Chapter 5: Transportation Impact Modeling Technologies

Numerous models are used today to evaluate potential environmental effects of transportation projects, primarily models of air quality, noise, water quality, and biological resources (see table below). Like all models, these transportation impact models are simplified representations of real conditions, based on assumptions and available knowledge. Although impact models themselves are not a new technology, there are a number of innovations in the collection of model input data and the types of models used to process the data. The discussion that follows addresses three techniques for data collection analysis through models:

• Gap analysis: Organize baseline data on existing features according to user specifications, but provide no interpretation.
• Integrated models: Generate high-level predictive data (model output) using single, integrated or multiple model systems.
• Expert systems: Generate knowledge using baseline data and model output data in a rule-based system.

Integrated models (where interactions among impact areas are modeled) could include a number of impact areas and interactions (e.g., land use and water quality, land use and air quality, various plant and wildlife species). The rapidly emerging area of integrative transportation-land use impact models are discussed as the main example here, based in part on their recently recognized importance and the relatively little attention received in past applications.

Expert systems (where conclusions about impacts are drawn) can be broad-based—designed to provide impact information on the full range of human and natural systems—or resource specific. Broad-based models are discussed as the main focus here because they more fully address the objectives of this research project.

Gap analysis is the basic use of GIS in a structured way to determine the nature and location of potential impacts on existing resources of the built, natural, or social environment. Most commonly, the user specifies buffers around sensitive natural resources or other features as a first step in identifying constraints related to a given project. Gap analysis is a screening tool that precedes quantitative and other analyses through the use of models and other tools. The examples discussed here represent the range of potential uses and scales, from local or regional land use analysis to statewide natural resources analysis.

GIS Grid Technique: Standard GIS data is encoded in polygons—irregular shapes that correspond to areas on a map. A common problem is that polygons in different layers are not directly comparable. Running scenarios using polygon data is either not possible or is very slow. In contrast, grid (or rastor)-based data allows for direct comparisons and faster model runs. Grid scales can be of variable size depending on the resolution that makes sense for the project. A 30-meter scale might be appropriate for a regional analysis; a 1-foot scale might be appropriate for a subarea (e.g., small city, neighborhood) analysis. Averaged over hundreds or thousands of grids, the accuracy is considered very good. Still, the results are best used to produce visualization tools and would not be appropriate for drawing an exact regulatory line. The grid technique can be used to build proximity models that are much harder to do using arc or polygon technology.

An example is the preparation of a GIS-based comprehensive plan in Athens, Georgia. Data provided up front included zoning, vacant land, parcel data, building footprints, pavement areas, digital aerial photos, and natural resources. Environmental constraints were mapped from aerial photos and studies of existing land use patterns, growth patterns, and demographics. Potential future land use scenarios were tested against existing systems, including transportation, environmentally sensitive areas, schools, parks, and special districts. For the transportation impacts study, household and employment projections were analyzed and assigned to existing transportation analysis zones. The zone projections were then used as the basis for traffic modeling. The results of this model were then imported into the GIS system and became part of the land use plan. Another impact study identified constrained lands based on conditions such as wetlands, open water, riparian corridors, floodplains, and groundwater recharge areas. Sensitive lands were identified on aerial photos by professionals familiar with the area. The constrained lands were designated as off-limits for future development.

Gap Analysis Program: Gap analysis in this example is a scientific method for identifying the degree to which native animal species and natural communities are represented in our present-day mix of conservation lands. Those species and communities not adequately represented in the existing network of conservation lands constitute conservation "gaps." The purpose of the Gap Analysis Program (GAP) is to provide broad geographic information on the status of ordinary species (those not threatened with extinction or naturally rare) and their habitats in order to provide land managers, planners, scientists, and policymakers with the information they need to make better-informed decisions.

GAP is sponsored and coordinated by the Biological Resources Division of the U.S. Geological Survey. Additional support at the national level has been provided by the Department of Defense and the Environmental Protection Agency. The program has a close working relationship with the National Mapping Division of the U.S. Geological Survey and with The Nature Conservancy. Mapping and analysis is conducted by GAP projects within each state. Additional analyses are conducted for large multi-state regions in partnership with state governments, federal agencies, and other cooperators. Presently, GAP is made up of more than 445 contributing organizations in 44 states. Contributors include business, universities, state and federal agencies, tribes, and nongovernment organizations.

Vegetation is mapped from satellite imagery and other records using the National Vegetation Classification System. Native animal species ranges are mapped by using museum and agency specimen collection records in conjunction with known general ranges and the animal’s affiliation with the previously mapped vegetation types and other physical characteristics. These data are combined and displayed with a computerized GIS at a cartographic scale of 1:100,000. Maps of vegetation types, individual species, or selected suites of species (depending on one’s interest) are overlain on maps of land ownership and land management, showing where land-based conservation efforts can be focused to achieve the conservation of selected elements of biodiversity most efficiently, preventing both conservation crises and land-use surprises.

Because GAP provides a standardized method and format, the data can be edge-matched with adjacent states as the state projects are completed. The importance of having data that is consistent across state boundaries is in revealing, for the first time, actual patterns of biodiversity at scales relevant to both the magnitude of present-day changes and the multiple levels of biological organization, from species and natural communities to large landscapes.

GIS-based gap analysis is extremely useful as a screening tool for identifying and locating resources in the built, natural, or social environments. It can be used for preliminary assessment of fatal flaws and potential effects at both project and regional levels. It allows for rapid integration of GIS data layers in assessing information and provides an accepted baseline for defining habitats and other resources at a coarse scale. GIS data may be linked to other software, expert systems, and resource depositories for use in more complex analyses. The GAP program is viewed as a very useful tool by biologists to assess regional biodiversity issues.

The main limitation of gap analysis is that it is only useful at a coarse scale and for screening-level analysis. It forms the base information layer for more detailed analyses. The grid technique is highly accurate on an areawide basis but is not precise at the individual grid level. The GAP program, because of its applicability at the statewide and larger levels, is not as useful at the project level.

Integrated models recognize the interdependence of resources and that modeling each impact independently may not accurately represent the ecology of the impacts. Although integrated models could be developed and used for a number of impact areas, some of the most visible examples are those being used to analyze the interaction between changes in transportation infrastructure and changes in land use.

Traditionally, transportation impacts and land use impacts have been analyzed independently, with little attention to the interdependence between the two systems. Today, there is mounting pressure to consider both the impact of transportation access on land development decisions and the effects of land use patterns on transportation demand more formally in long-range planning, major investment studies, and transportation alternatives analyses. A number of modeling and graphic tools are available for analyzing the interaction between changes in transportation infrastructure and changes in land use. These tools are commonly used to answer questions about alternative growth scenarios based on transportation system investments and public policies dealing with urban form and growth management.

The technologies discussed here include:

• Land use forecasting models that incorporate transportation system impacts
• Transportation models that incorporate land use impacts
• Multiple-model systems that seek to address both topics and others (e.g., econometrics) in a single model or model system

Land use forecasting models are formal methods used to forecast the distribution of socioeconomic activity within a planning region or area. The primary value of land use modeling is to provide a reproducible framework for estimating future development and activity allocations in response to land resource availability, transportation system infrastructure, resource constraints, and public policy. Potential future land use scenarios can be tested for their effects on transportation and natural resource systems. Typical model inputs include GIS data describing land use, transportation facilities, natural resources, housing, and employment. Model output includes sketches and/or visualizations that test land use or growth scenarios against impacts on built and natural systems such as transportation or sensitive natural resources. Because of their limited geographic resolution, most land use forecasting tools are best suited for preliminary evaluation of alternatives in order to select those that warrant advanced analysis using more sophisticated tools.

Most traffic and travel behavior projections used in the environment impact assessment process have been based on the traditional four-step travel forecasting model or a subset of it. The steps of the four-step travel modeling process include trip generation, trip distribution, modal choice, and trip assignment. Despite its common use, the four-step model has a number of limitations. These include a general shift from a concern for developing projections to be used in facility and travel network design toward evaluation of policies to promote sustainability and protect environmental quality. However, a new generation of travel model concepts and approaches is under development to improve and extend the traditional four-step travel modeling approach.

Multiple-model systems for transportation and land use attempt to address transportation and land use interactions in a single comprehensive model or through multiple models used as an iterative system. In addition to connecting traditional land use and transportation models, tools in this category also integrate economic considerations such as land prices and other real estate conditions. These models are truly at the leading edge, taking the innovations of the individual models and combining them into even more holistic representations of reality.

INDEX: INDEX is a customizable GIS-based community planning tool that includes scenario analysis and indicator tracking applications. Recently, a version called Smart Growth INDEX was developed for the U.S. EPA. With this model, sketches can be prepared and analyzed for regional growth management plans, comprehensive land use plans, transportation plans, neighborhood plans, environmental impact reports, and a number of other uses. Smart Growth INDEX is distinguished by an internal travel demand submodel that can estimate transportation outcomes from land use changes without the use of a traditional four-step transportation model. It can also run in tandem with transportation models.

Examples of typical model inputs include land use plan designations, existing housing and employment, existing and future street center lines, and transit routes. To use the model, these data are entered, stipulating growth constraints and incentives. Then growth is spatially allocated to a horizon year using a travel-based gravity submodel with land use and transportation interaction in each interval year. The finished sketch is scored and mapped by performance indicators. Smart Growth INDEX scores its sketches with a set of 24 performance indicators that measure outcomes such as land consumption, housing and employment density, proximity to transit, pollution emissions, and travel costs. Land allocations are also tabulated for land use classes and local jurisdictions.

Modified Four-Step: Many of the current practice versions of transportation models have incorporated more sophisticated elements and integration (some include add-on models) to make traditional travel modeling more sensitive to socioeconomic attributes and land use/urban form inputs, modal system elements, defined growth policies, and constraints. These adaptations of the earlier, more primitive and simpler four-step transportation modeling are attempts to better represent travel behavior, recognize land use feedback loops, and incorporate policy and urban form factors that promote sustainability and protect environmental quality. Among other things, these tools bring with them the ability to better model environmental consequences of transportation plans, including air quality, noise, disruption of social linkages and fabric, and consequences for natural resources and habitats.

The general thrust of transportation planning enhancements can be characterized around four domain areas:

• Activity-based Trip Generation: These improved behavior constructs consider time budgets for human activity and the allocations made by different individuals and households to travel needs. A second refinement is inclusion of nonmotorized as well as motorized travel as a function of accessibility and urban form in the immediate environs.
• Simultaneous Trip Distribution-Modal Choice Equilibrium: Recognition of joint decisionmaking in travel destination and mode choice is at the heart of these modeling enhancements. These improved modeling processes typically are in place in larger urban areas, where multimodal transportation systems and investment decisions arise. Additionally, modeling feedback loops are incorporated to assure equilibrium consistency. That is, network conditions assumed for the trip distribution and modal choice allocations reasonably conform to the network results after trips are distributed and assigned to modes.
• Multi-class Traffic Assignment and Integrated Highway/Transit Networks: More sophisticated traffic routing and link assignment methods now allow analysis of mixed-use and high-occupancy lane volumes as well as specific and separate routing logic algorithms for commercial vehicles. Similarly, integrated network representation maintains consistency in level of service attributes for autos and transit vehicles that share the same roadways/right-of-way.
• Stated Preference Surveys: Unlike attitudinal surveys, stated preference gauges user response to a set of options by describing each option as a series of economic tradeoffs. This allows the relative acceptability of each option to other options available to be evaluated in a consistent framework. By framing user options as trade-offs, the user is presented with a realistic and consistent scale of probable impacts of making a particular choice. Stated preference surveys can better reflect likely stakeholder response to changes in environmental and transportation conditions associated with planned projects and can inform planners on issues where information on user direct experience or historical data is not available.

For a new roadway near Cardiff, Wales, stated preference techniques were used to evaluate potential temporal, route, and mode shifting that could result from the project. Specifically, stated preference surveys were used to examine driver route changes, trip retiming, destination changes, modal shifts, and vehicle occupancy impacts. Four-step modeling for the project was informed by the stated preference work through translating likely user economic benefits and costs for a given choice to a cost value, which was then factored into the modeling of user route, destination, and mode choices.

In another example, in the Phoenix, Arizona, region, conventional travel forecasting techniques were married with stated preference to determine ridership impacts of "soft" variables such as rider comfort, security, and expected reliability resulting from replacing conventional bus service with light rail and/or advanced technology guideway vehicle systems such as group rapid transit. The economic benefits that could be attributed to such soft variables were used to modify estimated travel cost outputs of the conventional mode split calculations.

TRANSIMS: A third example is TRANSIMS, a travel demand forecasting model incorporating a new approach to the forecasting process. TRANSIMS is based on application of disaggregate micro-simulation techniques for both traveler and traffic behavior. TRANSIMS represents a major departure from current methods of preparing large area travel forecasts and is an attempt by the federal government to address identified issues and shortcomings of those methods. The model’s four major elements include a population synthesizer, activity generators, a route planner, and traffic micro-simulation.

This population synthesizer creates a synthetic set of transportation system users with associated behavioral and socioeconomic characteristics based on a seed population. The activity generators then simulate individual traveler needs and choices to project travel desires and then resolve those desires into trip planning. The route planner matches forecast trip activity with the possible routes and vehicles available for trip making. Finally, the traffic micro-simulation component performs traffic assignment (the process of translating trip patterns into traffic and traveler volumes on specific facilities and links) by simulating individual traveler and vehicle movements as a function of elapsed time and simulated interactions.

TRANSIMS is currently undergoing prototype field implementation and related modification and improvement using the Portland, Oregon, region as the major test area. Tests to be performed in Portland include developing synthetic substitutes for selected transportation features and attributes and subsequent evaluation of the impacts of using these substitutes on resulting outputs. The study plan includes subsequent phases to evaluate TRANSIMS integrated vehicle emissions models and to examine its effectiveness in simulating impacts of technology based on transportation system enhancements (intelligent transportation systems).

TRANSIMS is also being tested at the corridor level in Dallas, Texas, focusing on the micro-simulation traffic assignment model to evaluate computational and related hardwaresoftware requirements.

Metroscope: Metroscope consists of four models—a regional econometric model, residential and nonresidential real estate models, and a transportation model—embedded in a GIS-based accounting and visualization tool. Output from the econometric model is fed into a modified four-step transportation model; output from the transportation model is then used as input to the other models for subsequent iterations. Portland Metro has plans to use Metroscope in the following four work efforts:

• Evaluation of public welfare impacts of proposed additions to the urban growth boundary (UGB). Specifically, testing 600-acre additions with various combinations of land use in several areas of the region and comparing model output to a 2015 base case forecast.
• I-5 Trade Corridor Study. Involves determining whether to provide additional transportation capacity across the Columbia River between the Portland Metro Region and Clark County, Washington. Modeling work run in 5-year increments will determine land use and transportation impacts throughout the economic region. Of particular concern in this study will be the growth of households and employment in Clark County resulting from a particular transportation investment alternative.
• Year 2000 forecast update. This project will produce a detailed project by housing type, price/rent, category, etc. Modeling will occur in 5-year increments with land added to the UGB and transportation investments made according to the relevant regional transportation plan.
• Urban design evaluation of the 3,500 acres added to the UGB in 1999 (Pleasant Valley). Will use models to determine design and land use mix configurations that will maximize use and minimize vehicle miles traveled.

UrbanSim: The UrbanSim model has been implemented in several areas and is in development in several others. The Wasatch Front Regional Council in Utah is developing an application of this model. Their plans to use it are two-fold: for developing better socio-economic projections (e.g., 2020 or 2030) for the entire region and then as part of scenario development to determine potential changes in land use that would result from a specific transportation project. Examples of the latter would be at the MIS or EIS stage for a new road or substantial expansion of an existing road. The model would be sensitive enough for smaller scale projects (e.g., major arterial widening).

The new and emerging generation of integrated transportation-land use models are more accurate than their predecessors. Land use models can be used to predict and illustrate the potential effects of land use scenarios on a variety of transportation, environmental, and other indicators. The modified and integrated traffic models attempt to address the many limitations of the traditional four-step models. Advantages include the ability to predict localized impacts of traffic congestion (e.g., specific delays); behavior at intersections, including queuing and acceleration/deceleration; the influence of different modes of travel upon each other; shifts in trip making time due to network conditions; inclusion of nonmotorized travel; the incorporation of a transportation-land use feedback system to account for interactions over time; and others.

GIS-based land use visualization and scenario development tools require access to or creation of a GIS data layer for all applicable information types. For uses where little or none of the data are available electronically, the effort to digitize the data may be prohibitive. Scenarios developed in the GIS-based land use tools are useful for visualization and broad planning purposes but generally are not accurate enough to draw regulatory lines, for example. SmartGrowth INDEX, for example, is best suited for preliminary evaluation of multiple alternatives in order to select those that warrant advanced analysis using more sophisticated tools but would not be appropriate by itself for evaluation of major investments or documenting regulatory compliance. With the multiple-model systems, the main limitations currently are their complexity and the resources required to fully develop and implement them in a given geographic area. Although these barriers will be reduced with increasing use over the long term, in the short term, these models will only be accessible to a select group of users able to support and fund this activity.

Expert systems generally consist of a set of rules and user-supplied data that interact through an inference engine, an expert, or knowledge-based system that is able to derive or deduce new facts or data from existing facts and conditions. Expert system shells have become more widely available, allowing users to define the database and rule base without using artificial intelligence programming languages. Less commonly, individual organizations will create their own expert systems for specific purposes.

Expert systems have been developed and applied to large-scale and small-scale projects in North America and worldwide. For the most part, the systems have been used for various water resource, military, and other large-scale infrastructure development efforts where GIS applications were either planned or available for use in developing the decisionmaking system. The examples below are divided into applications of commercial systems and systems developed by agencies for specific uses.

Nobility EM: This model is a Windows-based, GIS-enabled expert system for environmental planning and management decision support. Nobility EM was developed to enable generalists to make decisions that often require access to multi-disciplinary knowledge and the analysis of multiple options. The knowledge-based decision support system is designed to streamline and simplify the complex process of environmental decisionmaking. Nobility EM captures and deploys both information (i.e., site-specific data and project characteristics) and multidisciplinary knowledge (i.e., information supplied by domain experts and coded as decision rules or chains of cause-and-effect). These decision rules describe the relationships and conditions under which impacts may occur for different project activities and link to the specific management actions that are necessary to minimize risk and achieve compliance.

Nobility EM integrates GIS, relational databases, reporting tools, and special-purpose models in a single system.

ESSA Environmental Assessment (Calyx EA): Calyx EA is a screening expert system that allows users to preview potential environmental impacts in an assessment process. The system is designed to enable users to identify and quantify a spectrum of potential environmental impacts and possible forms of mitigation. Users can produce concise, detailed reports based on the findings of the analysis. Calyx EA may be used in conjunction with other software products that involve forest and wildlife resource management, aquatic and fisheries resource management, and land use planning. Technologies that may be integrated included computer simulation models, relational databases, and decision support systems.

• Environmental Software and Services GmbH EIAxpert: EIAxpert is an interactive, rule-based expert system for environmental impact assessment. It is designed to provide a screening-level assessment of development projects at a pre-feasibility stage. The system can operate with a minimum of data and information about the project, its alternatives, and the environmental setting by drawing on a generic knowledge and rule base as well as a regional geographical information system and a set of databases. The system was designed for the assessment of water resources development projects and uses checklists and assessment rules derived from the Asian Development Banks Environmental Guideline Series (Fedra, et al 1991).

• Facet Decision Systems, Inc. Baseline Environmental Assessment Framework: Facet Decision Systems Framework prepares a regional, state, or countywide baseline assessment of environmental conditions based on a standard, digital set of topographic, land-use/land cover, and other maps. The system generates a status report by partitioning the landscape watershed boundaries or administrative units and applying a rule base that measures features such as road density, stream-buffer encroachment, and other factors. It is designed to organize decision processes so these decisions are repeatable, efficient, well informed, and rapid. The system also can include cause and effect functions that simplify development of decision rules.

Pennsylvania Department of Transportation Hydrologic and Hydraulic Joint Permit Application Expert System: In Pennsylvania, projects with potential hydrologic and hydraulic impacts are required to complete a Joint Permit Application (JPA) through the Pennsylvania Department of Transportation (Penn DOT). Although the JPA was developed to streamline the combined state-federal permit process, the application is complex and difficult, with multiple agencies involved. To simplify and improve the accuracy of the JPA, Penn DOT has developed the Hydrologic and Hydraulic (H&H) JPA Expert System. The system’s main features are assistance with the administrative side of the application process—determining which type of permit is required for a given project and where the permit should be filed—based on the project type and anticipated impacts. Currently, DOT staff can use the system online, and outside users can view and track the output of the process on the DOT Web site. Starting in 2001, users outside the agency will be able to use the online system to determine the type of permit needed and check to see that all required information has been provided. Eventually, it is anticipated that permits will be submitted electronically using this system. The current system is based on a limited set of rules. Because many of the permit questions are complex or subjective, building in knowledge to address the full extent of technical issues has been more difficult.

Penn DOT CEE Expert System: Under NEPA, certain projects without significant environmental impacts are classified as a "categorical exclusion" (CE) and require only limited environmental analysis. Despite the apparent simplicity of such evaluations, they can still involve detailed and cumbersome processes. Penn DOT is currently conducting prototype testing of a Categorical Exclusion Evaluation (CEE) Expert System to streamline the CE process. The CEE Expert System will allow the user to take a laptop computer into the field to gather necessary information that is automatically saved and converted to the CEE format when a CEE project is initiated, eliminating the need for redundant data entry and reducing opportunities for errors. When fully functional, the rule-based system will use Internet-based access to obtain information from other databases (e.g., threatened and endangered species, cultural resources, water quality, etc.) and incorporate that information directly into the electronic form. The system is designed to allow electronic approval and transmission, eliminating the delays associated with paper transmission. An extensive help system will include frequently asked questions and a detailed glossary and acronym listing, as well as hyperlinks to external sites for all regulatory references.

Mn/Model Archaeological Predictive Model: Mn/Model is a GIS-based, statewide archaeological predictive modeling project conducted for the Minnesota Department of Transportation, with funding from the FHWA. The purpose of the model is to provide new information about the probability of finding archaeological properties. High-resolution models were built by regions using Arc/Info Grid and then merged to form a statewide model. The foundation of the model is the statistical relationship between environmental variables and the locations of known archaeological sites and surveyed locations. The model was constructed using environmental data and information in existing historical and archaeological sites’ databases. In the course of evaluating preliminary site probability models, the non-random pattern of archaeological surveys in Minnesota became apparent. This led to the development of survey probability models, which predict survey bias. The high-, medium-, and low-probability zones in these models indicate the likelihood that archaeologists have surveyed places with different environmental characteristics. The final result is a survey implementation model that indicates, on a statewide basis, the potential for archaeological sites. The entire state is divided into categories of potential: unknown, possibly low, low, suspected medium, probably medium, medium, suspected high, and high. The models’ intended users are cultural resources staff; however, planners and engineers will be able to use survey implementation models to identify areas with a high potential for archaeological properties.

These systems have utility in being applicable to the planning phase and to the longer term mitigation and monitoring phases of a specific project. Additional applications can be developed after initial investment in software, systems development, and user training are made for a project. There may be more cost effectiveness considerations for multiple applications or long-term reanalysis of the same database as future projects or developments are considered in a defined geographic area.

Application of expert systems to a specific spatial area of sufficient scale (state, county, district, watershed, etc.) would increase the overall utility and benefits of an expert system approach to project planning, land use planning, and other long-term strategic analyses used in determining local and regional transportation needs.

The benefits of the system are derived from long-term use of the developed application and/or from the ability to run numerous scenarios to document the sensitivity of the decision analysis.

Expertise of resource specialists is required to establish the knowledge engineering elements of the systems, allowing the system to meet application-specific needs. In addition to having the resource specialists apply the criteria to the decision process, software and user training are also required for the environmental decision process.

When compared to existing technology and use of various GIS database information tools, the application of an expert system requires an additional initial investment of time and knowledge. Limited application of the expert systems technology when compared to existing technologies may be the single most critical factor in rating the benefits of this technology. Initial costs of software license, application development, and user training may contribute to the currently low level of usage worldwide. User perception and reality of learning to adapt and develop the software for each project application may also limit the utility and widespread use of these systems.