Inferring Aerosol Sources from Low-Cost Air Quality Sensor Measurements: A Case Study in Delhi, India
David H. Hagan, Shahzad Gani, Sahil Bhandari, Kanan Patel, Gazala Habib, Joshua S. Apte, Lea Hildebrandt Ruiz, and Jesse H. Kroll
Environmental Science & Technology Letters 2019 6 (8), 467-472
DOI: 10.1021/acs.estlett.9b00393
These data files are intended as additional supplementary information for this publication. If you would like to use any data herein in a publication, please contact the corresponding authors, David Hagan ([email protected]) and Jesse Kroll ([email protected]) first.
If you have any questions or comments about this data or manuscript, please contact either David Hagan or Jesse Kroll.
David H Hagan
Jesse H Kroll
The final manuscript can be downloaded directly from the ES&T Letters website, as it is open access.
Low-cost sensors (LCS) offer the opportunity for measuring urban air quality at a finer spatiotemporal scale than is currently practical with expensive research or regulatory-grade instruments. Recently, the LCS research community has largely focused on sensor calibration, pollution monitoring, and exposure assessment; here, we investigate a new application of LCS - the characterization of particulate pollution sources in an urban environment. Using an integrated, multipollutant LCS system (measuring both gases and particles), we collected air quality data for 2 winter months at a site in Delhi, India . Results were compared to measurements made by co-located research-grade particle instruments. Non-negative matrix factorization was used to deconvolve LCS data into unique factors which were then identified by examining the factor composition and comparing them to the research-grade measurements. The data were described well by three factors, a combustion factor characterized by high CO levels and two secondary factors characterized by larger particles, which agree with the results of factor analysis of online particle composition measurements. This work demonstrates that multipollutant LCS measurements, despite their inherent limitations (e.g., calibration challenges, inability to measure smallest particles), can provide insight into sources of fine particulate matter in a complex urban environment.
The environment, code, and most data used to produce this manuscript can be found on this page. The data that is currently missing (ACSM, SMPS) can be requested from Professor Josh Apte at the University of Texas at Austin. All low-cost sensor (LCS) data for this project are included here. By releasing the Dockerfile needed to build the analysis environment, there should be no issues related to software development environment (knock on wood...). The goal is to get as close as possible to reproducible, open science. If you are unfamiliar with Docker or how to use it, please check out this fantastic tutorial or the Docker documentation.