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.
oping denser activity centers. The Blueprint focuses on the four Ds: density, diversity in the mix of land uses, design related to pedestrian- and transit- friendly approaches, and destination or the utility clustering of complementary land uses. Activity- based modeling has a natural link to this land use policy focus. Activity- based models provide the level of detail needed to evaluate the impact of different land use plans and different land use patterns. ⢠The new SACOG activity- based travel demand model uses parcel- level data, rather than traffic analysis zones. The main reason for using the parcel- level data is that the data provide the detail needed to assess ques- tions related to development patterns, street patterns, and proximity to transit services. SACOG used PLACE3s, which is a parcel- based land use scenario analysis package, in the development of Blueprint. Since PLACE3s is parcel based, most of the land use data have been transitioned to the parcel level over the past few years. At the same time, SACOG coordinated a regional roadway geographic information system (GIS) coopera- tive focused on protocols and data standards for a regional roadway centerline GIS. SACOG has also devel- oped a transit GIS, which includes routes and stops, as part of a regional traveler information system. ⢠Based on these factors and other needs, SACOG initiated the development of an activity- based travel demand model at the parcel level. This approach does present data production challenges. Activity- based mod- els are data intensive, especially when they include the capability to capture the effects of land use, street pat- terns, and transit proximity. PLACE3s provides the capa- bility to display and analyze land use changes. It contains dwelling unit yields and constraint layers. Forecasts are developed starting at the parcel level. ⢠The estimation of the activity- based model was based on a 2000 household survey, and parcel- and point- level data on dwelling units, employment levels, street pat- terns, and accessibility to transit services. A comparison of the place type and development density between 2000 and 2030 can be made. In these areas, the forecast has to gen- erate an equivalent parcel- level detail on the street pattern and transit proximity. This task is a challenge. ⢠The street pattern is a geographical representation of how the streets appear and how supportive they are of nonmotorized travel and accessibility measures. Intersec- tions are used to help define street patterns. Three types of intersections or nodes are included in the GIS. These types include 1-link nodes, which are cul- de- sacs or dead- end streets; 3-link nodes, which are T- street intersections; and 4-link nodes, which are four- legged street intersec- tions. Based on the GIS definition, higher density or the prevalence of 3- and 4-link nodes is associated with âgoodâ street design and the prevalence of 1-link nodes is associated with âbadâ street patterns. A number of for- mulas are actually used in the model related to the street pattern. The most frequently used formula is the good intersection ratio, which is the sum of 3- and 4-link node intersections divided by the sum of all intersections, for a particular area. The reverse of this ratio is the bad inter- section ratio, which is the sum of all 1-link node intersec- tions divided by the sum of all intersections. ⢠The street pattern variables are critical inputs to many of the choice submodels and are very predictive. For work locations, work tour destinations, and nonwork and nonschool tour destinations, the good intersection ratio is a highly significant, positive variable. For school and other tour mode choice models, intersection density at the tour origin is a highly significantly positive variable for walk- to- transit, walk, and bicycle modes. The bad or dead- end street ratio is a highly significant negative variable for the intermediate stop location model. The difficulty is in fore- casting street patterns at the parcel level. ⢠Generation of forecast street pattern data that vary by parcel is used for input elements. These elements are the parcel- level dwelling and employment for the base year, the parcel- level dwelling and employment for the forecast year, the base- year roadway GIS, and a look- up table of densities of the three intersection types for dif- ferent types of areas. The generation process includes five steps. First, land use in the base year and the forecast year are compared by parcel, and parcels with changes are identified. Second, for parcels that are expected to change in use and are over a threshold acreage, synthetic points are generated in a grid pattern throughout the parcel. Third, each synthetic subparcel is populated with a computed number of each type of intersection using a look- up table. Fourth, the synthetic subparcel points in change parcels are merged with the real intersections from the base- year GIS for nonchange parcels. Finally, the merged points are buffered to parcel according to the specific street pattern variable definitions. ⢠A different approach is used with the transit prox- imity variables. The transit proximity variables are defined by straight- line distance or time from each parcel to the nearest transit stop in the GIS. Actual transit stop locations are used for estimation. For forecasts, transit stops must be added at reasonable locations. The future transit lines are overlaid on the existing routes. There are different methods of synthesizing future stop locations depending on the type of transit mode. There is a pro- gram to generate stops at reasonable spacing for fixed- route bus service. Most of the station locations for future light rail transit lines are known. These station locations are manually added depending on the alternative being modeled. For express routes, stops are only located in the neighborhoods being served. Julie Dunbar, Dunbar Transportation Consulting, mod- erated this session. 58 INNOVATIONS IN TRAVEL DEMAND MODELING, VOLUME 1
