Spatial Microanalysis and Equity Assessment of Joint Relationships among Destination Choice, Activity Duration, and Mode Choice (August 2021 - August 2022). Sponsor: Pacific Southwest Region 9 University Transportation Center, US Department of Transportation, and UCSB. Budget: $118,000. This project combines multiple strands of research and develops a new integrated framework for spatial choice modeling and simulation. We use ideas emerging from our own research in motif and sequence analysis based on the 2017 National Household Travel Survey for California. This is then complemented with other external data at the business establishment microlevel. We develop a structural behavioral model that shows spatial correlation among destinations, duration of activity at each destination and the mode chosen to travel to each activity location considered as forming activity chains/tours. Then, we post process the data to identify segments of the population that face unsurmountable barriers in activity participation and therefore become de facto excluded from opportunities. This project advances science in the combination of built environment data with individual/household behavior and in the formulation of a new generation of core models for modeling and simulation for large scale urban simulation model systems. The methods we develop here have applications in equity analysis.
A Before-After Intervention Experiment and Survey and Covid-19 Add-on Survey (August 2020 - August 2022). Sponsor: Pacific Southwest Region 9 University Transportation Center, US Department of Transportation, and UCSB Senate. Budget: $106,500. In this project, we first develop a strong theoretical model accompanied by data collection to test some of its aspects for behavioral change research. Data collection shifted from the originally planned public transportation intervention to the examination of COVID-19 impacts on the life of Los Angeles Metropolitan area residents. In terms of substantive findings, we verified that in this region as in other parts of the US people experienced loss of jobs, forced relocations, and major changes in working and studying. In terms of the attitude-behavior relationship, we also confirmed the existence of more diversity in attitudinal groups of people with respect to their position towards the private automobile and found that these attitudes are strongly correlated with the use of different modes. The survey design and conceptual model form the foundation for subsequent data collection and analysis based on the pilot examples of this project. A third pilot study within this project is the design of a smartphone application. Guidelines for survey design are provided in this report with a description of an ongoing research effort at UCSB that continues beyond the project reported here.
Connected and Autonomous Vehicle Demand Modeling for Integrated Transportation Modeling (BEAM Core) (May 2021 – June 2022). Sponsor: Lawrence Berkeley National Lab (LBNL) and UCSB. Budget: $80,138. In this project we develop Market Segments and their Potential Market Penetration of Connected Autonomous Vehicles (CAV). Then, we formulate pilot demand models by market segment and at the end we test and assess the pilot models for inclusion in the ActivitySim, BEAM, ands Atlas model system in development at the Lawrence Berkeley National Lab (LBNL). This project contributes to the development of new processes, policies, analytical tools, program designs, and business models to advance the state of the art in next-generation sustainable transportation solutions.
Revisiting the impact of teleworking on activity-travel behavior using recent data and sequence-based analytical techniques (January 2020 – December 2020). Sponsor: Pacific Southwest Region 9 University Transportation Center, US Department of Transportation, and UCSB. Budget: $109,520. In this project motif and sequence analysis are used in tandem to analyze differences and commonalities between telecommuters and usual commuters. Telecommuters are by far more diverse in their allocation of time to places, activities, and travel. Approximately 20% of telecommuters stay at home all day during a workday, while only 8% of commuters do. Telecommuters that have at least one trip during their workday accrue more vehicle miles travelled and number of trips than their commuter counterparts. However, they drive alone less and tend to have more complex schedules visiting more locations. A substantial proportion of traditional commuters display morning and afternoon peaks of arriving at and departing from work, and telecommuters do not show this pattern. In addition, telecommuters during a day travel to a variety of locations to either visit customers and exploit their spatio-temporal schedule flexibility to perform work tasks from locations other than home or workplaces. Similarly, seniors (60 years and older) enjoy higher activity and travel flexibility due to seniority in jobs or retirement and use telecommuting in a variety of different ways. We find that 15 distinct motifs can capture 82.17% and 86% of the total senior respondents on workdays and non-workdays, respectively. Seniors are more likely to have simple motifs with three or fewer distinct locations on non-workdays, while they present more complex motifs during workdays.
An Analysis of Accessibility, Social Interaction, and Activity-Travel Fragmentation in California (August 2018 – August 2019). Sponsor: Pacific Southwest Region 9 University Transportation Center, US Department of Transportation, and UCSB. Budget: $110,000. Sequence analysis is used in this project to measure fragmentation in activity participation and travel. Studying sequences of daily episodes (each activity at a place and each trip) is preferable over other techniques of studying activity-travel behavior because sequences include the entire trajectory of a person’s activity during a day while jointly considering the number of activities and trips, their ordering, and their durations. We first identify places visited and duration at each place on a minute-by-minute basis, then we derive representative daily behavior patterns using hierarchical clustering. Our study shows there are at least nine distinct daily patterns with different sequencing of activities and travel as well as travel time ratios and modal split. As expected, day of the week plays a major role in the type of daily activity-travel patterns. Travel time ratios are also examined for each daily pattern and we find differences in the role played within each pattern between central city, suburban, exurban, and rural dwellers. In a comparison of couples, we find systematically higher fragmentation in households that have children and their parents are employed with women showing higher fragmentation in the activity-travel patterns.
Activity Based Model for Qatar (August 2017 – April 2019). Sponsor: Ministry of Transport and Communication Qatar via ItalConsult. Budget: $610,000. In this project in collaboration with Chandra Bhat from UT Austin and Ram Pendyala from Arizona State University, a new version of an activity-based travel demand forecasting model is created for the State of Qatar. This simulator includes population synthesis that recreates the entire resident population of this region, provides locations for residences, workplaces, and schools for each person, estimates car ownership and type as well as main driver for each vehicle, and provides other key personal and household characteristics. Then, a synthetic schedule generator recreates for each resident person in the simulated region a schedule of activities and travel that reflects intra-household activity coordination for a day. These synthetic schedules are then converted to multiple Origin Destination (OD) matrices at different times in a day and used in other modeling tasks developed by ItalConsult.
Vertical Equity Statewide Pilot, Data Inventory, and Guidelines for Performance Based Planning (January 2018 – December 2018). Sponsor: Pacific Southwest Region 9 University Transportation Center, US Department of Transportation, and UCSB. Budget: $123,072. In this project, we develop a method for vertical equity analysis. Vertical equity is the analysis of the disadvantages groups of different incomes and other sociodemographic characteristics experience from the land use-transportation system. We first create a data inventory for the indicators needed to satisfy performance based planning in California. This inventory is comprehensive and covers many goals in CTP 2040. In parallel, using detailed in space and time database in the GeoTrans laboratory we create a first pilot geo-computation of equity indicators covering the entire State at fine spatial detail. We compare data available and this pilot to the literature on gentrification, equity analysis, and access to opportunities. We also develop a crowdsourcing method to collect input for the expert community in three different stages. First on the first pilot geo-computation online and at TRB 2018 and then after an improved method is developed we check with a wider spectrum of experts at an international conference in Santa Barbara. Input from the different expert sources that is used to create a second pilot geo-computation of equity is particularly important and for this reason requires special attention in a dedicated task. The project ends with the creation of guidelines for data collection, computational examples, and guidelines for research and practice.
Qatar Activity-based Model Preparation and Initial Design (May 2017 – August 2017). Sponsor: ItalConsult & Ministry of Transport and Communication Qatar. Budget: $50,000. In this project, in collaboration with Professor Chandra Bhat from UT Austin and Professor Ram Pendyala from Arizona State University, we conducted a series of meetings with all the local agencies involved in the project. We examined data and plans for data collection and developed a temporal sequence of model development and testing. Moreover, in this task will provided a series of seminars on data needs for policy analysis, modeling and simulation in activity-based approaches, synthetic population generation, household travel surveys, and the design of choice experiments.
Long Distance Travel in the California Household Travel Survey (CHTS) and Social Media Augmentation (July 2016 - June 2017). Sponsor: University of California Center /CALTRANS. Budget: $173,409. The objectives of this project are to: a) provide empirical evidence for long distance travel behavior analysis using synthetic population methods; and b) identify the determinants of long-distance travel behavior. The method provides CALTRANS with a powerful baseline inventory of travel demand statewide and creates the foundation for future updates. The archived data and their analysis also enable the study of vulnerable segments of the population as well as developing an estimate of long distance travel contribution to the statewide VMT. Our initial workshop in Sacramento illustrated the potential of these methods and the technical problems we solve in the project, and the final report explains and demonstrates the findings from CHTS, the potential of harvesting social media, the regression and structural equations models and latent class clusters developed here, their substantive findings, and a variety of ideas for using similar methods for other purposes.
Combining California Household Travel Survey data with harvested social media information to form a self-validating statewide origin-destination travel prediction method (April 2015 to March 2016). Sponsor: University of California Center /CALTRANS. Budget: $128,050. In this project we use data from multiple sources to produce statewide travel patterns and large scale estimates of induced travel demand. In this way we can develop a baseline short- and long- distance travel inventory that includes statewide vehicle miles traveled (VMT). A procedure is created to monitor the evolution of travel in California using data from social media adjusted by region and correlated with land uses at fine geographic areas. We use data from the California Household Travel Survey, Origin-Destination data from the Statewide Travel Demand Model, and social media harvested data. A conversion procedure is created to transform harvested data from the web into origin-destination travel and perform comparisons with other sources of OD matrices.
Spatial Transferability Using Synthetic Population Generation Methods (March 2015 to April 2016). Sponsor: University of California Transportation Center/CALTRANS. Budget: $84,094. In this project we develop a new method to transfer daily travel behavior data from one place to another. The basic ingredients of this new method are: a) the California Household Travel Survey (CHTS) data that includes household and person characteristics and an one day place-based diary that spans an entire 12 month period in 2012 and 2013; b) a database of all the business establishments in California that enables computation of land use indicators at many geographical scales; and c) highway and public transportation network information. In the project we develop a classification system of the different determinants of household travel behavior and then use variables at the person, household, and spatial organization levels to perform experiments and find the best transferability method.
Business Establishment Survival and Transportation System Level of Service (March 2015 to April 2016). Sponsor: University of California Transportation Center/CALTRANS. Budget: $115,205. In this research we empirically link the survival and economic success of business establishments to the performance of the transportation system that serves these establishments. We explore this relationship for the entire State of California while controlling in a statistically robust way for a variety of factors influencing business life cycle events, such as closures, formation/birth, and relocation. We use data from: a) longitudinal business establishment population event data that span two decades; b) highway and transit accessibility and level of service indicators at multiple time points; and c) geographical market size and mix from available US Census data. We draw lessons learned and develop suggested policies that increase economic development and business establishment performance.
Business Establishment Spatial Evolution Microsimulation (BESEM) (April 2012 - September 2014). Sponsor: University of California Transportation Center/CALTRANS. Budget: $44,000. The ultimate objective of this initiative is to create software that is able to replicate the change in location of each business establishment in California as a function of its relationship with other business establishments and the transportation infrastructure connecting all businesses. This is a much needed method to: a) show the spatial correlation between business location (and implicitly jobs) and infrastructure by each business type at a microlevel; and b) compute activity opportunity based accessibility indicators that capture observed changes due to businesses moving into the state, moving out of the state, and relocating from one region to another. Schemata for each business type (medical, retail, legal) will be first developed and tested with real world data using point process statistical models and measures of centrality and clustering. In addition, economic efficiency and relocation behavior is also analyzed to discern patterns of regularity/stability and change. Eventually, models of location/relocation will be developed and used in simulating urban environments. The tasks include data assembly and assessment of quality, testing of spatial statistics models, creation of the simulator framework, and testing.
SCAG Activity-based Travel Demand Model Development: Development of Simulator of Activities, Greenhouse (gas) Emissions, Networks, and Travel (SimAGENT) (April 2009 to June 2013). Sponsor: Southern California Association of Governments. Budget: $1,400,000. In this project in collaboration with Chandra Bhat from UT Austin and Ram Pendyala from Arizona State University, the requirements of California Senate Bill 375 and the regional transportation modeling guidelines are addressed by developing an activity scheduling model system and insert it into the overall model system of SCAG. The simulator includes population synthesis that recreates the entire resident population of this region, provides locations for residences, workplaces, and schools for each person, estimates car ownership and type as well as main driver for each vehicle, and provides other key personal and household characteristics. Then, a synthetic schedule generator recreates for each resident person in the simulated region a schedule of activities and travel that reflects intra-household activity coordination for a day. These synthetic activity and travel daily schedules are then converted to multiple Origin Destination (OD) matrices at different times in a day. These are in turn combined with other OD matrices (representing truck travel, travel from and to ports and airports, and travel generated outside the region) and assigned to the network in multiple periods in a day. The assignment output is then used in the software EMFAC to produce estimates of fuel consumed and pollutants emitted (including CO2) by different classes of vehicles. The overall model system also includes provision for finer spatial and temporal resolutions that is pilot tested using TRANSIMS and MATSIM. In addition, spatial allocation (geolocation) techniques are used to assign household to residential parcels and activities to all land parcels in a region. Moreover, testing of second by second vehicle emission estimation using the output of TRANSIMS and CMEM was also successful.
Development of Next Generation Agent-based Simulation (January 2009 to December 2012). Sponsor: UC Lab Fees Program. UC Office of the President. Budget: $870,000. In this project, realistic agents are created using observed and reported data from persons and their households including a variety of time use, activity participation, and travel surveys combined with large databases available from public agencies and private companies. Also key is the inclusion of weekly rhythms in the life of people, their interactions with other people within their strongest and most influential social network (i.e., the household), life cycle stages, and people’s complex interactions with the built environment. In this project, different modeling techniques are developed, tested, evaluated, and implemented to demonstrate them in applications. This project generated a base suite of tested models, provided core information for many new research proposals, strengthened the GeoTrans laboratory at UCSB, and offered unique opportunities for our graduate students in developing modeling and simulation careers.
California Household Travel Survey Pre-test Design and Management Consultant (April 2010 to June 2013). Sponsor: Southern California Association of Governments. Budget: $90,000. In this project Goulias with Dr. Morrison designed a pre-test for the California Household Travel Survey (an approximately 65,000 household survey) and developed a list of data items required for the modeling needs of California to address SB 375 policy questions today and to also prepare for new modeling needs in the future for large, medium, and small MPOs as well as CALTRANS. Goulias also supervised data collection efforts and analyzed outcomes using quality assurance and control techniques.
Forecasting with Dynamic Microsimulation: Design, Implementation, and Demonstration (August 2009 to December 2010). Sponsor: University of California Transportation Center. Budget: $102,000. In this project we develop a new travel demand forecasting system that integrates demographic microsimulation with urban simulation and travel demand model systems. The basic ingredients of this new model system are: a) a dynamic demographic simulator designed and tested with repeated observations of the same individuals in another context that will be transferred to a case study in Santa Barbara, CA; b) a modified version of the recently finalized Urbansim model that will also be calibrated with data from Santa Barbara, CA; and c) travel demand models that account for intra-household interactions and path based accessibility that were estimated with data from California. The model system is unique because it combines within a day and across years human behavior dynamics and it will push the frontier of modeling and simulation one step further. A pilot test of land use models was tested in Santa Barbara, CA, and a strategy for next steps was developed.