Generating traffic flow and speed regional model data using internet GPS vehicle records

Area of Science:

• Environmental Science
• Transportation Engineering
• Data Science

Background:

• Smartphones and vehicles commonly use Global Positioning System (GPS) for navigation.
• GPS data offers potential for deriving local vehicular flow and estimating emissions.
• Large volumes of GPS data present challenges in processing and analysis.

Purpose of the Study:

• To develop methods for handling large GPS datasets to extract high-resolution vehicular flow.
• To correct for spatial bias in GPS-derived traffic flow using traffic count observations.
• To inform traffic-related air quality modeling and local air pollution management.

Main Methods:

• Filtering GPS speed and acceleration data.
• Assigning buffers to the road network.
• Aggregating speed by street and filling missing lane data.
• Generating traffic flow from processed GPS data.
• Rescaling GPS traffic flow using independent traffic count data.

Main Results:

• Successfully generated high spatial resolution vehicular flow information for South-east Brazil from 124 million GPS recordings.
• Investigated and accounted for spatial bias in GPS data using local traffic count observations.
• Developed a robust description of spatial and quantitative traffic patterns through rescaled GPS traffic flow.

Conclusions:

• The presented methods effectively handle large GPS datasets to generate reliable vehicular flow data.
• The bias-corrected traffic flow data is valuable for air quality modeling and traffic management applications.
• The methodology is adaptable for various applications requiring localized or time-resolved traffic flow inputs.