Characterisation and calibration of low-cost PM sensors at high temporal resolution to reference-grade performance

Florentin M J Bulot; Steven J Ossont; Andrew K R Morris; Philip J Basford; Natasha H C Easton; Hazel L Mitchell; Gavin L Foster; Simon J Cox; Matthew Loxham

doi:10.1016/j.heliyon.2023.e15943

Characterisation and calibration of low-cost PM sensors at high temporal resolution to reference-grade performance

Heliyon. 2023 Apr 29;9(5):e15943. doi: 10.1016/j.heliyon.2023.e15943. eCollection 2023 May.

Authors

Florentin M J Bulot^{1

2}, Steven J Ossont³, Andrew K R Morris⁴, Philip J Basford¹, Natasha H C Easton^{2

4

5

6}, Hazel L Mitchell¹, Gavin L Foster^{2

5

6}, Simon J Cox¹, Matthew Loxham^{2

7

8

9}

Affiliations

¹ Faculty of Engineering and Physical Sciences, University of Southampton, Southampton, UK.
² Southampton Marine and Maritime Institute, University of Southampton, Southampton, UK.
³ BizData, 278 Collins St, Melbourne, VIC, 3000, Australia.
⁴ National Oceanography Centre, Southampton, UK.
⁵ Faculty of Environmental and Life Sciences, University of Southampton, Southampton, UK.
⁶ School of Ocean and Earth Science, National Oceanography Centre, University of Southampton, UK.
⁷ School of Clinical and Experimental Sciences, Faculty of Medicine, University of Southampton, Southampton, UK.
⁸ National Institute for Health Research Southampton Biomedical Research Centre, Southampton, UK.
⁹ Institute for Life Sciences, University of Southampton, Southampton, UK.

Abstract

Particulate Matter (PM) low-cost sensors (LCS) present a cost-effective opportunity to improve the spatiotemporal resolution of airborne PM data. Previous studies focused on PM-LCS-reported hourly data and identified, without fully addressing, their limitations. However, PM-LCS provide measurements at finer temporal resolutions. Furthermore, government bodies have developed certifications to accompany new uses of these sensors, but these certifications have shortcomings. To address these knowledge gaps, PM-LCS of two models, 8 Sensirion SPS30 and 8 Plantower PMS5003, were collocated for one year with a Fidas 200S, MCERTS-certified PM monitor and were characterised at 2 min resolution, enabling replication of certification processes, and highlighting their limitations and improvements. Robust linear models using sensor-reported particle number concentrations and relative humidity, coupled with 2-week biannual calibration campaigns, achieved reference-grade performance, at median PM_2.5 background concentration of 5.5 μg/m³, demonstrating that, with careful calibration, PM-LCS may cost-effectively supplement reference equipment in multi-nodes networks with fine spatiotemporality.

Keywords: Air pollution; Calibation; Low-cost sensors; Machine learning; PM2.5; Particulate matter.