A USC study of the Expo Line has finally proven what transportation planners have long theorized about urban rail projects: even the most effective ones are unlikely to reduce traffic congestion.
That runs counter to the way many rail projects have been marketed to the public — a strategy the study’s authors say overlooks more valid reasons for supporting transit infrastructure.
Their findings are also the first to come from a unique big data archive run by USC, one of the largest in the country looking at urban transportation. The archive hints at a future in which “smart cities” could measure service performance with scientific detail. That would help transportation agencies make better policy decisions and decide which taxpayer-funded projects are worth an investment.
The study, published in the Journal of Planning Education and Research, was authored by Genevieve Giuliano, holder of the Ferraro Chair in Effective Local Government at the USC Price School of Public Policy; Sandip Chakrabarti, a postdoctoral researcher at METRANS Transportation Center, a joint research partnership between USC and California State University, Long Beach; and Mohja Rhoads of South Bay Cities council of Governments.
For the past four years, Giuliano’s team has worked with Cyrus Shahabi and Ugur Demiryurek of the USC Viterbi School of Engineering to establish a “big data” archive — the Archived Data Management System (ADMS). The project, funded by Los Angeles County Metro, features data collected by the agency. The data store is vast and diverse, including four years of readings from traffic sensors, bus passenger counters and accident reports.
The uncompressed archive totals about 15 terabytes, according to Demiryurek, associate director of USC Viterbi’s Integrated Media Systems Center, who built much of the ADMS archive and its algorithms.
The study data was pulled from geo-located sensors along the I-10 Freeway and arterial roads over a three-month period, checking readings at 30-second to 1-minute intervals. The team used close to 1 million records from freeway sensors and almost 16 million from arterial road sensors.
These readings revealed that Phase 1 of the Expo Line had little to any impact on local street or freeway congestion. It did, however, initiate a significant overall rise in transit ridership across the Culver City-Downtown Los Angeles corridor along the Expo Line.
Chakrabarti said the study’s findings are in line with the concept of “triple convergence” — the idea that raising overall capacity for a transportation system will only improve congestion in the short-term before returning to previous levels (or even growing worse).
This study shows us how to be more realistic in what we should expect from transit.
“Looking into the future, this study shows us how to be more realistic in what we should expect from transit,” he said. “There is value in creating quality transit for those who use it by choice or by need. It improves productivity for a lot of people.”
Effective transit projects can expand residential and employment opportunities, he added, while also providing access to services that are critical for lower-income people. In the long term, they can positively influence development patterns and also help achieve regional environmental goals.
Rick Jager, a communications manager at Metro, pointed out that the Expo Line’s success in attracting riders highlights the demand for public transit in Los Angeles.
“The first phase of the Expo Line has shattered all ridership projections, now carrying over 30,000 daily riders,” Jager said. “When Phase 2 of the project to Santa Monica opens next year, thousands of commuters on the Westside will have a reliable alternative to get out of their cars and onto mass transit.”
The power of big data
Many transportation studies still require someone standing on a corner to count the number of cars that go by or survey bus riders on their overall experience, Chakrabarti said. Very few cities across the country are systematically storing real-time data sets like this.
“We’ve never had this level of granularity,” he said. “Many cities are collecting this kind of data, but it’s seldom archived in a systematic fashion. Mostly, it’s used for real-time operations and it goes down the drain.”
Chakrabarti said the ability to do before/after studies opens new possibilities as far as measuring, monitoring and evaluating the performance of systems and new infrastructure projects.
The research by Shahabi and Demiryurek influenced how the big data is crunched and applied in predictive route planning. Demiryurek said most existing route-planning algorithms use real-time data. But the archive learns from past history, effectively making decisions using predictive data as well.
He said this represents a leap beyond navigation algorithms used by Google Maps and Waze. The USC technology has already been licensed for commercial use by a company planning to compete with both mapping services.
“If you can understand traffic and predict traffic, you can plan your schedule much better,” Demiryurek said. “Instead of detect and react, the idea is to predict and avoid. Current navigation algorithms are all in the state of detect and react.”
Being able to predict how a particular traffic intersection, bus line or train stop functions at rush hour and how it fits into the whole of LA’s transit system could change which policies officials endorse. It could also prove which taxpayer-funded projects are worth investing in.
“The next step is to transition Los Angeles into a smart city,” Demiryurek said. “Right now, we’re talking to people in the city for new projects.”