Modeling Urban Air Quality Using Taxis as Sensors

TL;DR

This paper demonstrates that taxi fleets can serve as effective mobile sensors for high-resolution, city-wide air quality monitoring, enabling better environmental insights and policy decisions.

Contribution

It introduces a scalable method using taxi data for real-time urban air quality monitoring and modeling, highlighting its effectiveness and potential applications.

Findings

Taxis provide fine-grained, street-level air quality data.

Traffic and wind data can predict pollution variability effectively.

Taxi-based sensing offers scalable, near-real-time air quality insights.

Abstract

Monitoring urban air quality with high spatiotemporal resolution continues to pose significant challenges. We investigate the use of taxi fleets as mobile sensing platforms, analyzing over 100 million PM2.5 readings from more than 3,000 vehicles across six major U.S. cities during one year. Our findings show that taxis provide fine-grained, street-level air quality insights while ensuring city-wide coverage. We further explore urban air quality modeling using traffic congestion, built environment, and human mobility data to predict pollution variability. Our results highlight geography-specific seasonal patterns and demonstrate that models based solely on traffic and wind speeds effectively capture a city's pollution dynamics. This study establishes taxi fleets as a scalable, near-real-time air quality monitoring tool, offering new opportunities for environmental research and data-driven policymaking.