Integrating large-scale stationary and local mobile measurements to estimate hyperlocal long-term air pollution using transfer learning methods

Zhendong Yuan*, Jules Kerckhoffs, Youchen Shen, Kees de Hoogh, Gerard Hoek, Roel Vermeulen

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

Mobile air quality measurements are collected typically for several seconds per road segment and in specific timeslots (e.g., working hours). These short-term and on-road characteristics of mobile measurements become the ubiquitous shortcomings of applying land use regression (LUR) models to estimate long-term concentrations at residential addresses. This issue was previously found to be mitigated by transferring LUR models to the long-term residential domain using routine long-term measurements in the studied region as the transfer target (local scale). However, long-term measurements are generally sparse in individual cities. For this scenario, we propose an alternative by taking long-term measurements collected over a larger geographical area (global scale) as the transfer target and local mobile measurements as the source (Global2Local model). We empirically tested national, airshed countries (i.e., national plus neighboring countries) and Europe as the global scale in developing Global2Local models to map nitrogen dioxide (NO 2) concentrations in Amsterdam. The airshed countries scale provided the lowest absolute errors, and the Europe-wide scale had the highest R 2. Compared to a "global" LUR model (trained exclusively with European-wide long-term measurements), and a local mobile LUR model (using mobile data from Amsterdam only), the Global2Local model significantly reduced the absolute error of the local mobile LUR model (root-mean-square error, 6.9 vs 12.6 μg/m 3) and improved the percentage explained variances compared to the global model (R 2, 0.43 vs 0.28, assessed by independent long-term NO 2 measurements in Amsterdam, n = 90). The Global2Local method improves the generalizability of mobile measurements in mapping long-term residential concentrations with a fine spatial resolution, which is preferred in environmental epidemiological studies.

Original languageEnglish
Article number115836
Number of pages10
JournalEnvironmental Research
Volume228
DOIs
Publication statusPublished - 1 Jul 2023

Bibliographical note

Funding Information:
The project received funding from the Environmental Defense Fund (United States), Google (United States), EXPOSOME- NL (NWO; project number 024.004.017) and EXPANSE (EU-H2020 Grant number 874627). This work made use of the Dutch national e-infrastructure with the support of the SURF Cooperative using grant no. EINF-1923.The authors gratefully acknowledge the help and coordination (data, discussions) of Karin Tuxen-Bettman and Natalie Smailou, Google, United States. Help, support, and coordination (discussions) of Harry van Bergen, Paul Coops, and Imke van Moorselaar, the Municipality of Amsterdam. We thank Kees Meliefste for the design and instrumental setups of the car and processing of data by Fares Al Hasan.

Funding Information:
The project received funding from the Environmental Defense Fund (United States) , Google (United States) , EXPOSOME- NL (NWO; project number 024.004.017 ) and EXPANSE (EU-H2020 Grant number 874627 ). This work made use of the Dutch national e-infrastructure with the support of the SURF Cooperative using grant no. EINF-1923 .

Publisher Copyright:
© 2023 The Authors

Keywords

  • Air pollution mapping
  • Google street view
  • Mobile monitoring
  • Transfer learning

Fingerprint

Dive into the research topics of 'Integrating large-scale stationary and local mobile measurements to estimate hyperlocal long-term air pollution using transfer learning methods'. Together they form a unique fingerprint.

Cite this