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.
11 Design Features of Activity- Based Microsimulation Models for U.S. Metropolitan Planning Organizations A Summary Mark Bradley, Mark Bradley Research and Consulting John L. Bowman, Bowman Research and Consulting This paper provides a concise summary of importantdesign features of various activity- based model sys-tems that have been implemented or have recently been designed for planning agencies in the United States. The models described are for Portland, Oregon; San Fran- cisco, California; New York; Columbus, Ohio; Atlanta, Georgia; Sacramento, California; the Bay Area of Califor- nia; and Denver, Colorado. These models were selected because they are in the same family of activity- based mod- els, and one or both of the authors have been involved in the design of all of them except for New York. Two other examples have also been included in the summary table and supplementary text of activity- based models in the United States: the CEMDAP model for Dallas, Texas, and the FAMOS model for southeast Florida (see sidebars, pages 14 and 17). Not included is the TRANSIMS model or the TLUMIP model for Oregon. Although those models share some of the features discussed here, the authors are not suf- ficiently familiar with them to compare them at the level of detail included here; that, however, could be a useful exten- sion of this paper. All model systems described in this paper share a similar overall structure, with a hierarchy of levels from top to bottom, with the lower choices predicted con- ditionally on higher- level choices. The levels are ⢠Population synthesis: geographic allocation of households; ⢠Longer- term decisions: auto ownership and (in some cases) work and school locations; ⢠Personâhousehold day level: number of tours and activities made for various purposes; ⢠Tour level: main destination and mode, begin and end times, and number of stops; and ⢠Trip level: intermediate stop location and the mode and departure time of each trip. Within this structure, several important design fea- tures distinguish the models, and these are summarized in Table 1. The models are listed in the table roughly chronologically, with the earliest ones on the left and the later ones on the right. At the time of writing, the Bay Area Metropolitan Transit Commission (in California) and the Denver (Colorado) Regional Council of Govern- ments models are in the design stage; therefore, the design characteristics shown for these models are those currently envisioned. Each following paragraph is a more detailed annotation of a row in the comparison table. CONTROLS AND CATEGORIES FOR POPULATION SYNTHESIS All model systems simulate persons one by one and require a representative sample of households and persons for the base year and forecast years. All regions use zone- level data and forecasts of household size and income as control variables for sampling households from the regional Public Use Microdata Samples households. In addition, most regions have used the number of workers in the household as a third control variable, both because it is important
