for early successes, the model can be adjusted more rapidly to current modeling needs, and the approach helps the team to gradually advance their modeling expertise as the model becomes more sophisticated over time. This research effort revealed that integrated land use/transport models have a significant influence on decision making. The interviewed agencies reported that model results are used as additional evidence to support (or reject) proposed policies. Given that these models represent the land use/transport feedback cycle explicitly, agencies reported that their modeling efforts generally received more credibility than using a transport model alone. Although data limitations for land use modeling remain a concern, new crowdsourcing sites (such as OpenStreetMap, Foursquare, Google, Rome2rio, and Twitter) can provide alternative data sources, which often may be accessed for free.