Calibration of low-cost NO2 sensors in an urban air quality network

V.M. Van Zoest (Corresponding Author), F.B. Osei, A. Stein, Gerard Hoek

Research output: Contribution to journalArticleAcademicpeer-review

2 Downloads (Pure)

Abstract

Low-cost air quality sensors measuring air quality at fine spatio-temporal resolutions, typically suffer from sensor drift and interference. Field calibration is typically performed at one location, while little is known about the spatial transferability of correction factors. We evaluated three calibration methods using a year of hourly nitrogen dioxide (NO2) observations from low-cost sensors, collocated at two sites with a conventional monitor as reference: (1) an iterative Bayesian approach for daily estimation of the parameters in a multiple linear regression model, (2) a daily updated correction factor and (3) a correction factor updated only when concentrations are uniformly low. We compared the performance of the calibration methods in terms of temporal stability, spatial transferability, and sensor specificity. We documented drift within the 1-year period. The correction factor updated under uniformly low concentrations performed poorly. The iterative Bayesian approach and daily correction factor reduced the root mean squared error (RMSE) by 21–46% at the calibration locations, but did not reduce RMSE at the other location. By examining the posterior distributions of the regression coefficients, we found that the poor spatial transferability is consistent with different responses of individual sensors to environmental factors. We conclude that the spatial and temporal variability in the calibration parameters requires them to be updated regularly, including sensor-specific recalibrations.
Original languageEnglish
Pages (from-to)66-75
Number of pages10
JournalAtmospheric environment
Volume210
Early online date25 Apr 2019
DOIs
Publication statusPublished - 1 Aug 2019

Fingerprint

air quality
sensor
calibration
cost
nitrogen dioxide
environmental factor
method
parameter

Keywords

  • ITC-ISI-JOURNAL-ARTICLE

Cite this

@article{27380ee2b50245b68be5c2c8f4b68554,
title = "Calibration of low-cost NO2 sensors in an urban air quality network",
abstract = "Low-cost air quality sensors measuring air quality at fine spatio-temporal resolutions, typically suffer from sensor drift and interference. Field calibration is typically performed at one location, while little is known about the spatial transferability of correction factors. We evaluated three calibration methods using a year of hourly nitrogen dioxide (NO2) observations from low-cost sensors, collocated at two sites with a conventional monitor as reference: (1) an iterative Bayesian approach for daily estimation of the parameters in a multiple linear regression model, (2) a daily updated correction factor and (3) a correction factor updated only when concentrations are uniformly low. We compared the performance of the calibration methods in terms of temporal stability, spatial transferability, and sensor specificity. We documented drift within the 1-year period. The correction factor updated under uniformly low concentrations performed poorly. The iterative Bayesian approach and daily correction factor reduced the root mean squared error (RMSE) by 21–46{\%} at the calibration locations, but did not reduce RMSE at the other location. By examining the posterior distributions of the regression coefficients, we found that the poor spatial transferability is consistent with different responses of individual sensors to environmental factors. We conclude that the spatial and temporal variability in the calibration parameters requires them to be updated regularly, including sensor-specific recalibrations.",
keywords = "ITC-ISI-JOURNAL-ARTICLE",
author = "{Van Zoest}, V.M. and F.B. Osei and A. Stein and Gerard Hoek",
year = "2019",
month = "8",
day = "1",
doi = "10.1016/j.atmosenv.2019.04.048",
language = "English",
volume = "210",
pages = "66--75",
journal = "Atmospheric environment",
issn = "1352-2310",
publisher = "Elsevier",

}

Calibration of low-cost NO2 sensors in an urban air quality network. / Van Zoest, V.M. (Corresponding Author); Osei, F.B.; Stein, A.; Hoek, Gerard.

In: Atmospheric environment, Vol. 210, 01.08.2019, p. 66-75.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Calibration of low-cost NO2 sensors in an urban air quality network

AU - Van Zoest, V.M.

AU - Osei, F.B.

AU - Stein, A.

AU - Hoek, Gerard

PY - 2019/8/1

Y1 - 2019/8/1

N2 - Low-cost air quality sensors measuring air quality at fine spatio-temporal resolutions, typically suffer from sensor drift and interference. Field calibration is typically performed at one location, while little is known about the spatial transferability of correction factors. We evaluated three calibration methods using a year of hourly nitrogen dioxide (NO2) observations from low-cost sensors, collocated at two sites with a conventional monitor as reference: (1) an iterative Bayesian approach for daily estimation of the parameters in a multiple linear regression model, (2) a daily updated correction factor and (3) a correction factor updated only when concentrations are uniformly low. We compared the performance of the calibration methods in terms of temporal stability, spatial transferability, and sensor specificity. We documented drift within the 1-year period. The correction factor updated under uniformly low concentrations performed poorly. The iterative Bayesian approach and daily correction factor reduced the root mean squared error (RMSE) by 21–46% at the calibration locations, but did not reduce RMSE at the other location. By examining the posterior distributions of the regression coefficients, we found that the poor spatial transferability is consistent with different responses of individual sensors to environmental factors. We conclude that the spatial and temporal variability in the calibration parameters requires them to be updated regularly, including sensor-specific recalibrations.

AB - Low-cost air quality sensors measuring air quality at fine spatio-temporal resolutions, typically suffer from sensor drift and interference. Field calibration is typically performed at one location, while little is known about the spatial transferability of correction factors. We evaluated three calibration methods using a year of hourly nitrogen dioxide (NO2) observations from low-cost sensors, collocated at two sites with a conventional monitor as reference: (1) an iterative Bayesian approach for daily estimation of the parameters in a multiple linear regression model, (2) a daily updated correction factor and (3) a correction factor updated only when concentrations are uniformly low. We compared the performance of the calibration methods in terms of temporal stability, spatial transferability, and sensor specificity. We documented drift within the 1-year period. The correction factor updated under uniformly low concentrations performed poorly. The iterative Bayesian approach and daily correction factor reduced the root mean squared error (RMSE) by 21–46% at the calibration locations, but did not reduce RMSE at the other location. By examining the posterior distributions of the regression coefficients, we found that the poor spatial transferability is consistent with different responses of individual sensors to environmental factors. We conclude that the spatial and temporal variability in the calibration parameters requires them to be updated regularly, including sensor-specific recalibrations.

KW - ITC-ISI-JOURNAL-ARTICLE

UR - https://ezproxy2.utwente.nl/login?url=https://doi.org/10.1016/j.atmosenv.2019.04.048

UR - https://ezproxy2.utwente.nl/login?url=https://library.itc.utwente.nl/login/2019/isi/zoest_cal.pdf

U2 - 10.1016/j.atmosenv.2019.04.048

DO - 10.1016/j.atmosenv.2019.04.048

M3 - Article

VL - 210

SP - 66

EP - 75

JO - Atmospheric environment

JF - Atmospheric environment

SN - 1352-2310

ER -