Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 38
THE SAN FRANCISCO MODEL IN PRACTICE 27 ibrated against the 1990 MTC survey data for primary the average daily results. All targets of highway assign- destinations by purpose and distribution of trip length ment validation were met except for two screen lines and frequencies. The results reflect reasonable allocation of one neighborhood. For transit assignment, all modes destinations among four areas of the city and those des- were within 5% of the observed transit boardings. How- tinations located outside the city. The estimate of ever, there were some distinct differences by time of day, employment that results from the work location model with estimated bus boardings significantly greater than compared with actual employment by neighborhood observed boardings in the a.m. peak period. Matching showed that results were reasonable when compared the number of a.m. bus boardings within 5% would with estimated values by neighborhood. The biggest dif- require a 30% reduction in work transit tours compared ferences were the two neighborhoods in the core busi- with the observed data from the 1990 MTC household ness district, which were underestimating employment, survey. An independent estimate of U.S. Census journey- but calibration results also showed that the destinations to-work data indicates that the observed transit share of in the core were within 3% for each tour type and were work tours (35%) is reasonable. Therefore, the observed actually overestimated in these results. The destination work walktransit share was held constant, causing an choice model was also calibrated by comparing trip overestimation of a.m. period local bus trips. length and duration frequency distributions. These results showed reasonable average trip lengths and dura- tions for all tour types. The validation of the intermedi- COMPARISONS WITH TRIP-BASED MODEL ate stop choice model was challenging because similar models of destination choice had not included separate The comparisons of the San Francisco tour-based model validation of the intermediate stop choice component for with the MTC regional trip-based model showed comparison. The results of this validation test were that expected differences in the base year model and some both work and other tours were overestimated slightly interesting differences in the forecast year model. by the model, while work-based tours were underesti- Because the base year models were both validated to mated. Additional calibration adjustments to try to rec- observed data sets, the authors did not see as many dif- oncile these differences were not pursued because further ferences in those as in data for the future, when impacts adjustments would have negatively affected the results of of various forecasts showed different effects in the mod- the highway assignments by time period. eling systems. · Mode Choice (Tour-and-Trip) Models: The tour- and-trip mode choice models were calibrated by tour purpose. The calibration results for tour and trip modes Base Year 1998 showed a close match between estimated and adjusted observed tours and trips by mode and purpose. Initially, SF-CHAMP predicted tours by type rather than trips, so estimated transit boardings were discovered to be much a direct comparison of the home-based work trips was higher than observed boardings, particularly for local difficult (3). The 1996 MTC survey was used for cali- bus and Muni Metro transit modes; it was concluded bration because the number of trips within San Fran- that either the transit calibration target values generated cisco County was very low in the 1990 MTC survey from the household survey were too high or the observed (used to calibrate the MTC trip-based model) due to transit boardings were low. Because the transit board- underreporting of trips that occurred in that survey. The ings were calculated annually by Muni, they were held underreporting of trips was not consistent across periods constant, and both the observed and estimated transit or across trip purposes, which may have influenced shares were adjusted to match boardings better. model estimation that was based on the 1990 MTC sur- · Trip Assignment: There were two major modes for vey. Off-peak periods and work-based and other tours assignment validation: highway and transit. These were were all underestimated as a result. validated separately by using observed volumes of vehi- Trip rates per household were compared by trip pur- cles and passengers on the highway and transit systems, pose and showed that trip rates overall were similar, but respectively. Assignment validation at the county level the trips per household by trip purpose were quite dif- was completed by means of aggregated volumes by cor- ferent. For example, the model underestimates work and ridor (identified by screen lines), type of service (facility school trips compared with the MTC survey, but this dis- type, mode, or operator), size (volume group), and crepancy can be attributed to the survey's definition of a period. Speeds and travel times were also used in high- trip to work or home as containing all trips to and from way and transit validations to ensure that these were work or school. The San Francisco model differentiates accurately represented in the models. The highway between trips to work or school with an intermediate assignment results were compared for five periods and stop from those without an intermediate stop and thus