Low-cost air pollution monitoring system—an opportunity for reducing the health risk associated with physical activity in polluted air

General information

As part of the scientific project, known as the Storm&DustNet implemented at the Jagiellonian University in Kraków (Poland), the Jagiellonian University Network of Stations (JUNS) was created, dedicated to monitoring the concentration of PM in the atmosphere, as well as selected meteorological parameters (including air temperature, humidity, and atmospheric pressure). The first stations began working under JUNS at the beginning of 2016.

The central element of JUNS is a data server with software written in the script programming language Python. This server receives data from measuring stations and allows the administrator full access to data, as well as remote diagnosis of the technical condition of individual university measuring stations (UMS), which usually work far in the field. The data server processes the data flowing from individual stations on an ongoing basis, prepares the necessary information to be displayed on the website, and places it on the web-server. As a result, every Internet user can observe the results of the online measurements in the public domain via a website (http://inhalation.uj.edu.pl).

The general schematic diagram of JUNS is presented in Fig. 1A. The network is built in the star topology, and each measuring station connects to the data server via the Internet using one of two techniques: (a) wireless transmission using module GSM/UMTS/LTE; or (b) standard cable connection TCP/IP. The JUNS network architecture ensures easy scalability of the solution in both the hardware and software layers.

Characteristic of servers

Both servers included in JUNS are embedded in the technical Internet infrastructure of the Faculty of Physics, Astronomy, and Applied Computer Science of the Jagiellonian University, and work on the basis of the Linux system (version SUSE 4.4.162). Communication of the measuring stations with the data server is carried out by cryptographic network protocol (SSH), and the results of the measurements are stored in the MySQL relational database. All access to the measurement results is therefore made through the query language of the SQL Server Database.

Characteristics of measuring stations

The university measuring stations (UMS) have a uniform structure and software written in Python. The components of each of them are: (a) minicomputer BeagleBone Black (Arrow Electronics, USA), which is a low cost and popular development board for developers and the avid hobbyist; (b) wireless communication module GSM E3131h-2 (Huawei, China); (c) digital laser dust sensor SEN0177 (DFRobot, China) which measures the particle concentration of suspended particulate matter in the air; (d) integrated humidity, temperature, and pressure sensor BME280 (Bosch, Germany); (e) precision real-time clock module DS3231 (Maxim Integrated, USA); (f) module GPS MIKROE-1032 (MikroElektronika, Serbia); (f) analog-digital converter ADS1015 (Texas Instruments, USA), which is used to monitor the input supply voltage of the station and 3.3V in the internal power bus; (g) step-down voltage converter D24V50F5 (Pololu, USA), which enables DC power supply in from 6.0 to 38.0 V by a standard power supply or accumulator. The average power consumption of the station is equal to 5.4 W (450mA@12V). The quality of measurements made with the DFRobot sensor has been confirmed in former studies (Rogulski, 2017; Markowicz & Chiliński, 2020).The block diagram of the station is presented in Fig. 1B, and the view of the arrangement of the components in the housing of the measuring station is shown in Fig. 1C. According to the technical specifications of the manufacturers, all components of the measuring stations can be operated in the humidity range of 5–95% and the temperature from −20 °C to 50 °C. The external dimensions of the station are 20 × 15 × 7.3 cm (LxWxH), and the weight is 0.6 kg. The station can work as a stationary or mobile device. The total cost of building the station was approximately 400 USD, estimated at the beginning of 2020.

Data quality

The station measured the concentration of particle matter and air temperature, humidity, and pressure several dozen times per minute. The parameters were averaged over one minute intervals. Then, the average values were sent to the data server and saved to a database. Measurements of the air temperature, relative humidity (RH) and pressure were taken using a BME280 detector installed inside the UMS (temperature range: −40 to +85 °C with accuracy: ±1 °C; humidity rage: 10 to 80% RH with accuracy: ±3% RH; pressure range: 300 to 1100 hPa with accuracy: ±1 hPa).

Laboratory verification of PM₁₀ data quality

Although UMS measures three indices (PM₁₀, PM_2.5 and PM₁), we verified the most popular PM₁₀ mass index for which was determined in Poland a daily permissible level of 50 µg m⁻³ (Chief Inspectorate For Environmental Protection, 2015) . The UMS’s final accuracy of the concentration measurements of suspended particulate matter PM₁₀ was determined on the basis of test measurements carried out for this purpose. Results recorded by UMS have been compared with data obtained from the reference analyzer EDM107 produced by GRIMM (GRIMM Aerosol Technik, Germany). Both apparatus were operated in the same place during the test. The EDM107 was our calibration standard of known accuracy, with the measurement error amounting to ±2 µg m⁻³. This analyzer has a certificate of calibration and equivalency to a gravimetric method (Grimm & Eatough, 2009) .

Quality verification of data was carried out simultaneously for 15 UMS stations, which were placed side by side with the analyzer EDM107 in the laboratory next to the window (about 1 m away). The distance between the air inlet and the analyzer, and between the air inlet and each UMS station, did not exceed 50 cm. One additional UMS station (named St4) worked outside the building near the lab window.

The results of the comparative test measurements is presented in Fig. 2A, where the thin black lines represent data recorded by 15 UMSs, the bold black line is the average from measurements made by 15 UMSs and standard deviation; the bold red line represents data recorded by the analyzer EDM107 and its error; and the blue line represents data recorded by the additional station St4.

The comparative measurements lasted 95 min (07:10-08:45) and consisted of two periods. The first period lasted 25 min (07:10-07:35) and the window was closed in the laboratory which had clean air. All UMSs registered low concentrations of PM₁₀, andthe mean values were 14.0 µg m⁻³ (SD = ±1.5 µg m⁻³). The EDM107 analyzer recorded a mean of 14.2 µg m⁻³ (SD = ±0.5 µg m⁻³).

During the second period, the window was completely open (07:36-08:45). After opening the window, the polluted air from outside flowed into the laboratory. The window opening time and the rapid influx of polluted air are marked with the yellow rectangle in Fig. 2A. During the comparative measurement, the PM₁₀ values recorded individually by 15 UMSs did not differ more than ±10 µg m⁻³ from the results recorded by the EDM107 analyzer. Time distributions of the PM indicator registered by each UMS station correlated with the results recorded by EDM107 (Spearman coefficient in range 0.86–0.94).

In summary, the comparative measurements described above included natural ambient air and various particulate matter concentrations. As a result, the comparison showed that measurements of PM₁₀ made by UMS stations are encumbered with an average error that does not exceed ±10 µg m⁻³.

Practical verification of PM₁₀ data quality

The other method of verification was the comparison of measurements made by the validated monitoring station belonging to the Provincial Environmental Protection Inspectorate (PEPI) in Kraków, and the results made by one of the UMS stations (St10), which operated together in this same place and time. Comparative measurements were taken in the time period 19-31.12.2017 in Skawina, in close proximity to the city of Kraków. The distribution of hourly averages of particulate matter PM₁₀ measured by both stations is presented in Fig. 2B, and the correlation coefficient was very high (R = 0.96). The average value of PM₁₀ calculated based on the measurements taken by the St10 station was 15% higher than the average calculated on the basis of the measurements taken by the validated station PEPI. As can be seen in Fig. 2B, the results of measurements from the St10 station are higher than those of the PEPI station mainly for PM₁₀ greater than 100 µg m⁻³. For PM₁₀ below 100 µg m⁻³, the average error for the St10 station does not exceed 9 µg m⁻³.

Summary of the PM₁₀ data quality verification

The comparative analysis of the PM₁₀ data quality recorded by UMS stations using two approaches was completed. In the data range below 100 µg m⁻³, both the laboratory and the practical analysis gave the comparative inaccuracy equal ±10 µg m⁻³ and ±9 µg m⁻³. Considering the wider range of data (above 100 µg m⁻³), the UMS stations made measurements with an accuracy of 15%.

The spatial distribution of the JUNS

The JUNS made their first multipoint observations over the Kraków Metropolitan Area during 2017 in 8 localizations, as presented in Fig. 3A. Locations were selected that represent a diverse spectrum of terrain conditions in which the Kraków agglomeration community undertakes physical activity. Therefore, the types of buildings, traffic intensity, and the presence of vegetation were taken into account. The UMS stations were installed mainly on buildings, at a height around 2.5 m above the ground. Four stations were installed in villages close to Kraków, where the building density is low: Czajowice (St0), Brzoskwinia (St7), Zielonki (St8), and Siepraw (St9). One station monitored air quality in the town Wieliczka (St1) next to Kraków, and three stations monitored air quality in the city of Kraków: al. Mickiewicza (St2), one of the main streets of the city; ul. Zapolskiej (St3), a housing estate of compact houses; ul. Łojasiewicza (St4), the Jagiellonian University Campus, which is close to green areas.

It should be mentioned that some of our measurement stations belonging to the JUNS were placed in the regions of the Krakow Metropolitan Area, where people used to practice varied forms of outdoor physical activities. For example, station (St2) in al. Mickiewicza is located close to the Planty Krakowskie park area and to the eastern part of the grassland zone called Błonia Krakowskie. Similarly, the station (St1) in Wieliczka is located very close to the athletics stadium, where sportsmen practice varied sports and participate in competitions. As can be seen in Fig. 3A, popular locations chosen by the people to practice varied outdoor physical activities - such as Planty Krakowskie and the eastern part of the zone Błonia Krakowskie, located close to station (St2) in al. Mickiewicza (Fig. 3A), as well as the athletics stadium in Wieliczka (St1), are unfortunately characterized by poorer air quality than the other locations studied within our project (see e.g., (St0) Czajowice - a village situated just on the border of the town).

Choosing a place for physical activity during the winter (or cold season) through the lens of air quality

It should be noted that physical activity is undertaken in different seasons in the year, both in warmer and colder months. Unfortunately, in winter, as a result of low temperatures and increased combustion, as well as unfavorable meteorological situations (no wind) (Scibor, Bokwa & Balcerzak, 2020), there are episodes of high concentrations of particulate matter which pose a real threat to human health.

In 2017, particularly long periods of high air pollution were recorded in January. In all eight monitored locations, PM₁₀ exceeded the permissible level (50 µg m⁻³) on most days of January. The largest number of days exceeding the permissible level occurred in Wieliczka (29), and the smallest in Czajowice (17). The average monthly value of PM₁₀ for January 2017 was the highest in Kraków-al. Mickiewicza (289 µg m⁻³), and the lowest in Czajowice (86 µg m⁻³). The highest daily PM₁₀ concentration occurred on 30.01.2017, when the stations recorded maximal PM₁₀ values in Kraków-al. Mickiewicza (696 µg m⁻³), as well as high levels in Wieliczka (605 µg m⁻³). The monthly average values of PM₁₀ recorded by stations in rural areas were in the range of 86–203 µg m⁻³, and the distribution of daily values in January 2017 is presented in Fig. 3 (panels B, C, D, E). The monthly average values of PM₁₀ recorded by stations in urban areas were in the range of 126-289 µg m⁻³, and the distribution of daily values in January 2017 is presented in Fig. 3 (panels F, G, H, I). The PM_2.5/PM₁₀ ratio observed in our study was 0.82, which is very close to the values previously reported by others (Wilczyńska-Michalik & Michalik, 2017).

Example of choosing the optimal place and time for open-air physical activity in the future

The concentration of PM₁₀ depends on several factors, such as location, season, and time of day. All of them have their basis in landform, weather conditions, economic activities, and daily human life. Current decisions about the choice of place and time of physical activity in the open air can be made on the basis of available air quality measurements. However, in long-term planning, e.g., for future major sporting events, it is good to make a decision after reviewing the archival results of air quality measurements in the considered locations. Figure 4 presents examples of results recorded in 2017 by four select stations, for which an example analysis was made. The distribution of PM₁₀ concentration as a function of time of day and season is color-coded. Vertical areas of red shades are clearly visible, and correspond to periods of several days in which the stations recorded very high concentrations of PM₁₀ occurring 24-hours per day and day by day. During this time, the average hourly PM₁₀ values were over 80 µg m⁻³ (colors: orange, red and burgundy) exceeding around twice (or more) of the daily limit value (50 µg m⁻³).

In 2017, the minimum and maximum value of the average PM₁₀ concentration in the examined hour occurred between 10:00–14:00 and 18:00-22:00, regardless of location (Table 1), the minimum value was in village Czajowice (20.6 µg m⁻³), and the maximum in Wieliczka (120.1 µg m⁻³). At all locations, the largest percentage of time with PM₁₀ with a Good Air Quality Level (AQL) occurred between 10:00–14:00 (maximum 70% in Czajowice), and the lowest between 18:00–22:00 (minimum 11% in Wieliczka). The biggest percent of time with Extremely Poor AQL were between 18:00–22:00, independent of location (maximum 25% in Wieliczka).

Minimizing the health risks from exposure to air pollution

Minimizing the health risks from exposure to air pollution on an annual scale can be determined using the results in Table 2. For most months of 2017, minimum monthly average 4-hour PM₁₀ concentrations were recorded between 10:00–14:00 regardless of location. In general, the minimum average monthly value of PM₁₀ was recorded between 14:00–18:00 in Czajowice (4.9 µg m⁻³), and the maximum between 18:00–22:00 in Wieliczka (339.0 µg m⁻³).

It should be emphasized that the results presented in Fig. 4 come from locations of different natures and degrees of urbanization. Czajowice and Siepraw are definitely rural areas, Wieliczka is a small town, and the area of the III Campus of the Jagiellonian University represents the edge of the recreation area in the southwest of Kraków. The biggest area of green color in Fig. 4A indicates good and very good Air Quality Level, which was in Czajowice.

The results showed that, among the four locations selected, the village Czajowice is most often the best choice for physical activity in the open air. In contrast, as seen in Fig. 4D, the Wieliczka area is characterized by frequent periods with elevated PM₁₀ concentrations, indicating that this is not a suitable place for activity in the open air.

Despite the important general differences in locations of the UMS stations, short periods of high PM₁₀ concentrations are clearly visible at all locations (Fig. 4), although with varying intensity. This indicates the large spatial coverage of this phenomenon covering at least the area of the Kraków agglomeration. In this case, the recommendation should be that the all unnecessary outdoor activities should be abandoned.

Possible applications of the system

In recent years, one has witnessed growing popularity of low-cost air pollution monitoring systems and their applications in several fields. Simple low-cost systems can track short-lived pollution events (Bulot et al., 2019), monitor pollution in smart cities (Chen et al., 2017; Johnston et al., 2019), provide aggregated information about observed air quality outdoors (Castell et al., 2017; Rogulski, 2018), in flats and houses and in the zoological garden area (Pawlak & Nieckarz, 2020). Generally, it is recommended that a credible validation of low-cost sensors in a future study is performed (Chojer et al., 2020).

As shown above, simple systems, such as the unit used in the present study, can be used to monitor and register the quality of air, which could firstly be applied to the industrial areas of densely populated parts of a country. It could be also suitable in villages, especially those offering an attractive touristic service. Finally, this system could be valuable for monitoring air conditions in parks and sports areas attracting a large amount of people who practice various kinds of open air activities.

Knowledge concerning the quality of air in a given place might have a pivotal role in planning the duration of time to spend in the open air, as well as for planning the amount and intensity of training for a given person on a certain day. This seems to be especially important since any form of sustained exercise (including physical labor) increases the body’s need for oxygen, which enhances the amount of the air ventilated by the human lungs (for an overview see, e.g., Bowen, Benson & Rossiter, 2019; Wasserman et al., 2011; Zoladz, Grassi & Szkutnik, 2019). For example the minute ventilation (V_E) which amounts to about 6–8 liters of air for humans at rest, increases to 30–50 liters per minute during exercise of moderate intensity, and exceeds 100 L per minute during strenuous exercise (Wasserman et al., 2011). In some athletes, maximal V_E during strenuous exercise can exceed 200 L per minute, i.e., about 30 times more than at rest (Åstrand & Rodahl, 1986). Increasing the amount of air taken into to the respiratory system increases the quantity of the inhaled suspended particles matter, and thus its deposition in the respiratory track and in other organs of the body. This increases the risk and enhances the severity of several illnesses in humans (for review see, e.g., Bai, Li & Niu, 2020; Lauwers et al., 2020; Pope et al., 2019). Furthermore, knowledge concerning the quality of air can be very useful for patients suffering from cardio-pulmonary insufficiency, and who practice their physical rehabilitation program in the open air (for overview see, e.g., Dyer & Pugh, 2019).

Summing up, the practical outcome of our study can be expressed in the following three points: (i) the best hours to practice physical activities in Kraków in the open air, as judged by air quality, are between 10:00 and 14:00, whereas the worst time for practicing sports outdoors in Kraków is between 18:00 and 22:00; (ii) physical activities in the open air in Kraków during the winter (the season when the air is most polluted) should be limited and moved outside of the town; (iii) elderly people should consider the possibility of spending the winter outside of Kraków or at least to adjust the time spent in the open air in the town according to the air quality on a given day.