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.
has fewer trips identified as work or school trips and many more identified as non- home- based trips. A relative comparison for trip distribution was the summary of employment attracted to each zone as part of the work tour primary- destination choice model. This comparison required the estimation of non- San Fran- cisco residents who work in San Francisco by zone, which, to some degree, may have biased the comparison results. Another comparison was the trip table at the district- to- district level for intracounty trips; this table showed a strong correlation in percentage distribution of trips by district between the San Francisco and MTC models but a difference in total trips due to the underes- timation of trips discussed in trip generation. Trips by mode and superdistrict showed a strong sim- ilarity between the results of the mode shares by superdistrict, which resulted from the fact that both mode choice models were developed from the same 1990 MTC travel survey data. A comparison of the vehicle trips showed there is a significant difference between the trip- based and the tour- based auto mode shares. Drive- alone trips are slightly overestimated in SF- CHAMP, and carpool trips are underestimated compared with the MTC model. A comparison with the Census Transporta- tion Planning Package (CTPP) for trips within San Fran- cisco showed that drive- alone trips were 89% of total vehicle trips and shared ride trips were 11% of total vehi- cle trips, which bore a strong correlation to the San Fran- cisco model results. Forecast Year 2030 MTC produced Year 2030 forecasts for its regional transportation plan. The SF- CHAMP model used the same land use projections, road improvements, and regional transit improvements as the MTC model. This consistency allowed for convenient comparison of results from the mode choice steps of each model. The overall trip rates per household remained similar in the 2030 forecasts for both models: about 9.2 trips per household. As in the base case, for the two models, the distribution across the various trip purposes was dif- ferent, due again to the impact of intermediate stops; the MTC model predicts more home- based trips, particu- larly work trips, and fewer non- home- based trips than the SF model. This accounting issue is well understood. Examination of the geographic distribution of trips revealed more differences. In the base year, the San Fran- cisco and MTC models predicted similar overall levels of trip- making among the four quadrants (defined by MTC as âsuperdistrictsâ) of San Francisco; comparison of the trip distribution patterns for the San Francisco and MTC models showed that all movements between all superdis- tricts varied by less than 3% on relative terms. Again, the absolute trip- making rates were different due to trip gen- eration issues described earlier. When the 2030 distributions of the two models were compared, larger differences emerged. Compared with its base year forecast, SF- CHAMP showed a small reduc- tion in intradistrict movements for all quadrants except the Sunset, the lowest- density and most suburban car- oriented part of the city. The Sunset district was the only quadrant that increased its share of trip making to and from all other quadrants, by up to four percentage points. No district- to- district movement changed by more than four percentage points when the base was compared with the 2030 forecast with SF- CHAMP. The MTC model showed larger swings in trip distri - bution in a somewhat similar pattern. Again, the Sunset district showed growth, but the MTC model also pre- dicted a relative increase in trips to downtown and an increase in intradowntown trips. These data contra- dicted the SF- CHAMPâs 2% reduction in trips to down- town. This finding echoed other studies that have found the gravity model used in trip- based distribution models to be quite sensitive to changes in travel time. From the perspective of mode split, the two models behaved in similar manners in the base and future years. In both the base year and 2030, the SF- CHAMP model predicted more walk trips, fewer transit trips, and more drive trips than did MTC. The relative size and direction of these differences was about the same in both base and future years, except for walk trips. MODEL APPLICATIONS Equity Analysis SFCTA developed an application of the San Francisco tour- based model to estimate impacts on mobility and accessibility for different populations so as to support development of a countywide transportation plan (4). Equity analyses based on traditional travel demand fore- cast models were compromised by aggregation biases and data availability limitations. Use of the disaggregate (individual person- level) San Francisco microsimulation model made it possible to estimate benefits and impacts to different communities of concern on the basis of indi- vidual characteristics such as gender, income, auto avail- ability, and household structure. Tenderloin Residents A recent study of the predominantly low- income Tender- loin neighborhood took advantage of disaggregate model outputs to explore the differences between travel patterns of Tenderloin residents and other trip makers in 28 INNOVATIONS IN TRAVEL DEMAND MODELING, VOLUME 2