Portable direct-reading instruments by light-scattering method are increasingly used in airborne fine particulate matter (PM2.5) monitoring. However, there are limited calibration studies on such instruments by applying the gravimetric method as reference method in field tests.
An 8-month sampling was performed and 96 pairs of PM2.5 data by both the gravimetric method and the simultaneous light-scattering real-time monitoring (QT-50) were obtained from July, 2015 to February, 2016 in Shanghai. Temperature and relative humidity (RH) were recorded. Mann-Whitney U nonparametric test and Spearman correlation were used to investigate the differences between the two measurements. Multiple linear regression (MLR) model was applied to set up the calibration model for the light-scattering device.
The average PM2.5 concentration (median) was 48.1μg/m3 (min-max 10.4–95.8μg/m3) by the gravimetric method and 58.1μg/m3 (19.2–315.9μg/m3) by the light-scattering method, respectively. By time trend analyses, they were significantly correlated with each other (Spearman correlation coefficient 0.889, P<0.01). By MLR, the calibration model for the light-scattering instrument was Y(calibrated) = 57.45 + 0.47 × X(the QT – 50 measurements) – 0.53 × RH – 0.41 × Temp with both RH and temperature adjusted. The 10-fold cross-validation R2 and the root mean squared error of the calibration model were 0.79 and 11.43 μg/m3, respectively.
Citation: Shi J, Chen F, Cai Y, Fan S, Cai J, Chen R, et al. (2017) Validation of a light-scattering PM2.5 sensor monitor based on the long-term gravimetric measurements in field tests. PLoS ONE 12(11): e0185700. https://doi.org/10.1371/journal.pone.0185700
Editor: Alexander Larcombe, Telethon Institute for Child Health Research, AUSTRALIA
Received: February 21, 2017; Accepted: September 18, 2017; Published: November 9, 2017
Copyright: © 2017 Shi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information file.
Funding: The study was supported by National Key R&D Program of China (2017YFC0702701), National Key R&D Program of China (2016YFC0207400), the State Key Basic Research Program (973) project (2015CB553404), Public Scientific Research Project of Ministry of Environmental Protection of China (No. 201409080) and Shanghai 3-Year Public Health Action Plan (GWTD2015S04).
Competing interests: The authors have declared that no competing interests exist.
Airborne particulate matter with the aerodynamic diameter equal to or less than 2.5μm(PM2.5) has been linked to respiratory or cardiovascular diseases and all-cause mortality in epidemiological studies worldwide including developing countries [1–4]. Exposure assessment is crucial for the accurate estimate on the PM2.5 health effects. Many previous epidemiological studies relied on the ground fixed-site monitoring stations[5, 6] and were unavoidably with potential exposure misclassification. The monitoring of PM2.5 in microenvironment or by personal monitoring is a good supplement to the ambient fixed-site monitoring. However, by a lack of calibration on these portable PM2.5 instruments, their accuracy and reliability are often unknown which limit their application in scientific research.
The portable and direct-reading PM2.5 instruments are generally based on the light-scattering and particle absorbance theories. The particle number concentration is counted and then transferred to mass concentration as the output. The measurements of particles by light-scattering method, however, are often affected by water vapor or droplets in the air which can be assessed by measuring the high relative humidity (RH). Previous results showed that light-scattering technology tends to overestimate particulate levels when compared with the gravimetric method, especially under higher RH [8–13]. Other factors such as the particle size distribution, particle morphology and chemical constituents also influence the measurements of PM2.5, of which the influence magnitude may vary by different pollution sources. So it is necessary to set up the calibration model for the light-scattering instrument based on the local long-time sampling with temperature and RH adjusted. By comparing with the filter-based gravimetric measurements, regarded as the reference standard method, the calibration model was constructed in this study based on an 8-month parallel field test between the light-scattering method and the gravimetric method.
The objectives of this study were to set up the calibration model by investigating the relationship between the measurements by the light-scattering PM2.5 monitor (QT-50, Hivron, Beijing, China) and by the parallel gravimetric measurements, and further establish the method of field calibration for low-cost optical sensors. We are going to investigate the factors influencing the differences between the two sets of measurement data and eventually construct a validation model for the PM2.5 sensor instrument with temperature and RH adjusted, based on long-time field tests. Because the temperature and RH are largely different between indoor and outdoor environment, both indoor and outdoor samplings and comparisons were conducted at the same time to enable the validation model be applicable in a variety of environment.
Materials and methods
PM2.5 sensor instrument
The light-scattering PM2.5 sensor instruments (QT-50, Hivron, Beijing, China) were pocket-size (weight<300g, volume<510cm3). It consisted of a PM sensor (DS-01D-V1), a microprocessor, a real-time clock, a data logger, a temperature and relative humidity sensor, a network module, and a small light emitting diode (LED) display screen. In the PM2.5 monitoring sensor(DS-01D-V1), an infrared light emitting diodes at 650 nm was used as the light source and a photodiode detector was used as the detector for scattered lights at a scattering angle of 90°(Patent number: CN201430196873.5). The schematics of QT50 was presented in Fig 1 as follows. QT-50 has a fan installed for a better airflow. This is different with some other sensors, such as Shinyei and Samyoung[14, 15]. The Shinyei and Samyoung PM sensors were on the principle by electrically heating a resistor near the sensor inlet to sample particles through sensing volumes. In this study, QT-50 instruments were connected to 220 V wall outlet power during the long-term monitoring.
RH-calibrated PM2.5 sensor
The MicroPEM (RTI International, Research Triangle Park, NC, USA) is used as a RH-calibrated light-scattering instrument in this study. MicroPEM is a wearable PM personal exposure monitor based on laser light-scattering method, which is able to monitor the PM mass concentration both in a real-time mode and in collecting PM2.5 samples on filters. The readout of MicroPEM is based on its build-in calibration curve, the principles and validation of this instrument have been given in detail elsewhere [16–18]. MicroPEM was fitted with a two-stage mini-impactor set with initial and final cut points at 4.0μm and 2.5μm, respectively. Before each sampling, MicroPEM was zeroed with an in-line HEPA filter, and pre-calibrated at 0.50 LPM by a TSI model 4100 mass flowmeter (TSI, Inc., Shoreview, MN, USA) by using Docking Station software (RTI International, Research Triangle Park, NC, USA). MicroPEM overcomes the disturbance of water droplets or water vapor by applying the internal RH correction coefficient.
Gravimetric filter sampling
SKC pumps (Aircheck Sampler, Model 224-PCXR8, SKC Inc., PA, USA) were used to collect the PM2.5 samples using a 2.5μm impactor (SKC Inc., PA, USA). A flow rate calibration chamber (SKC Gulf Coast Inc., Houston, TX, USA) was used before and after the sampling to facilitate the measurements of the flow rate (Liter per minute, LPM). For each sampling, a 37mm Quartz filter (Pall Corp., Ann Arbor, MI, USA) was pre-heated at 900°C for 3h to remove carbon contamination. All Quartz filters were pre- and post-weighed in a temperature and RH constant room using a Mettler-Toledo AG285 electronic microbalance with ±0.01mg sensitivity. Filters were repeatedly weighed until the differences between replicate weights were <20 mg and <10 mg for samples and blanks, respectively. The number of field blank samples (which were placed into the filter holders in the field but did not sample any air) accounted for 15% of all gravimetric samples. The integrated gravimetric PM2.5 mass concentration was calculated by dividing the net mass changes(μg) by the total air volume sampled(m3). To minimize the evaporation of volatile components, sampled filters were stored at 4°C until analysis.(1)
- ρ:Average particulate matter mass concentration, μg/m3;
- w1:Mass of the membrane before sampling, μg;
- w2:Mass of the membrane after sampling, μg;
- Vn:Sampling volume transformed in standard state(273.15K,101.325KPa), L.
The PM2.5 concentrations were monitored by 3 types of instruments for 8 continuous months (from July, 2015 to February, 2016) both indoor and outdoor in 3 representative locations. Totally 96 measurements were completed, twice a month in each location both indoor and outdoor simultaneously (2 x 8 x 2 x 3 = 96).
The 3 types of instruments were 1) PM2.5 sensor instrument (QT-50), 2) PM2.5 sensor reference instrument (MicroPEM) and 3) the gravimetric method (by SKC pumps) as described above. To be representative of the ordinary indoor environment, the 3 locations were selected in the urban area of Shanghai: 1) a university office in the Jiangwan Campus of Fudan University, 2) a 4-person dormitory room in the Jiangwan Campus of Fudan University and 3) a residential building (on the 8th floor), 2 km far from Jiangwan Campus of Fudan University. All three instruments were put in the same locations at the same height to the floor. Temperature and RH during the whole sampling were measured and recorded by the HOBO data logger (U12-012, Onset Computer Corporation, Pocasset, MA, USA).
For the two filer samples in each location in each month, one sample was performed in 4 continuous weekdays (the inlet flow rate was 2L/min) and the other was in the weekend for 2 continuous days (the inlet flow rate was 4L/min matched with the impactor). There were eventually 54 filter samples of PM2.5 obtained successfully excluding the missing data due to instrumental failure or filter membrane damage. For the PM2.5 sensor instruments (QT-50 and MicroPEM), data were continuously monitored and recorded every 10min. According to the beginning and ending time of the filter sampling, the corresponding real-time measurement data by QT-50 or MicroPEM were calculated into the 2-day or 4-day average values in order to be matched and comparable with the integrated average value by the gravimetric method. The data in 3 locations were polled together to set up the calibration model of QT-50 instrument.
The original real-time monitoring data (every 10 min) by QT-50 and MicroPEM were firstly calculated into the 24-hour average level. The 24-hour average data were then paired by time between the two sensor instruments. For the comparisons between the sensor monitoring data and the gravimetric data, the 24-hour sensor monitoring data were further calculated into the integrated average level corresponding to each specific sampling time period of the gravimetric sampling as mentioned above. Eventually 75 pairs of data were obtained between the QT-50 instrument and the gravimetric method and 54 groups of complete data were obtained with all 3 types of instruments.
Mann-Whitney U nonparametric test was used to compare the differences between 3 groups of data since PM2.5 concentrations were abnormally distributed. Spearman correlation analyses were performed between any two measurements of QT-50, MicroPEM and gravimetric measurement data. The calibration model for QT-50 was set up by multiple linear regression with the gravimetric measurements (the reference method) as the dependent variable and QT-50 data as the independent variables, adjusted for temperature and RH. The similar model was set up for MicroPEM as for a comparison. The 10-fold cross-validation (10-fold CV) was performed to validate the multiple linear regression model for QT-50. The overall fit R2 and root mean squared error (RMSE) between the predicted and measured concentrations of CV were calculated to evaluate the model performance . All analyses were conducted with R software (Version 2.15.3, R Development Core Team) and SPSS (Version 22.0, IBM).
In the 8-month sampling, the indoor median level of PM2.5 was 47.8 μg/m3 (19.2–135.1μg/m3) by the real-time QT-50 sensor instrument, 36.7 μg/m3 (11.7–142.3μg/m3) by the real-time MicroPEM instrument and 39.7 μg/m3(10.4–95.8μg/m3) by the gravimetric method (Table 1). The indoor levels measured by QT-50 were higher than the ones by MicroPEM and the gravimetric method. For outdoor air PM2.5, the ranking of the concentrations by the 3 instruments were similar with the indoor ones. By polling the indoor and outdoor measurement data together, QT-50 measurements were significantly higher than MicroPEM and the gravimetric method. In the whole sampling, the median indoor temperature and RH was 20.5°C and 61.6% and the outdoor ones were 14.4°C and 54.3%.
Correlations between QT-50, MicroPEM and the gravimetric method
There were high correlations between the measurements by QT-50, MicroPEM and the gravimetric method. By comparing the 54 groups of matched data by 3 instruments, the spearman correlation coefficients were 0.891 (P = 0.0001) between QT-50 and MicroPEM (the hourly average data, n = 6549), 0.817 (P = 0.0001) between QT-50 and the gravimetric data (n = 54) and 0.875 (P = 0.0001) between MicroPEM and the gravimetric method (n = 54) (Fig 2).
To evaluate and validate QT50 for PM2.5 in field measurements, parallel sampling was performed by QT50, an established light-scattering instrument (MicroPEM) and by gravimetric method. QT50 and MicroPEM recorded PM2.5 data for every 10 min continuously and the filters were sampled for 2 days in the weekend or 4 days in weekdays. According to the beginning and ending time of the filter sampling, the real-time measurement data by QT-50 or MicroPEM were calculated into the 2-day or 4-day average values to be matched and comparable with the integrated average value by the gravimetric method. Eventually, 54 pairs of matched values were obtained and presented in Fig 2. The date points refer to each paired average values. Y-axis refers to PM2.5 average levels and X-axis refers to sampling date.
When analyzing the data distribution of PM2.5 by temperature and RH, it was found that (Fig 3A), with the increase of RH in the air, there was a significant increase of PM2.5 by QT-50. However, no such a trend was observed in the data by MicroPEM and the gravimetric method. On the other hand, with the increase of temperature, no obvious trend was observed for PM2.5 (Fig 4A).
The distribution of PM2.5 concentration by QT-50, MicroPEM and gravimetric method at different RH before QT-50 calibration (A) and after calibration (B). In both Fig 3(A) and Fig 3(B), each point refers to a PM2.5 concentration data and the lines refer to the PM2.5 trend by RH. The points by QT-50 and MicroPEM were the original 10-min data of PM2.5. The points indicated as “Gravimetric” were the integrated average in each sample by SKC pumps. Y axis refers to the PM2.5 concentration level (μg/m3) and X axis refers to RH.
The distribution of PM2.5 concentrations by QT-50, MicroPEM and gravimetric methods at different temperatures before QT-50 calibration (A) and after calibration (B). In both Fig 4A and 4B, each point refers to a PM2.5 concentration data and the lines refer to the PM2.5 trend by temperature. The points by QT-50 and MicroPEM were the original 10-min data of PM2.5. The points indicated as “Gravimetric” were the integrated average in each sample by SKC pumps. Y axis refers to the PM2.5 concentration level (μg/m3) and X axis refers to temperature.
Calibration of the QT-50 instrument
Between the QT-50 instrument and the gravimetric method, a total of 75 pairs of data were obtained (S1 File). In comparison with the gravimetric data, QT-50 measurements were on average 1.51 (SD 0.66, range 0.58–4.46) times of the gravimetric data. In order to find out the appropriate calibration model for QT-50, a stepwise linear regression was performed. Firstly, the basic regression model was set up with the gravimetric data as the Y and QT-50 data as the X. In order to adjust the influence of RH on PM2.5, the second regression was performed with RH added in the basic model, and finally, both RH and temperature were added in the regression model (Table 2).
In this model, the R2 was 0.75 and the 95%CI of the slope (0.48) was 0.42–0.55. Compared with the theoretically ideal measurements, e.g. Y = X, it demonstrated that the differences (d = 0.52x-14.15) between the gravimetric and the QT-50 data were from both the system bias but also associated with the PM2.5 concentration levels.
With RH and temperature added in, the R2 and the 10-fold CV R2 of the model increased to 0.80 and 0.79, respectively, while the RMSE between the predicted and measured concentrations of CV decreased to 10.87μg/m3 and 11.43μg/m3, respectively (Table 2). A higher R2 and a lower RMSE R2 was obtained, which indicated a more fit regression model was achieved. Therefore, the QT-50 calibration model was determined as follows: (3) where Y refers to the calibrated PM2.5 levels and X was the original PM2.5 measured values by the QT-50 monitor.
QT-50 data comparisons before and after RH calibration
To test how the original PM2.5 data measured by QT-50 were improved after applying the calibration model, further comparison analyses were performed. Firstly, the median and the range of calibrated QT-50 data became closer to the level of the gravimetric data (Fig 5). By calculation, the ratio of calibrated QT-50 and gravimetric measurements decreased from 1.51(0.66) to 1.09(0.38). Secondly, the potential measurement bias by RH and temperature were significantly corrected. In the scattering plot of QT-50 after calibration, the PM2.5 concentration was not increased with the increase of RH anymore (Fig 3B), and the PM2.5 level was much closer to gravimetric measurement across different temperature (Fig 4B). Thirdly, by plotting the QT-50 data against the gravimetric data, the QT-50 calibration line was much closer to the theoretically ideal line of Y = X compared to the original QT-50 data line (Fig 6). The correlation between QT-50 and gravimetric measurement was improved by the increased adjusted R2 from 0.75 to 0.81 (Y = 0.81X+8.94). The MicroPEM data, which were considered as the reference light-scattering data, showed the adjusted R2 of 0.81 (Y = 1.29X-10.27).
Abbreviation: QT50 calibrated, QT50 calibrated by model, QT50 cf, QT50 calibrated by correction factor (CF), CF = 1 + 0.25 * RH * RH/(1 − RH), if RH>6.
Y axis refers to PM2.5 concentration based on the light-scattering method, and X axis refers to PM2.5 concentration measured by gravimetric instrument. The red line Y = X is achieved when light-scattering measurements (QT50, MicroPEM) were in an ideally perfect agreement with the gravimetric measurement.
In this study, a parallel 8-month sampling by both the light-scattering method (QT-50, MicroPEM) and the gravimetric method was performed. It was found that the light-scattering instrument overestimated the PM2.5 compared with the gravimetric method and the monitoring data were influenced by RH and temp. A calibration model (Y = 57.45 + 0.47 × X − 0.53 × RH − 0.41 × Temp) was further constructed for QT-50 by multiple linear regression against the gravimetric method with RH and temperature adjusted.
By comparing with the gravimetric method, the PM2.5 levels measured by QT-50 were on average 1.5 times higher. Further analyses suggested that RH was a major potential influencing factor and was adjusted in the final calibration model. In previous research [20–24], to control the influence of RH, a RH correction factor was used. The RH correction factor (CF) for light-scattering instrument was calculated as follows if the RH>60% : (4) In this study, we compared the data by using the RH correction factor when RH >60% with the calibrated data by regression models. By calibration using the CF and our regression model, the average QT-50 PM2.5 levels were 44.62 μg/m3 and 44.27 μg/m3 (Fig 5), respectively. The adjusted R2 was 0.79 and 0.81, respectively (Fig 6). The ratios between the CF corrected data/gravimetric data and the model calibrated data/gravimetric data were both 1.09. All these results indicated the high consistency between these two correction methods. Also, some other studies applied the model calibration. Ying Zhu et al showed the R2 between DustTrak (TSI Inc, Shoreview, MN, USA) and gravimetric measurement was 0.86, and Avril Challoner et al demonstrated the R2 of the above two different instruments was 0.50.
On the other hand, the data by MicroPEM instrument were calibrated for RH from a build-in program. So it was reasonable that no significant trend was observed between PM2.5 by MicroPEM and RH (Fig 3). The calibrated QT-50 data were comparable with the data by MicroPEM with the same adjusted R2 (Fig 6). Considering the convenience applied in the field tests and the higher cost-effectiveness, the QT-50 instrument can be superior with no need of filter weighing and treatment as necessitated by gravimetric methods.
Our calibration model was derived from a long-term field test covering a wide seasonal variation and meteorological conditions. In order to demonstrate the field calibration for a light-scattering sensor in a natural environment of interest, so that the findings and methodology may be extended and replicated by researchers who are interested in the utility of low-cost sensors such as QT-50. The method of field calibration for this class of sensors was also conducted in California at a regulatory monitoring site. A good linear relationship between the PM2.5 mass concentration and the responses of low-cost optical PM sensors were also reported by previous studies [15, 28–30].
Calibrations should be conducted under different seasonal and environmental conditions to test how well this calibration model hold . Limitations of our study should be noted. Firstly, the sample size in this study was relatively small. A larger number of sampling is needed to confirm our results. Secondly, the calibration study was performed in Shanghai, China, a place with relatively higher annual RH and temperature. More tests are needed to understand how stable of our calibration model application in other places with similar PM2.5 pollution patterns and climate characteristics.
Light-scattering instruments such as QT-50 could overestimate the PM2.5 levels. The calibration model was set up after a long-term sampling covering a wide range of PM2.5 concentration, temperature and RH.
- 1. Pope CA, 3rd, Dockery DW. Health effects of fine particulate air pollution: lines that connect. J Air Waste Manag Assoc. 2006;56(6):709–42. Epub 2006/06/30. pmid:16805397.
- 2. Schwartz J. The effects of particulate air pollution on daily deaths: a multi-city case crossover analysis. Occup Environ Med. 2004;61(12):956–61. Epub 2004/11/20. pmid:15550600; PubMed Central PMCID: PMC1740692.
- 3. Beelen R, Raaschou-Nielsen O, Stafoggia M, Andersen ZJ, Weinmayr G, Hoffmann B, et al. Effects of long-term exposure to air pollution on natural-cause mortality: an analysis of 22 European cohorts within the multicentre ESCAPE project. Lancet. 2014;383(9919):785–95. Epub 2013/12/18. pmid:24332274.
- 4. Chen R, Li Y, Ma Y, Pan G, Zeng G, Xu X, et al. Coarse particles and mortality in three Chinese cities: the China Air Pollution and Health Effects Study (CAPES). Sci Total Environ. 2011;409(23):4934–8. Epub 2011/09/20. pmid:21925709.
- 5. Chen R, Kan H, Chen B, Wei H, Bai Z, Song G, et al. Association of Particulate Air Pollution With Daily Mortality The China Air Pollution and Health Effects Study. Am J Epidemiol. 2012;175(11):1173–81. pmid:22510278
- 6. Wong CM, Teams TPP. Public Health and Air Pollution in Asia (PAPA): A Multicity Study of Short-Term Effects of Air Pollution on Mortality. Environ Health Perspect. 2008;116(9):1195–202. pmid:18795163
- 7. Koehler KA, Peters TM. New Methods for Personal Exposure Monitoring for Airborne Particles. Current Environmental Health Reports. 2015;2(4):399–411. pmid:26385477
- 8. Yanosky JD, Williams PL, MacIntosh DL. A comparison of two direct-reading aerosol monitors with the federal reference method for PM2.5 in indoor air. Atmospheric Environment. 2002;36(1):107–13. http://dx.doi.org/10.1016/S1352-2310(01)00422-8.
- 9. Chan LY, Lau WL, Zou SC, Cao ZX, Lai SC. Exposure level of carbon monoxide and respirable suspended particulate in public transportation modes while commuting in urban area of Guangzhou, China. Atmospheric Environment. 2002;36(38):5831–40. http://dx.doi.org/10.1016/S1352-2310(02)00687-8.
- 10. Ramachandran G, Adgate JL, Pratt GC, Sexton K. Characterizing Indoor and Outdoor 15 Minute Average PM 2.5 Concentrations in Urban Neighborhoods. Aerosol Science and Technology. 2003;37(1):33–45.
- 11. McNamara ML, Noonan CW, Ward TJ. Correction factor for continuous monitoring of wood smoke fine particulate matter. Aerosol Air Qual Res. 2011;11(3):315–22. Epub 2011/06/01. pmid:25364330; PubMed Central PMCID: PMC4214921.
- 12. Wallace LA, Wheeler AJ, Kearney J, Van Ryswyk K, You H, Kulka RH, et al. Validation of continuous particle monitors for personal, indoor, and outdoor exposures. J Expo Sci Environ Epidemiol. 2011;21(1):49–64. Epub 2010/05/27. pmid:20502493.
- 13. Alexandrova OA, Boyer DL, Anderson JR, Fernando HJS. The influence of thermally driven circulation on PM10 concentration in the Salt Lake Valley. Atmospheric Environment. 2003;37(3):421–37. http://dx.doi.org/10.1016/S1352-2310(02)00803-8.
- 14. Liu D, Zhang Q, Jiang J, Chen DR. Performance Calibration of Low-cost and Portable Particular Matter (PM) Sensors. Journal of Aerosol Science. 2017.
- 15. Austin E, Novosselov I, Seto E, Yost MG. Laboratory Evaluation of the Shinyei PPD42NS Low-Cost Particulate Matter Sensor. Plos One. 2015;10(9):e0137789. pmid:26367264
- 16. Sloan CD, Philipp TJ, Bradshaw RK, Chronister S, Barber WB, Johnston JD. Applications of GPS-tracked personal and fixed-location PM(2.5) continuous exposure monitoring. J Air Waste Manag Assoc. 2016;66(1):53–65. Epub 2015/10/30. pmid:26512925.
- 17. Chartier R, Phillips M, Mosquin P, Elledge M, Bronstein K, Nandasena S, et al. A comparative study of human exposures to household air pollution from commonly used cookstoves in Sri Lanka. Indoor Air. 2016. Epub 2016/01/23. pmid:26797964.
- 18. Sloan CD, Philipp TJ, Bradshaw RK, Chronister S, Barber WB, Johnston JD. Applications of GPS-tracked personal and fixed-location PM2.5 continuous exposure monitoring. Journal of the Air & Waste Management Association. 2016;66(1):53–65.
- 19. Meng X, Fu Q, Ma Z, Chen L, Zou B, Zhang Y, et al. Estimating ground-level PM(10) in a Chinese city by combining satellite data, meteorological information and a land use regression model. Environ Pollut. 2016;208(Pt A):177–84. Epub 2015/10/27. pmid:26499934.
- 20. Zhu Y, Smith TJ, Davis ME, Levy JI, Herrick R, Jiang H. Comparing gravimetric and real-time sampling of PM(2.5) concentrations inside truck cabins. J Occup Environ Hyg. 2011;8(11):662–72. Epub 2011/10/14. pmid:21991940; PubMed Central PMCID: PMC3321380.
- 21. Challoner A, Gill L. Indoor/outdoor air pollution relationships in ten commercial buildings: PM2.5 and NO2. Building and Environment. 2014;80:159–73.
- 22. Wu CF, Delfino RJ, Floro JN, Samimi BS, Quintana PJ, Kleinman MT, et al. Evaluation and quality control of personal nephelometers in indoor, outdoor and personal environments. J Expo Anal Environ Epidemiol. 2005;15(1):99–110. Epub 2004/03/25. pmid:15039794.
- 23. Chang L-T, Suh HH, Wolfson JM, Misra K, Allen GA, Catalano PJ, et al. Laboratory and Field Evaluation of Measurement Methods for One-Hour Exposures to O3, PM25, and CO. J Air Waste Manag Assoc. 2001;51(10):1414–22. pmid:28086693
- 24. Tasić V, Jovašević-Stojanović M, Vardoulakis S, Milošević N, Kovačević R, Petrović J. Comparative assessment of a real-time particle monitor against the reference gravimetric method for PM10 and PM2.5 in indoor air. Atmospheric Environment. 2012;54:358–64.
- 25. Li M, Huang X, Zhu L, Li J, Song Y, Cai X, et al. Analysis of the transport pathways and potential sources of PM10 in Shanghai based on three methods. The Science of the total environment. 2012;414:525–34. Epub 2011/11/29. pmid:22119031.
- 26. Zhu Y, Smith TJ, Davis ME, Levy JI, Herrick R, Jiang H. Comparing Gravimetric and Real-Time Sampling of PM2.5 Concentrations Inside Truck Cabins. Journal of occupational and environmental hygiene. 2011;8(11):662–72. pmid:21991940
- 27. Challoner A, Gill L. Indoor/outdoor air pollution relationships in ten commercial buildings: PM 2.5 and NO 2. Building & Environment. 2014;80:159–73.
- 28. Holstius DM, Pillarisetti A, Smith KR, Seto E. Field calibrations of a low-cost aerosol sensor at a regulatory monitoring site in California. Atmospheric Measurement Techniques. 2014;7(4):1121–31.
- 29. Sousan S, Koehler K, Thomas G, Park JH, Hillman M, Halterman A, et al. Inter-comparison of low-cost sensors for measuring the mass concentration of occupational aerosols. Aerosol Science & Technology. 2016;50(5):00-.
- 30. Wang Y, Li J, Jing H, Zhang Q, Jiang J, Biswas P. Laboratory Evaluation and Calibration of Three Low-Cost Particle Sensors for Particulate Matter Measurement. Aerosol Science & Technology. 2015;49(11):1063–77.
- 31. Gao M, Cao J, Seto E. A distributed network of low-cost continuous reading sensors to measure spatiotemporal variations of PM2.5 in Xi'an, China. Environ Pollut. 2015;199:56–65. Epub 2015/01/27. pmid:25618367.