Low-Cost Particulate Matter Sensors for Monitoring Residential Wood Burning

Conventional monitoring systems for air quality, such as reference stations, provide reliable pollution data in urban settings but only at relatively low spatial density. This study explores the potential of low-cost sensor systems (LCSs) deployed at homes of residents to enhance the monitoring of urban air pollution caused by residential wood burning. We established a network of 28 Airly (Airly-GSM-1, SP. Z o.o., Poland) LCSs in Kristiansand, Norway, over two winters (2021–2022). To assess performance, a gravimetric Kleinfiltergerät measured the fine particle mass concentration (PM2.5) in the garden of one participant’s house for 4 weeks. Results showed a sensor-to-reference correlation equal to 0.86 for raw PM2.5 measurements at daily resolution (bias/RMSE: 9.45/11.65 μg m–3). High-resolution air quality maps at a 100 m resolution were produced by combining the output of an air quality model (uEMEP) using data assimilation techniques with the network data that were corrected and calibrated by using a proposed five-step network data processing scheme. Leave-one-out cross-validation demonstrated that data assimilation reduced the model’s RMSE, MAE, and bias by 44–56, 38–48, and 41–52%, respectively.

with observations 11 , 59.The model and/or its underlying input data do not fully capture the variations of air pollution at the high spatial resolution, and the assimilation of sensor data into the model can provide better estimates of the spatiotemporal variation of air pollution at regional scales.

SI.2 Airly sensor systems
The Airly sensor kit measures temperature and RH with a DHT22 temperature/RH sensor.The Airly unit delivers hourly averaged data; however, the original sampling rate is one measurement per five minutes.The data is communicated using the Global System for Mobile Communications

Description
The Small Filter Devices (Kleinfiltergeräte) LVS3 and MVS6 are designed for outdoor use at very high as well as very low temperatures.The devices can also be used indoors.
The flow rates of the samplers are controlled in compliance with basic physical principles by means of a temperature-and pressure-compensated orifice plate according to Bernoulli's law and by conversion into operating-m³/h resp.standard-m³/h according to BoyleMariotte`s law.The sampled air volume is displayed in operating-m³ and standard-m³ with a sensitivity of 0,01 m³ on the digital display.In case of a pressure drop across the filter of more than 300 mbar the device will automatically shut down.
All relevant data are shown on the display and can be stored on a memory stick.In case of a power failure, all data stored in the micro controller and in the system's memory will be safe for several years thanks to a built-in high-capacity battery.The device's housing consists of stainless-steel sheet metal with a lockable door.The device's solid construction guarantees a high availability.
Because of their low noise emission level, the Small Filter Devices can be used in urban areas at any time of the day and indoors as well.The sampling head can also be set up externally, e.g.,.directly at kerbsides or in living rooms by using a hose connection to the instrument placed in a greater distance.

Subject to alterations Ed. 07/11
Model Variations

LVS3
This model can be operated with controlled flow rates between 1,0 and 2,3 m³/h.In the uncontrolled mode (UMODE), the device is identical with regard to its function to the type GS 050/3 described in the guideline VDI 2463 Part 7.

MVS6
This model can be operated with controlled flow rates between 2,3 and 3,5 m³/h.Its design is identical with the model LVS3.The controlled flow rate of 2,7 m³/h meets the requirements of VDI 2463 parts 7 and 8.

Inlets
For ambient air, indoor and workplace measurements The dust collected on the filters can also be analysed on ions (sulphate, nitrate etc.) as well as radioactivity.

(
GSM) and transferred to a Norwegian Institute for Air Research (NILU) sensor data platform for further analysis (https://nordicpathlive.nilu.no/,accessed in Nov 2022).Additionally, the sensor system is already assembled/mounted and must only be connected to the power supply.It has a weatherproof casing with dimensions of 74 × 77 × 83.5 mm and a weight of 440g.According to the manufacturer, the sensor system can measure PM with a declared accuracy of ±10 µg m -3 in the three fractions; it operates in the temperature range between -10 and +60 °C and RH range of 0 -100%.Airly (optical) PM sensors measure the micro-particle concentration (aerodynamic diameters of 0.3 -10 μm) through the light-scattering principles; the sensor's effective detection range is 0 to 500 µg m -3 .Among the early and limited studies evaluating the performance of Airly PM LCSs, Vogt, Schneider, Castell and Hamer 25 analyzed the performance of three Airly sensors against reference instrument FIDAS 200 (Palas GmbH, Germany) at Kirkeveien air quality station in Oslo (2020-09-09 until 2020-10-19); the results showed a post-calibration MB and RMSE between 1.37 and 7.89 µg m -3 and 4.39 and 10.9 µg m -3 for PM2.5, respectively.However, the Airly PM sensor relatively showed a post-calibration poor performance measuring PM10 (MB = 3.19 -9.71 µg m -3 , and RMSE = 5.26 -12.15 µg m -3 ).This behavior is typical for other PM sensor systems using light scattering methods, and recent literature shows that PM LCSs have better performance for PM2.5 than PM10 37 , 60.However, all the evaluation metrics in Vogt, Schneider, Castell and Hamer 25 analysis, such as MB or RMSE, are based on the calibrated data, not the raw output of the sensor.Correspondingly, the results of applying the sensor-specific multilinear regression models by Vogt, Schneider, Castell and Hamer 25 to three Airly sensor systems against reference gravimetric method in the field co-location at the Kirkeveien also showed that sensors overestimate the 24-h averages of PM2.5 and division by scale factors of 2.08, 1.87, and 2.46 is required to correct the data (Theil slope with an intercept of zero).For comparison, we found a slope scale factor of 1.88, with intercept equal to zero.The evaluation of the dependency of the sensors on RH against reference-equivalent optical instrument FIDAS 200 (Palas GmbH, Germany) showed a change in Bias for RH exceeding 70%, with the most considerable Bias being 5 -10 µg m -3 when the RH reaches a value of 100%.During a co-location campaign between 2021-01-22 and 2021-02-10, three Airly PM LCSs were installed by Hofman, Peters, Stroobants, Elst, Baeyens, Van Laer, Spruyt, Van Essche, Delbare and Roels 31 on the roof of a regulatory air quality station in Antwerp, Belgium equipped with a FIDAS 200, Palas reference-grade instrument.They found slope calibration factors (Sensor = scale factor × Ref) of 1.91, 1.63, and 1.90 for the three sensors' PM2.5 measurements (sensor-tosensor rs between the three Airly sensors were between 0.986 and 0.989).The pre-calibration rs, MAEs, and RMSEs between the three Airly sensors and the reference measurements were 0.89 -0.92, 9.63 -13.58 µg m -3 , and 11.72 -16.45 µg m -3 , respectively.However, they ran another calibration campaign in late Spring and early summer (2021-05-18 until 2021-06-08) and found that sensors underestimated the PM2.5 and the slope scale factors were respectively 0.75, 0.70, and 0.73 (sensor-to-sensor rs between the three Airly sensors were between 0.921 and 0.956.We have access to the data of a co-location campaign at Bjørndalssletta reference station equipped with Grimm EDM180 optical dust monitor in Kristiansand during the summer of 2022(2022-07-25   until 2022-08-13), represented in Supplementary Figure20a.We also analyzed the results of the co-location at Haga station in Gothenburg during the same period in the summer of 2022 (Supplementary Figure20b).We see similar results here, i.e., overall better performance and underestimation of PM2.5 during the summer.Hofman, Peters, Stroobants, Elst, Baeyens, Van Laer, Spruyt, Van Essche, Delbare and Roels 31 attribute this to less RH during the summer.-weight construction for outdoor use (stainless steel), small set-up area The device will be automatically heated (Winter operation) and ventilated Controlling of operating-m³/h (ambient air conditions) and standard-m³/h (0 °C or 20 °C, 760 mm Hg) by an orifice plate Impactor inlets with exchangeable jets (8 pieces) for PM10 -PM4,0 -PM2,5 -PM1,0 Inlets for TSP, PU foam (with and without ozone denuder) and bioaerosols Easy and self-explanatory 3-key menu-guided operation Data storage on memory stick

Supplementary Table 1: Airly low-cost Particulate Matter sensors operated by citizens in Kristiansand, Norway.
Sampling resolution: 3,600s.28 of 30 sensors were analyzed in this study.Zone code 1 = Grim, Zone code 2 = Kvadraturen, and Zone code 3 = Lund.

Table 2 : Studies on spatial correlation of PM2.5 in urban environments.
Ye, Ji, Yang, Yao, Chan, Cadle, Chan and Mulawa 56 Shanghai, China Two sites, Tongji University and Hainan Road, Starting in March 1999 and was conducted for one year Average concentrations over the period at the two sites were 67.6 and 64.6 μg m -3 , with a high correlation value between sites (r > 0.94) consumption, available heating technology, location of chimneys, and meteorology 11.Residential emissions have been extensively validated for Norwegian cities by comparing modeling results