Citizen-Operated Low-Cost Sensors for Estimating Outdoor Particulate Matter Infiltration

doi:10.21203/rs.3.rs-4618450/v1

Download PDF

Article

Citizen-Operated Low-Cost Sensors for Estimating Outdoor Particulate Matter Infiltration

https://doi.org/10.21203/rs.3.rs-4618450/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Indoor air pollution poses a significant environmental concern, leading to adverse health effects. Fine particulates (PM_2.5) observed indoors exhibit high variability, influenced by both indoor emission sources and the infiltration of outdoor particles through open spaces and the incomplete building insulation. This study examines the relationship between indoor and outdoor PM_2.5 levels using data from a network of citizen-operated low-cost air quality sensors, deployed in Legionowo, Poland. Our results showed that generally, indoor PM_2.5 was lower than outdoor levels, with occasional peaks attributed to potential indoor emission sources. Statistical analysis identified emission events, particularly during cooking and household-heating periods, occurring more frequently from October to April. In the absence of indoor sources, outdoor particles accounted for 29–75% of indoor particle concentration, highlighting the significance of infiltration. This study shows how citizen-generated data using low-cost sensors, after post-processing, can provide decision-ready information as for example outdoor particulate matter infiltration factors for each building. This information can help decision-makers in devising effective interventions such as prioritizing indoor ventilation or addressing outdoor pollution sources.

Earth and environmental sciences/Climate sciences/Atmospheric science

Earth and environmental sciences/Environmental sciences/Environmental impact

Indoor Air Quality

Particulate Matter

Indoor-to-Outdoor Ratio

Infiltration Factor

Low-Cost Sensor

Household air pollution accounts for three million deaths worldwide according to the World Health Organization (WHO) ¹. Fine particulate matter (PM_2.5) — particulates with an aerodynamic diameter < 2.5 µm — is a major atmospheric pollutant attributed to both outdoor and indoor sources. Due to their small sizes, fine particles suspended to the air can easily penetrate into the respiratory system through inhalation, causing adverse health effects ². Therefore, assessing the indoor air pollutant concentrations and human exposure to indoor air pollution is crucial, considering that people typically spend from 65% to more than 90% of their time in indoor environments ^3,4.

Indoor particulate matter (PM) is regarded as a mixture of: particles generated by indoor sources and outdoor generated particles that have infiltrated into the indoor environment. Due to different origins, such particles exhibit diverse chemical compositions, toxicities, and sizes. When indoor sources are absent or not significant, particles with outdoor origin potentially affect the indoor air quality. Outdoor particles can enter a building through open doors and windows, but also from porous media and cracks of the building shell ^5–10. The presence of potential indoor emission sources coupled with the absent or inadequate ventilation could lead to higher indoor PM_2.5 concentrations compared to outdoor levels. Therefore, for a better understanding of indoor air quality, it’s essential to understand the ratio at which outdoor PM affects indoor conditions. Cooking, smoking, candles and wood burning appliances are mainly responsible for indoor fine PM, while coarser particles (PM_2.5 – PM₁₀) are typically the result of non-combustion activities like cleaning and particle resuspension due to residents’ movements ^11,12.

The initial step to investigate the connection between indoor and outdoor concentrations involves the calculation of Indoor-to-Outdoor concentration ratio (I/O). The I/O ratio varies dynamically throughout the day and across seasons ¹⁰. A ratio greater than unity shows the dominance of indoor emission sources. In general, since I/O depends on various factors and processes, it cannot solely offer reliable information about the relationship between indoor and outdoor air quality ⁸ unless additional information on ventilation and the frequency and duration of window openings is available ¹³. A detailed overview of I/O is provided in ⁸ and ¹⁴, showing a large range of values as explained by the different types of buildings, emission sources, monitoring periods, type of ventilation, etc. ⁸.

The impact of outdoor-generated particles on indoor air quality is mainly represented by the infiltration factor (f_inf) which quantifies the proportion of ambient PM entering indoor environments and remaining suspended under steady state conditions. The f_inf depends on building characteristics and operation, indoor and outdoor meteorological conditions, season and time of the day, and particle size ^{4,10,12,15–24}, while socioeconomic and demographic factors may also have indirect influences ^18,23. Various methodologies have been evaluated to estimate f_inf ¹³. The simplest approach when both information about indoor and outdoor air pollutants is available involves regressing indoor concentration on outdoor concentrations. The gain and the offset of the linear relationship act as the infiltration factor and the average concentration of the indoor-generated particles, respectively ^25–27. However, this approach estimates a “global“ f_inf and indoor-generated concentration neglecting the variability of both variables in time and their dependencies on auxiliary factors. The relationship between indoor and outdoor concentration levels was also statistically addressed using distributed-lag regression models ^22,23,28. These models use indoor concentrations as an independent parameter and outdoor PM levels in several lags combined with meteorological parameters or other auxiliary factors as independent information. The f_inf was derived by summing the regression coefficients of the lagged outdoor PM levels. Using this approach, Krebs et al. ²² estimated f_inf from 40–70% in the state of California, USA. A large extent of f_inf in single-family homes across USA was also estimated by Burke et al. ²³, where f_inf relationship with the household income and the outdoor concentration levels was also investigated.

Some other studies have calculated f_inf using the average I/O ratio over long-time intervals (e.g., day and week) excluding the extreme indoor emission sessions ^15,17,18,24. In those studies, it is assumed that the indoor concentrations closely mirror the corresponding outdoor concentrations when indoor emissions are minimal. Indoor emission-related PM concentrations typically exhibit abrupt, high-magnitude changes, and can be identified using rule-based approaches. Comparisons between indoor and outdoor concentrations, along with their time derivatives (consecutive concentration changes in time) help the detection of indoor emission events. Usually, higher values of f_inf are observed in summer compared to winter because of increased window opening frequency, with values varying depending on the building type.

This study explores the relationship between indoor and outdoor PM_2.5 concentrations in typical buildings of Legionowo, Poland. This is accomplished by integrating meteorological and PM_2.5 data from low-cost air quality sensors (LCSs) operated by citizens and installed in various locations across the city. In recent years, the use of LCSs for indoor air quality monitoring has risen. They not only provide information about air pollutant concentrations but also help identify pollution sources ^29–32. Despite challenges, particularly concerning data reliability compared to regulatory-grade instruments ³³, LCSs provide insights into both indoor and outdoor air quality levels ³⁴. This paper shows how data collected by citizens can be converted to decision-ready information. The ultimate goals are to explore the linkage between indoor and outdoor PM_2.5, to estimate the proportion of indoor PM_2.5 that originates from infiltration from the outdoor environment and to understand the factors affecting indoor PM_2.5 variability, such as indoor emissions and f_inf. The identified f_inf values for each building provide insights for policymakers. For instance, low f_inf indicates that dominant sources of pollution are indoor, suggesting that more interventions are required to improve indoor air quality, such as enhancing ventilation or regulating indoor heating sources. Conversely, high infiltration values suggest that the building has better insulation, but there may be a need to address outdoor pollution sources.

2.1. Study area

Legionowo is a city in central-eastern Poland (Longitude: 20.9369°E, Latitude: 52.4012°N), approximately 23 km north of Warsaw (Figure S1a). It covers a surface of 13.5 km² featuring a relatively flat terrain, predominantly surrounded by forests. The highest elevation in the area is around 80 km with no notable hills or mountainous ranges nearby. The population in Legionowo area is ~ 54170 inhabitants with a population density of 4000 people km^–2 (https://pl.wikipedia.org/wiki/Legionowo, assessed on 06.05.2022).

The climate is classified as temperate oceanic of type Cfb based on the Köppen-Geiger climate classification system. Figure S1b shows the climograph for Legionowo area. Meteorological data from 1951 to 2013 were obtained from the Polish Institute of Meteorology and Water Management—National Research Institute (IMGW-PIB) via https://dane.impw.pl/ using the “climate” R package ³⁵ (red star in Figure S1a). The average temperature is 8.3^°C, with January being the coldest month at − 2.3^°C and July the warmest at 19.2^°C. The yearly average precipitation is 537.2 mm per year, with the majority occurring during the summer period (341.1 mm). July stands out as the wettest month with 74.5 mm while the lowest precipitation amounts are recorded in February (27.9 mm) and March (27.1 mm).

In Legionowo, based on local experts, multi-family buildings are typically heated using central heating provided by a district heating system. For structures not connected to the central heating network, individual systems are used. Natural gas is the predominant choice for heating, followed by less common options such as hard coal, wood, electricity, and heat pumps. However, wood or coal is used for heating purposes ³⁶.

2.2. Low-cost sensor network

PM_2.5 and basic meteorological parameters are measured using low-cost sensors in various buildings situated in the Legionowo area. Figure 2 shows the spatial distribution of the air quality sensors. Indoor air quality was measured at nine locations while seven outdoor sensors were also installed in the same or nearby buildings. In six out of nine locations, both indoor and outdoor sensors were installed in the same building (yellow triangles in Fig. 1). For sensors 5485 and 5736 (red diamonds in Fig. 1), information about the outdoor air quality conditions was obtained through sensors installed in nearby buildings (green stars in Fig. 2). For 6030, outdoor air quality was determined through the nearest outdoor sensor (6033) which is positioned at a distance of 445 m.

Outdoor air quality was assessed using Airly sensors (https://airly.org/en). The Airly sensor system incorporates a Plantower PMS5003 sensor (https://www.plantower.com/en/products_33/74.html), which uses laser-based light scattering for measuring airborne particulate matter with aerodynamic diameters ranging from 0.3 to 10 µm. PM mass concentrations are provided in hourly resolution in the fractions of PM₁, PM_2.5 and PM₁₀. According to the manufacturer, the measurement range is 0–1000 µg m^–3 with an accuracy of ± 1 µg m^–3 for PM_2.5. In addition to PM, the Airly system also records air temperature (^°C), humidity (%), and air pressure (hPa).

For indoor air quality monitoring, AIRTHINGS Cloudberry systems (https://airthings.com) were deployed. The sensor system is equipped with an air quality sensor that utilizes laser light scattering technology to measure air particulates. PM₁ and PM_2.5 are recorded within the measurement range of 0 to 1000 µg m^–3. The claimed accuracies include ± 10 µg m^–3 for concentrations lower than 100 µg m^–3 and ± 10% for concentrations above 100 µg m^–3. The AIRTHINGS Cloudberry kit also includes sensors for measuring air temperature (^°C), relative humidity (%), air pressure (hPa), and volatile organic compounds (µg m^–3). It provides data at irregular sub-hourly time intervals. Indoor observations are hourly aggregated for being temporally consistent with outdoor data.

Indoor and outdoor air quality data were obtained through the AIRTHINGS dashboard and the Airly Data Platform from April 2022 to December 2023 based on the availability of indoor air quality measurements.

In general, the indoor concentration of an atmospheric pollutant, such as PM_2.5, can be modeled through a single mass-balance equation (Eq. 1) considering that indoor concentration changes mainly depend on a) the infiltration of outdoor particles, b) indoor particles that escape outdoors, c) deposition of indoor particles and d) indoor emissions. Eq. 1 assumes uniform mixing of the air pollutants and negates concentration changes due to gas-phase interactions or variations in the environmental parameters between the indoor and outdoor environments ¹³.

$$\frac{\text{d}{\text{C}}_{\text{i}\text{n}}\left(\text{t}\right)}{\text{d}\text{t}}=\text{p}{\alpha }{\text{C}}_{\text{o}\text{u}\text{t}}\left(\text{t}\right)-{\alpha }{\text{C}}_{\text{i}\text{n}}\left(\text{t}\right)-\text{k}{\text{C}}_{\text{i}\text{n}}\left(\text{t}\right)+\frac{\text{S}\left(\text{t}\right)}{\text{V}}$$

where dC_in is the change in indoor PM_2.5 concentration (µg m^–3) during the time interval dt, C_in and C_out are the indoor and outdoor particle’s concentrations (µg m^–3), t is the time (h), p is the penetration efficiency of particles (dimensionless), α is the air exchange rate (h^–1), k is the deposition rate of particles (h^–1), S is the indoor emission rate (µg h^–1) and V is the volume of the building or the room (m³). Under the assumption that the instantaneous temporal change of indoor concentration is substantially lower than the average indoor concentration over a considerable time interval, the steady-state form of Eq. 1 can be re-arranged as follows:

$${\text{C}}_{\text{i}\text{n}}=\frac{\text{p}{\alpha }}{{\alpha }+\text{k}}{\text{C}}_{\text{o}\text{u}\text{t}}+\frac{\text{S}}{\text{V}\left({\alpha }+\text{k}\right)}={\text{f}}_{\text{i}\text{n}\text{f}}{\text{C}}_{\text{o}\text{u}\text{t}}+{\text{C}}_{\text{i}\text{n}, \text{g}\text{e}\text{n}}$$

$$\frac{{\text{C}}_{\text{i}\text{n}}}{{\text{C}}_{\text{o}\text{u}\text{t}}}={\text{f}}_{\text{i}\text{n}\text{f}}+\frac{{\text{C}}_{\text{i}\text{n}, \text{g}\text{e}\text{n}}}{{\text{C}}_{\text{o}\text{u}\text{t}}}$$

with ${\text{f}}_{\text{i}\text{n}\text{f}}=\frac{\text{p}{\alpha }}{{\alpha }+\text{k}}$ the infiltration factor and ${\text{C}}_{\text{i}\text{n}, \text{g}\text{e}\text{n}}=\frac{\text{S}}{\text{V}\left({\alpha }+\text{k}\right)}$ the indoor concentration (µg m^–3) generated solely by indoor activities.

The indoor-to-outdoor concentration ratio converges to the f_inf when the outdoor concentration is significantly high or during periods with minimal indoor contribution (Eq. 3). The latter requires the detection and the removal of the indoor-generated peaks before the calculation of the infiltration factor. The optimal approach to detect such specific instances or/and periods is to record in detail the indoor activities. In this case, the peaks can also be assigned to the major activities that occurred in the indoor environment. If this kind of information is missing, machine learning and rule-based approaches can be evaluated for peak detection ^{15,17,18,24,28,37}. Here, the indoor emission cases were identified through analysis of the indoor PM_2.5 series using the Robust Extraction of Baseline Signal (REBS) methodology ^38,39. REBS has been applied for detecting local sources in various air pollutants time series ^39–41. According to Ruckstuhl et al. ³⁹, the indoor time series can be decomposed as:

$${\text{C}}_{\text{i}\text{n}}\left(\text{t}\right)={\text{C}}_{\text{B}}\left(\text{t}\right)+{\text{C}}_{\text{R}}\left(\text{t}\right)+{\epsilon }$$

where C_B(t) is the background concentration levels, C_R(t) is the concentration due to indoor emissions and other contributions (e.g., outdoor concentration) and ε is the normally distributed and independent errors. The local emissions that are responsible for spikes in time series then can be identified using the REBS method through a two-stage approach. Initially, the background concentration is determined using local linear regression over a moving window of a specific duration. Then, any data points greater than a designated threshold relative to the background concentrations are classified as emissions:

$${\text{C}}_{\text{i}\text{n}}\left(\text{t}\right)>\widehat{{\text{C}}_{\text{B}}}\left(\text{t}\right)+{\beta }\times {\sigma }$$

where $\widehat{{\text{C}}_{\text{B}}}\left(\text{t}\right)$ is the estimated background curve, σ is the standard deviation of the data falling below the background curve. β is a user-defined parameter. β controls the width of the threshold curve with higher values attributing to wider threshold concentrations. Here β is set equal to 3 as initially proposed in ³⁹. In this study, the REBS method was implemented using the rfbaseline function of the “IDPmisc” R package ⁴².

Once the peak emission events were detected, the f_inf at each timestamp was determined, assuming steady-state conditions after T hours (Section S1 of the supplementary material):

$${\text{f}}_{\text{i}\text{n}\text{f}}\left(\text{t}\right)=\frac{{⟨{\text{C}}_{\text{i}\text{n}}\left(\text{t}\right)⟩}_{\text{T}}}{{⟨{\text{C}}_{\text{o}\text{u}\text{t}}\left(\text{t}\right)⟩}_{\text{T}}}$$

with < C_in(t)>_T and < C_out(t)>_T the T-hour rolling averages for the indoor and the outdoor PM_2.5 concentrations, considering only the timestamps that were not identified as indoor emission events and at least 30% data coverage within the T-hour window. A steady state is reached when the instantaneous temporal change of indoor concentration is significantly lower than the average indoor concentration over a considerable time span (Eq. 1). Following the methodology outlined in section S1, the hourly concentration change is calculated by the concentration difference of the current and previous timestamp, ΔC_in(t) = C_in(t) – C_in(t–1). Then, running averages at various temporal intervals T are computed for the indoor concentration, <C_in(t)>_T, and the respective changes < ΔC_in(t)>_T. It is expected that < ΔC_in(t)> – <C_in(t)>_T almost equals to < C_in(t)>_T at increasing temporal intervals (Fig. S6 and S7). Section S1 and Fig. S6 and S7 showed steady state conditions are achieved after 48 hours. Based on this result, f_inf is derived through Eq. 6 using 48-hour running averages of indoor and outdoor PM_2.5.

4.1. Indoor and outdoor PM_2.5 concentrations

PM_2.5 was simultaneously measured in both indoor and outdoor environments using LCSs from February 2022 to August 2023. The data availability for each sensor is presented in Fig. S2. Regarding the outdoor PM_2.5 time series, most sensors provided continuous measurements with minimal data gaps (Fig. S2b). However, substantial data gaps were present in the indoor PM_2.5 measurements, primarily due to sensor malfunctions (Fig. S2a). Sensors 5189 and 5165 yielded almost complete time series for over five months, while the other sensors exhibited significant data gaps (Fig. S2a). Fig. S2c presents the periods when both indoor and outdoor PM_2.5 recordings were simultaneously available. Notably, sensors 5165 and 6030 were active for less than one month and provided the lowest data availability over the whole deployment period. Since this study focuses on the relationship between indoor and outdoor PM_2.5 levels, only the periods where both sensors were operational were included in the subsequent analysis.

Summary of the indoor and outdoor hourly PM_2.5 LCS measurements for the buildings is shown in Fig. 2a. Only timestamps with both indoor and outdoor measurements are used to generate the boxplots. Indoor PM_2.5 concentrations generally appear lower than the outdoor readings. The median ranges for indoor and outdoor PM_2.5 are 1.9–17.3 µg m^–3 and 6.7–27.9 µg m^–3, respectively (Table A1). Although significant peaks are present in indoor PM_2.5 time series (outlier points in Fig. 2a), mainly attributed to dominant indoor emission sources, 75% of the data were lower than 28 µg m^− 3.

Figure 2b and 2c illustrate the diurnal distribution of average indoor and outdoor PM_2.5 based on the available measurements of the entire period for each sensor. It is important to mention that the diurnal distribution for some sensors was drawn using a limited amount of data. For example, sensor 6030 measures indoor and outdoor PM_2.5 only for a two-month period implying that the diurnal pattern is not representative for the general temporal pattern at this specific location. Analyzing outdoor levels, all sensors exhibited a U-shaped pattern throughout the entire period, with lower concentrations between 09:00 and 18:00 local time (LT). However, analysis of the diurnal profiles on a seasonal basis revealed a different pattern in winter compared to other seasons, primarily displaying a bimodal distribution with morning and evening peaks (Fig. S3). This pattern is commonly observed in urban settings, where the morning peak is typically attributed to traffic, and the elevated concentrations during late afternoon and nighttime hours are due to heating activities and traffic ^34,43. The outdoor concentration remains consistently at high levels throughout the evening period. In winters, the low temperature conditions increase the heating activities while the low boundary layer height also traps the air pollutants close to the ground surface, resulting in relatively high PM_2.5 concentrations ⁴⁴. Conversely, the diurnal distribution of indoor PM_2.5 is primarily influenced by indoor activities. Most locations exhibit early morning and early afternoon and nighttime peaks. The sensor 5189 shows similar diurnal patterns for both indoor and outdoor PM_2.5. This behavior may be because indoor concentration is influenced by outdoor concentrations. It is worth noting that three sensors (5165, 5377, and 6030) were installed in kindergartens. Indoor PM_2.5 shows minimal variation throughout the day mainly because of the absence of significant indoor activities. In kindergartens, indoor emissions primarily stem from particle resuspension due to children's activities, as combustion activities are generally absent ^45,46. The type of ventilation used in indoor environments also affects the observed PM_2.5 levels. Sensor 6033 was installed in a kindergarten with a mechanical ventilation system, whereas other buildings rely on natural ventilation. In this particular location (sensor 6033), intra-day variations of the average indoor PM_2.5 are low ranging between 7.3 and 10.5 µg m^–3.

Figure 3a presents the Cumulative Distribution Functions (CDF) for the calculated hourly I/O ratios. On an hourly timescale, I/O exhibited significant variability both within a specific location and between different locations. In most places, 90% of the hourly I/O ratios were below unity implying that indoor PM_2.5 levels mainly reflect the contribution from outdoor particles. However, sensors 5351 and 5736 showed I/O > 1 in over 40% of hourly measurements, suggesting that in these two locations, the contribution and the strength of the indoor emission activities are greater compared to the other sensor locations. CDFs are also drawn based on the building type (kindergarten vs. other). Due to the absence of significant indoor sources in kindergartens, it is expected the diurnal distribution of I/O ratio to be lower than 1. Statistical analysis showed significant differences in the I/O distributions between the two building types (two-sample Kolmogorov-Smirnov test at 95% confidence level: KS statistic = 0.08 and p–value < < 0.05). Less than 10% of the I/O ratios is > 1 for kindergartens, while the for the other buildings, timestamps with I/O ratios > 1 are near 20% of the total calculated I/O ratios.

Figure 3b shows the diurnal distribution of the I/O ratio. The findings of Fig. 3b align with those discussed in X ¹⁰ concerning the diurnal distribution of I/O. Generally, I/O exhibits higher values from 09:00 to 18:00 LT, primarily due to the decrease in outdoor PM_2.5 during this period and increased indoor activities emitting PM_2.5 (Fig. 2c). Locations where the diurnal patterns of indoor and outdoor PM_2.5 closely resemble each other tend to display less variability in I/O throughout the day, as observed in sensors 5189 and 5485. A nearly stable I/O variation is presented in sensor 5189 which might be attributed to the impact of mechanical ventilation within the building and the relatively stable intra-day indoor PM_2.5 levels. Sensor 5736 showed I/O > 1 after 06:00 LT and the discrepancies between the diurnal indoor and outdoor PM_2.5 extended from − 0.1 µg m^–3 (I/O = 0.99) to 15 µg m^–3 (I/O = 2.9).

4.2. Detection of indoor emission events

The detection of elevated indoor PM_2.5 due to intense indoor activities was conducted using the REBS method with an illustrative example provided in Fig. 4 for sensor 5351. As illustrated in Fig. 4a, the REBS method efficiently captures the spikes present in the time series. Overall, 895 out of 8706 hourly data points (~ 10%) are attributed to significant indoor emission events. Figure 4b displays the statistical distributions of the indoor PM_2.5 levels for cases with PM_2.5 > threshold (“Emission”) and PM_2.5 < threshold (“Non-Emission”). The distributions are statistically different (two-sample t-test at 95% confidence level: p–value < < 0.05), with average (± 1 standard deviation) concentrations are 17.2 ± 11.1 µg m^–3 and 54.0 ± 27.7 µg m^–3 for the “Non-Emission” and “Emission” classes, respectively.

The methodology for identifying indoor emission events was applied to all indoor sensors with the results summarized in the boxplots of Fig. 5a. The average (± 1 standard deviation) PM_2.5 for the emission events ranged from 7.5 ± 2.1 µg m^–3 (sensor 5165) to 62.4 ± 68.1 µg m^–3 (sensor 5736). The ‘Emission’ and ‘Non-Emission’ classes were statistically different among the various sensors (based on the pairwise t-test under the 95% confidence level).

Cooking and household-heating activities potentially influence indoor PM_2.5 patterns. Consequently, an increase in the frequency of emission events is anticipated during the periods with high levels of indoor activities. As already mentioned in the data session, the use of coal or wood for residential heating purposes is very common in Poland. In general, Polish households have the highest coal consumption rate across the European Union ⁴⁷ while strict measures of solid-fuel burning for heating purposes have not widely implemented. As pointed out in ⁴⁸, fine PM concentrations shows a strong positive correlation with the number of heating systems in private-homes in Poland .

Figure 5c illustrates the diurnal distribution of detected emission events, expressed as a percentage of the total number of indoor LCSs measurements in each hour of day, from October to April (“household-heating” period) and from May to September (“non-household-heating” period). The percentages were calculated using detected emission events from all sensors. During the “non-household-heating” period, the percentage of emission events fluctuates between 4% and 10% of total measured events, with higher values observed in the early morning hours and after 16:00 LT. Assuming that indoor activities like cooking remain consistent during a day/year, the difference between the diurnal curves for the two periods could provide insights about household-heating activity contribution to the number of detected indoor emission events. In the “heating-household” months, people stay more at home, engage in more indoor activities, and the demand for heating increases. During “household-heating” period, nearly 25–27% of the indoor PM_2.5 measurements were identified to have a dominant indoor source of emission after 18:00 LT, indicating a 15% increase compared to the “non-household-heating” period. Similar analyses were conducted at each sensor location separately (Fig. S6). Only three out of nine sensors have available data for both “household-heating” and “non-household-heating” periods.

4.3. Infiltration factor, f_inf

The box plots in Fig. 6a and the summary statistics in Table A1 present the results of the f_inf calculation for various locations in Legionowo. The estimated f_inf values ranged from 0 and 1, with average values extending between 0.29 (sensors 5165 and 5189) and 0.75 (sensor 5751). This suggests 29%-75% of indoor measured concentrations are of outdoor origin. The standard deviation of f_inf is nearly similar across all locations (0.13–0.16). f_inf values are consistent with those reported in the literature ⁸. Seasonal analysis of f_inf is presented in Fig. 6b. Overall, during the “household-heating” period, f_inf was 0.45 ± 0.20, and 0.53 ± 0.22 during the “non-household-heating” period. Significant seasonal differences in the average f_inf were also observed for individual locations (based on t-tests at the 95% confidence level), ranging from 0.03 (for sensor 6033) to 0.19 (for sensor 5485). During the “non-household-heating” period, windows are opened more frequently, allowing the infiltration of outdoor particles into the indoor environment.

The f_inf depends on the penetration efficiency, the deposition rate of the indoor air pollutants, and the air-exchange rate (Eq. 5). In addition to building characteristics and residents’ activities, meteorological conditions may also influence f_inf ^8,24,49. In order to assess the relation between meteorology and f_inf, the differences between outdoor-indoor temperatures during the “household-heating” and “non-household-heating” periods are plotted against f_inf in Fig. S8a an Fig. S8b, respectively. Similar analysis was conducted in Lunderberg et al. ²⁴. Based on these results, the contribution of outdoor/indoor temperature difference to the f_inf is rather marginal, suggesting that this parameter cannot be considered as significant proxy for describing the f_inf variations.

This study investigated the link between indoor and outdoor PM_2.5 levels using a network of citizen-operated LCSs in Legionowo, Poland. Sensors were deployed both indoors and outdoors at nine locations throughout the city, providing data from April 2022 to December 2023. One limitation of this study is the absence of information regarding the daily residents’ activities contributing to indoor PM_2.5. Nonetheless, simultaneous measurements of indoor and outdoor PM_2.5 helped to detect indoor emission events and estimating proportion of the outdoor particles infiltrating into the buildings. The key findings are outlined as follows:

Indoor PM_2.5 levels were significantly lower than the outdoor levels, with median concentration ranging from 1.9 to 17.3 µg m^–3 indoors and 6.7 to 27.9 µg m^–3 outdoors. Occasional spikes in indoor PM_2.5 were attributed to potential indoor emission sources.
The diurnal distribution of outdoor PM_2.5 concentrations exhibited a U-shape pattern throughout the measurement period peaking in the morning and late afternoon to evening hours. Indoor PM_2.5 varied during the day driven by the indoor activities, with evening peaks likely associated with dinner cooking and household heating activities. Sensors installed in kindergartens (5165, 5377 and 6030) showed minimal diurnal PM_2.5 variation, mainly reflecting the influence of children’s activities on indoor particle levels.
The I/O concentration ratio was analyzed to understand the relationship between indoor and outdoor PM_2.5. I/O varied significantly both within a specific location and between different locations, mostly falling below unity. Sensors 5351 and 5736 exhibited significant indoor emissions, with over 40% of hourly measurements showing I/O > 1. Higher I/O ratios were observed from 09:00 to 18:00 LT on a diurnal basis, with less variability in intra-day I/O ratios observed in locations with similar indoor and outdoor PM_2.5 patterns.
A statistical approach was used to detect the potential indoor emissions events. The number of emission events was higher during early afternoon and nighttime hours, where activities like dinner cooking and house heating were more prevalent. The emission events were more distinguishable from October to April due to the contribution of household heating activities.
The influence of outdoor PM2.5 on indoor air pollution was also quantified in the f_inf, showing that 29–75% of indoor measured concentrations originated from outdoor sources. The f_inf was higher during warmer months (May – September) with little to no household-heating activities, attributable to more-frequent door and window openings. The contribution of outdoor-indoor temperature difference to the f_inf was found minimal.

Data availability: Data supporting the findings of this study will be made available upon reasonable request.

Funding: This research leading to these results has been received from the EEA Grants 2014–2021 via the National Center for Research and Development. The grant is provided under “Applied Research” Program for the “GREEN HEAT – towards collaborative local decarbonization” project (acronym GREEN HEAT); grant number PL-Applied Research-0043. Also, this research has received partial funding from the European Union’s Horizon 2020 Research and Innovation programme under grant agreement No 952433, VIDIS project. We also acknowledge funding for SOCIO-BEE project from the European Union’s Horizon Europe research and innovation programme under grant agreement No. 101037648.

Acknowledgments: The contribution of all citizens who participated in the GREEN HEAT project is acknowledged.

Conflicts of Interest: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

WHO. Household air pollution. https://www.who.int/news-room/fact-sheets/detail/household-air-pollution-and-health (2022).
Pope, C. A., Coleman, N., Pond, Z. A. & Burnett, R. T. Fine particulate air pollution and human mortality: 25 + years of cohort studies. Environmental Research 183, 108924 (2020).
Schweizer, C. et al. Indoor time–microenvironment–activity patterns in seven regions of Europe. J Expo Sci Environ Epidemiol 17, 170–181 (2007).
Liu, Y., Ma, H., Zhang, N. & Li, Q. A systematic literature review on indoor PM2.5 concentrations and personal exposure in urban residential buildings. Heliyon 8, e10174 (2022).
Liu, D.-L. & Nazaroff, W. W. Modeling pollutant penetration across building envelopes. Atmospheric Environment 35, 4451–4462 (2001).
Liu, D.-L. & Nazaroff, W. W. Particle Penetration Through Building Cracks. Aerosol Science and Technology 37, 565–573 (2003).
Bennett, D. H. & Koutrakis, P. Determining the infiltration of outdoor particles in the indoor environment using a dynamic model. Journal of Aerosol Science 37, 766–785 (2006).
Chen, C. & Zhao, B. Review of relationship between indoor and outdoor particles: I/O ratio, infiltration factor and penetration factor. Atmospheric Environment 45, 275–288 (2011).
Liu, C. et al. Influence of natural ventilation rate on indoor PM2.5 deposition. Building and Environment 144, 357–364 (2018).
Stamp, S. et al. A critical evaluation of the dynamic nature of indoor-outdoor air quality ratios. Atmospheric Environment 273, 118955 (2022).
Isaxon, C. et al. Contribution of indoor-generated particles to residential exposure. Atmospheric Environment 106, 458–466 (2015).
Li, J., Fan, G., Ou, Y. & Deng, Q. Characteristics and control strategies of indoor particles: An updated review. Energy and Buildings 294, 113232 (2023).
Diapouli, E., Chaloulakou, A. & Koutrakis, P. Estimating the concentration of indoor particles of outdoor origin: A review. Journal of the Air & Waste Management Association 63, 1113–1129 (2013).
Ilacqua, V., Scharko, N., Zambrana, J. & Malashock, D. Survey of residential indoor particulate matter measurements 1990–2019. Indoor Air 32, (2022).
Allen, R., Larson, T., Sheppard, L., Wallace, L. & Liu, L.-J. S. Use of Real-Time Light Scattering Data To Estimate the Contribution of Infiltrated and Indoor-Generated Particles to Indoor Air. Environ. Sci. Technol. 37, 3484–3492 (2003).
Rim, D., Wallace, L. & Persily, A. Infiltration of Outdoor Ultrafine Particles into a Test House. Environ. Sci. Technol. 44, 5908–5913 (2010).
Kearney, J. et al. Residential indoor and outdoor ultrafine particles in Windsor, Ontario. Atmospheric Environment 45, 7583–7593 (2011).
MacNeill, M. et al. Factors influencing variability in the infiltration of PM2.5 mass and its components. Atmospheric Environment 61, 518–532 (2012).
Tang, C. H. et al. Development of a modeling approach to estimate indoor-to-outdoor sulfur ratios and predict indoor PM2.5 and black carbon concentrations for Eastern Massachusetts households. J Expo Sci Environ Epidemiol 28, 125–130 (2018).
Vu, T. V. et al. Assessing the contributions of outdoor and indoor sources to air quality in London homes of the SCAMP cohort. Building and Environment 222, 109359 (2022).
Liang, Y. et al. Wildfire smoke impacts on indoor air quality assessed using crowdsourced data in California. Proc. Natl. Acad. Sci. U.S.A. 118, e2106478118 (2021).
Krebs, B., Burney, J., Zivin, J. G. & Neidell, M. Using Crowd-Sourced Data to Assess the Temporal and Spatial Relationship between Indoor and Outdoor Particulate Matter. Environ. Sci. Technol. 55, 6107–6115 (2021).
Burke, M. et al. Exposures and behavioural responses to wildfire smoke. Nat Hum Behav 6, 1351–1361 (2022).
Lunderberg, D. M. et al. Assessing residential PM _2.5 concentrations and infiltration factors with high spatiotemporal resolution using crowdsourced sensors. Proc. Natl. Acad. Sci. U.S.A. 120, e2308832120 (2023).
Ott, W., Wallace, L. & Mage, D. Predicting Particulate (PM ₁₀) Personal Exposure Distributions Using a Random Component Superposition Statistical Model. Journal of the Air & Waste Management Association 50, 1390–1406 (2000).
Wallace, L. & Ott, W. Long-Term Indoor-Outdoor PM2.5 Measurements Using PurpleAir Sensors: An Improved Method of Calculating Indoor-Generated and Outdoor-Infiltrated Contributions to Potential Indoor Exposure. Sensors 23, 1160 (2023).
Mousavi, A. & Wu, J. Indoor-Generated PM _2.5 During COVID-19 Shutdowns Across California: Application of the PurpleAir Indoor–Outdoor Low-Cost Sensor Network. Environ. Sci. Technol. 55, 5648–5656 (2021).
Nguyen, P. D. M. et al. Using Low-Cost Sensors to Assess Fine Particulate Matter Infiltration (PM2.5) during a Wildfire Smoke Episode at a Large Inpatient Healthcare Facility. IJERPH 18, 9811 (2021).
Morawska, L. et al. Applications of low-cost sensing technologies for air quality monitoring and exposure assessment: How far have they gone? Environment International 116, 286–299 (2018).
Ródenas García, M. et al. Review of low-cost sensors for indoor air quality: Features and applications. Applied Spectroscopy Reviews 57, 747–779 (2022).
Bousiotis, D., Alconcel, L.-N. S., Beddows, D. C. S., Harrison, R. M. & Pope, F. D. Monitoring and apportioning sources of indoor air quality using low-cost particulate matter sensors. Environment International 174, 107907 (2023).
Higgins, C., Kumar, P. & Morawska, L. Indoor air quality monitoring and source apportionment using low-cost sensors. Environ. Res. Commun. 6, 012001 (2024).
Castell, N. et al. Can commercial low-cost sensor platforms contribute to air quality monitoring and exposure estimates? Environment International 99, 293–302 (2017).
Hassani, A. et al. Low-cost sensors and Machine Learning aid in identifying environmental factors affecting particulate matter emitted by household heating. Atmospheric Environment 314, 120108 (2023).
Czernecki, B., Głogowski, A. & Nowosad, J. Climate: An R Package to Access Free In-Situ Meteorological and Hydrological Datasets For Environmental Assessment. Sustainability 12, 394 (2020).
Karpinska, L., Śmiech, S., Gouveia, J. P. & Palma, P. Mapping Regional Vulnerability to Energy Poverty in Poland. Sustainability 13, 10694 (2021).
Chan, W. R. et al. Quantifying fine particle emission events from time-resolved measurements: Method description and application to 18 California low‐income apartments. Indoor Air 28, 89–101 (2018).
Ruckstuhl, A. F., Jacobson, M. P., Field, R. W. & Dodd, J. A. Baseline subtraction using robust local regression estimation. Journal of Quantitative Spectroscopy and Radiative Transfer 68, 179–193 (2001).
Ruckstuhl, A. F. et al. Robust extraction of baseline signal of atmospheric trace species using local regression. Atmos. Meas. Tech. 5, 2613–2624 (2012).
et al. Identification of spikes associated with local sources in continuous time series of atmospheric CO, CO₂ and CH₄ Atmos. Meas. Tech. 11, 1599–1614 (2018).
Cristofanelli, P. et al. Identification of spikes in continuous ground-based in situ time series of CO ₂, CH ₄ and CO: an extended experiment within the European ICOS Atmosphere network. Atmos. Meas. Tech. 16, 5977–5994 (2023).
Locher, R. IDPmisc: ‘Utilities of Institute of Data Analyses and Process Design (. (2024).
Hassani, A., Schneider, P., Vogt, M. & Castell, N. Low-Cost Particulate Matter Sensors for Monitoring Residential Wood Burning. Environ. Sci. Technol. 57, 15162–15172 (2023).
Barlow, J. F. Progress in observing and modelling the urban boundary layer. Urban Climate 10, 216–240 (2014).
Morawska, L. et al. Airborne particles in indoor environment of homes, schools, offices and aged care facilities: The main routes of exposure. Environment International 108, 75–83 (2017).
Kalimeri, K. K., Bartzis, J. G., Sakellaris, I. A. & De Oliveira Fernandes, E. Investigation of the PM2.5, NO2 and O3 I/O ratios for office and school microenvironments. Environmental Research 179, 108791 (2019).
Sokołowski, J. & Bouzarovski, S. Decarbonisation of the Polish residential sector between the 1990s and 2021: A case study of policy failures. Energy Policy 163, 112848 (2022).
Zgłobicki, W. & Baran-Zgłobicka, B. Air pollution in major Polish cities in the period 2005–2021: Intensity, effects and attempts to reduce it. Environmental Research 240, 117497 (2024).
Liang, D. et al. Estimating climate change-related impacts on outdoor air pollution infiltration. Environmental Research 196, 110923 (2021).

(Not answered)

AppendixA.docx
SupplementaryMaterialSalamalikisetal.docx
Supplementary Material: Citizen-Operated Low-Cost Sensors for Estimating Outdoor Particulate Matter Infiltration

Download PDF

Editorial decision: revise
29 Jul, 2024
Review #2 received at journal
26 Jul, 2024
Review #3 received at journal
23 Jul, 2024
Reviewer #3 agreed at journal
05 Jul, 2024
Reviewer #2 agreed at journal
04 Jul, 2024
Reviewer #1 agreed at journal
04 Jul, 2024
Reviewers invited by journal
02 Jul, 2024
Editor assigned by journal
01 Jul, 2024
Submission checks completed at journal
01 Jul, 2024
First submitted to journal
27 Jun, 2024
Unknown event
25 Jun, 2024

You are reading this latest preprint version

Citizen-Operated Low-Cost Sensors for Estimating Outdoor Particulate Matter Infiltration

Status:

Version 1

Abstract

Figures

1. Introduction

2. Data

2.1. Study area

2.2. Low-cost sensor network

3. Methodology

4. Results

4.1. Indoor and outdoor PM_2.5 concentrations

4.2. Detection of indoor emission events

4.3. Infiltration factor, f_inf

5. Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1

Citizen-Operated Low-Cost Sensors for Estimating Outdoor Particulate Matter Infiltration

Status:

Version 1

Abstract

Figures

1. Introduction

2. Data

2.1. Study area

2.2. Low-cost sensor network

3. Methodology

4. Results

4.1. Indoor and outdoor PM2.5 concentrations

4.2. Detection of indoor emission events

4.3. Infiltration factor, finf

5. Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1

4.1. Indoor and outdoor PM_2.5 concentrations

4.3. Infiltration factor, f_inf