兰德-STM3-2011-base-frequency,-mode_61页_1mb

Summary of STM3 2011 Base Frequency, Mode-Destination and Car Ownership Models

Core Content

This report documents the updates to the Sydney Strategic Transport Model (STM3) to reflect a 2011 base year instead of the previous 2006 base year. The modifications were made in the second half of 2014 at the request of the Bureau of Transport Statistics (BTS) of Transport for New South Wales. The aim was to improve the accuracy of the model in representing current travel behavior in Sydney, especially after the zone structure was expanded to include Newcastle and Wollongong, and the number of zones approximately tripled.

The updates included:

• Choice data: Updated using more recent Household Travel Survey (HTS) data, with different waves selected based on the variability of tour rates.
• Level-of-service (LOS) data: Enhanced to reflect updated travel conditions.
• Attraction data: Updated to align with the new zone structure.
• Car cost data: Adjusted to reflect more current costs.
• Taxi cost data: Updated accordingly.

The model components updated include frequency models, mode-destination (MD) models, and car ownership models. These models were re-estimated using more recent data to ensure they accurately reflect the 2011 base year.

Main Points

1. Model Updates

• The 2011 base version of STM3 was developed using data from 2009 to 2012 for purposes with high variability in tour rates (e.g., commute and home-tertiary education).
• For other purposes with lower variability, data from eight waves was used.
• The updated models were validated and implemented to improve the representation of current travel patterns.

2. Overall Model Fit

• The log-likelihood per observation (LL/obs) was used to assess model fit.
• For most purposes, the 2011 base model shows a better fit than the 2006 base model.
• The home-tertiary education and home-other travel models show a slight worsening in fit, but no large changes in LL/obs across the model set.

3. Accessibility Parameters

• Accessibility parameters were incorporated into the frequency models for home-based tours.
• These parameters account for differences in accessibility based on home zone and socio-economic segment.
• The significance of most accessibility parameters is low, and none of the changes are statistically significant.

4. Public Transport Out-of-Vehicle Time Ratios

• Out-of-vehicle components such as access/egress time, wait time, and boardings were considered in the MD models.
• For most purposes, the same treatment of boardings was used as in the 2006 base model (5 minutes of in-vehicle time).
• Home-shopping and home-other travel models used separate estimates for boardings in the 2011 base version.
• The 2011 base model parameters for out-of-vehicle time are more plausible compared to the 2006 base model, especially in terms of walk time and wait time valuation.

5. Toll Choice Model

• The toll choice model was updated to reflect more recent travel behavior and data.
• The changes included adjustments to the toll bonus and distance effect terms, which are detailed in Table 12.

Key Information

• The report is intended for a technical audience familiar with transport modeling terminology.
• The 2011 base version of the model system is referred to as STM3.
• The log-likelihood per observation (LL/obs) is used to compare model performance across different versions.
• The validation of the MD models included analysis of implied values of time (VOTs) and elasticities.
• Car ownership models were updated to reflect more accurate ownership rates and cost data, with improvements in the company car ownership model and a slight worsening in the total car ownership model.
• The car availability adjustment and car ownership pivot were modified to align with the 2011 base year.

Conclusion

The update to the 2011 base year in STM3 represents a significant effort to improve the accuracy and relevance of the model for current travel behavior in Sydney. While most model components show improved performance, some, such as home-other travel, show only minor changes. The report highlights the importance of using updated data and model specifications to reflect evolving urban transport patterns and user behavior.