Seasonal Variability of Resuspension

. Particulate air pollution in cities is caused by a variety of sources. One of the less -studied contributors is wind-induced particle resuspension. As the wind speed increases, particles are removed from surfaces. These particles cause an increase in the to tal concentration in the air. It is known that particles of 10 - 2.5 µm in size can be resuspended (PM 10-2,5). M odern emission monitoring in cities also allows the monitoring of fine particles of 10, 2.5 and 1 µm in size. The size fractions can then be sorte d into PM 10-2,5 , PM 2,5 -1 and PM 1 . When breathed in, particles of different sizes cause various serious health risks. This paper focuses on the identification of the resuspension process of different particle size fractions by a data processing method. Data measured by automatic emission monitoring are used. It is confirmed that the concentration increase can be dominated by the fraction PM 10-2,5 . However, a concentration increase of fractions PM 2,5 -1 and PM 1 is also evident with increasing wind speed. Although the increase in the PM 1 frac tion is smaller than PM 10-2,5 , it is more severe due to the respiratory deposition dose. The resuspension of particles of different fractions has different behaviours during the year. PM 10-2,5 particles are dominantly resuspended in the summer months. In w inter, on the other hand, the proportion of PM 2.5 -1 and PM 1 particles increases, which may be related to the heating season


Introduction
PM10 air pollution is still a problem in most of the world's metropolises. PM10 is the term for the size of particles with an aerodynamic diameter of less than 10 micrometres. There are many sources of particles in the urban environment. Transport is one of the most emissionsintensive activities. Specifically, it accounts for 25 % of particle pollution in cities on a global scale [1]. As the rise of electric vehicles progresses, the focus is shifting from exhaust emissions to non-exhaust emissions [2]. Their amount may already exceed the exhaust particulate emissions [3]. This trend will only increase in the future. Other sources of urban particulate matter include industry, domestic fire burning, natural resources [1]. Study [1] points to a high degree of uncertainty in the determination of the source of the particles. The number of particles that cannot be classified as the aforementioned sources are then referred to as "unspecified sources of human origin". Wind-induced particle resuspension is indirectly related to human activity. In this process, already deposited particles are released from surfaces into the air by the wind, as already pointed out by [4]. There are several studies showing the characteristics of the wind-induced resuspension process [4][5][6][7][8]. However, the quantification is different each time, as the studies are based on different assumptions.
Wind-induced resuspension depends on many parameters. Among the main ones are time aspects. The initial number of resuspended particles is significantly greater than the number of particles after a period of higher wind speeds. This is caused by the weakly attached particles being lifted at the beginning of the process [9]. Resuspension is a process that occurs in a matter of seconds. Other factor affecting the resuspension is the humidity of the air. The humidity of the air can substantially affect the detachment of particles from the surface if it exceeds a value of about 60 % [10]. Availability of resuspendable particles varies throughout the year [11]. The total concentration of particles in the city is influenced by background concentrations. However, with respect to health risks, we are interested in the overall increase in particle concentration due to wind. It is known that background concentration is also strongly correlated with wind speed [1]. These factors should be taken into account when identifying wind-induced resuspension in the data. The PM10-2,5 concentration can be divided into two components depending on the wind speed. The first component dilutes with increasing wind speed. The second component increases with increasing wind speed [4]. The authors of the study assume that the PM10-2,5 fraction can be resuspended by the wind and the PM2.5 fraction is diluted by the wind. The PM10-2,5 component is diluted by increasing wind speed at the same rate as the PM2.5 fraction. Plotting the dependence of PM10-2,5 on wind speed then results in a characteristic U-shape function.
Other studies [6][7][8], that have dealt with wind-induced resuspension have focused on the PM10-2,5 and PM2.5 fractions. Their assumption is that only PM10-2,5particles are capable of resuspension and the concentration of PM2. 5  each fraction on the wind speed will show how they are able to resuspend. There is a fundamental difference if the wind in a city is lifting larger or smaller particles, especially in terms of health risks.
Due to their size, PM1 particles enter the human respiratory system very easily and deeply, which has a non-negligible impact on health. Compared to larger particles at the same mass, these fractions have a much higher concentration, active surface area, high deposition efficiency and a greater potential for harmful biological interactions with respiratory tissues [12][13][14].
Millions of people walk or run daily on streets and in parks as part of their daily routine, either on their way to work or recreationally. During these activities, large amounts of pollutants are inhaled into the respiratory system. Epidemiological studies [15][16][17][18][19] show a link between respiratory diseases and atmospheric PM. Most of those studies look at total deposition, with deposition of different parts of the respiratory system being important for assessing the potential hazard of PM [20]. Inhaled particles may be deposited in different regions of the respiratory system by the complex action of different deposition mechanisms or may be exhaled. The most important mechanisms are impaction, deposition and diffusion [21]. Particles that encounter the walls of the respiratory tract settle there and are not reabsorbed. The extent and location of particle deposition depends on the size, density and shape of the particles, the geometry of the airways and the mode of respiration. Many mathematical models exist to predict total and regional particle deposition. One possibility is to calculate the respiratory deposition dose (RDD), which expresses the entrainment between regional deposition and mass concentration. The aim of this work is to find a link between resuspension and RDD. The possibility of monitoring PM1 particles raises the question of whether these particles are also capable of resuspension and what health risks arise from this.

Data measurement
The data were obtained from automatic emission monitoring (AIM) in Brno. The AIM in Brno includes five measuring stations: • Svatoplukova -located in a populated part of the city. The measuring equipment is located at the four-lane road. • Výstavište -a location characterised by the presence of a major traffic junction. The measuring equipment is located on a less frequented four-lane road. • Zvonařka -a site characterised by the presence of significant construction activity in 2019. The measuring device is located along a four-lane road. • Arboretum -the measuring device is located directly in the center of the city park.
• Lány -a site characterised by the presence of residential development. The measuring equipment is located between houses. Data were measured throughout 2019. The frequency of recording was 1 second. However, the data is reprocessed into 10-minute averages by default. Measured variables included: temperature, humidity, wind direction, wind speed, pressure and PM10, PM2.5 and PM1 concentrations.

Data processing
Resuspension is a process that does not occur under all conditions. There are times of the year when, despite increased wind speeds, there is no increase in concentration. In order for the resuspension process to be recognized from the data, the data must be prepared according to predetermined conditions. Due to the availability of particles at different times of the year, the resuspension process has seasonal variability. Thus, the data were sorted into four seasons. Hourly averages of the measured data are mostly used to indicate the resuspension history [4,5]. Resuspension occurs within seconds [9], the shortest available measured time intervals, 10 minutes, were thus used. Air humidity above 60 % can significantly affect the removal of particles from the surface [10]. Therefore, only data with humidity below 60 % were used. Subsequently, the particle concentrations were sorted into three size fractions: The function of concentration on wind speed for particle size fractions, the sorted data were fit with a polynomial. The degree of the polynomial was determined by the k-Fold crossvalidation method [22], where the dataset is split into training and test subsets. Subsequently, the root mean square error (RMSE) determining the deviation of the test subset of the real data from the polynomial curve is determined. This process was carried out k times and after its completion, the average of k RMSE values was calculated for a given polynomial degree. Subsequently, the polynomial with the lowest resulting RMSE was selected. The degree of the polynomial can therefore vary for each data filtering condition (season, location and fraction) just based on the RMSE results. Within this paper, the value of k was set to 10, which is the value standardly used in this method across disciplines. Local extremes were then determined on the curves for subsequent analysis of resuspension rates.

Evaluation of the data
The original methodology for evaluating resuspension characteristics is based on the procedure [4]. The component of particles that can resuspend from the PM10-2,5 fraction is obtained by subtracting two functions. The first function is the rate at which particles of the PM10-2,5 fraction are diluted with increasing wind speed. This rate is the same as the rate at which PM2.5 particles are diluted. The second function is the dependence of the PM10-2,5 concentration on wind speed itself. Thus, the wind speed is identifiable when the resuspension event is visible. At the same time, the rate of increase in particle concentration due to resuspension is determined. However, based on the known concentration of PM1, the behaviour of other size fraction of PM2.5-1 and PM1 can be determined. These classes may have different behaviour depending on the wind speed.
Since the aim of this work is to evaluate how the concentrations of different particle size fractions behave as a function of wind speed, this approach is not applicable. Thus, the presented method of data processing and evaluation provides an indication of which particle size fraction concentration varies the most due to wind in the specified seasons. The increment of the concentration of the size fractions to the total increase in PM10 concentration is known. The method itself consists of the following steps: 1. Identifying the best-fit of the relationship between particle concentration and wind speed. 2. Locating the extremes of given functions. 3. Determining the wind speed interval where the PM10 increase occurs. 4. Deriving the concentration increase or decrease from the extremes of the functions for each particle size fraction in the specified wind speed interval. 5. Determination of the concentration increment of the individual particle size fractions to the total particle concentration increment. Furthermore, health risks from wind-induced particle resuspension may be identified. At different seasons, particles of different size fractions are resuspended. Regarding health risks, it then makes a difference which size fraction caused the increase in PM10 concentration due to increasing wind speed.

Estimation of the respiratory deposition doses
Atmospheric particles inhaled during breathing are deposited on the lining of the airways and are not exhaled, which can cause serious breathing disorders. A mathematical model based on calculations and data from the International Commission on Radiological Protection (ICRP) has been used to determine the extent of particle deposition in different regions of the respiratory tract. Respiratory deposition dose [23][24][25] is calculated according to the equation (1): where VT is the breathing rate (m 3 /inhale), f is the frequency of breathing, DFi is the deposition fraction of the size fraction i, PMi is the mass concentration of size fraction i. Breathing volume and respiratory rate depend on physical activity, but also on gender and age. The different values are given in Table 1. People usually perform little activity, mostly walking, when exposed to particles from resuspension, and because of this, the values for light physical activity were chosen and the calculation was calculated for men as well as women and children. The deposition size fraction is calculated in three different regions of the respiratory system -the head airway (HD), the tracheobronchial region (TB) and the alveolar region (AL) according to Hinds [20]. They are calculated as follows: = � 1 1 + exp (6.84 + 1.183 ln ) DFHD is the deposition fraction for the head airways, dp is the particle size, IF is the inhalable fraction, which is calculated as follows: DFTB denotes the deposition fractions in the tracheobronchial region and is given by the following equation: The particle mass concentration was used for the calculation, which corresponds to the increase in particles due to wind-induced resuspension. Consequently, the calculated RDD value corresponds to the value that a person would breathe while walking in the city with wind blowing at the speed at which the increase in particle concentration was identified.

Results and discussion
Based on the explained methodology, the data were processed and evaluated. The data are divided into four seasons in five locations. For each location, the concentration of PM10, PM10-2,5, PM2.5-1, PM1 (PM2.5 fraction is also evaluated for illustration) is evaluated depending on the wind speed. The data were fitted with a polynomial in a way that best fits the nature of the data. An example of the polynomial translation of the data is shown in the following figure. For each season and location, it was determined at which wind speeds PM10 particles increase. Within a specified range of wind speeds, the contribution of each size fraction to the increase in PM10 was determined. The concentration increase was calculated as the difference in the concentration maxima and minima of the polynomials for each particle size fraction over the observed wind speed range. The particle size fraction dominantly responsible for the increase in PM10 concentration due to wind is summarized in the following table. Table 2. Fraction of particles dominantly responsible for the increase in total concentration due to particle resuspension.

Location Arboretum Lány Výstavište Svatoplukova Zvonařka
Spring The table shows that PM10-2,5 is a fraction, which is predominantly resuspended by wind, confirming the assumption of the studies [4,8]. However, a further assumption of a decrease in PM2.5 concentrations with increasing wind speeds could not be confirmed. In the vast majority of cases, there were increasing concentrations of both PM2.5-1 and PM1 fractions due to wind. This phenomenon can be caused by fitting the data with a polynomial. Previous studies used a second-degree polynomial, which could suppress small changes in concentration with a wind speed. The behaviour of the size fractions on the wind speed interval where the increase in PM10 occurred is evident in Tab. 2. The locations where there was an increase in concentration, a decrease in concentration, or the concentration was constant are marked.  Table 3. Concentration behaviour of the particle fraction due to an increase in wind speed. It can be seen from Tab. 3 that the increase in concentration was evident in almost every season and location. Based on Tab. 2, it is known that the PM10-2,5 fraction is dominantly responsible for the increase in PM10 due to wind. However, Tab. 3 shows that the concentration increase is also recorded for other fractions. From the point of view of health risks, it is important to know how concentrations increase for all fractions and seasons. Almost every location is dominated by an increase in the PM10-2,5 fraction. Sites that showed similar behaviour across seasons were Výstavište, Zvonařka and Lány. Here, the contribution of the fine fraction PM1 to the overall concentration increase decreased during the spring and summer months. The increase in the PM2.5-1 fraction remained approximately constant or did not change at all during the year. During the autumn and winter months, the increase in the PM1 fraction increased. However, the overall increase in the concentration of individual fractions does not indicate a problem related to health risks. Particle size is key with respect to health risks, as smaller particles settle deeper in the respiratory system. Although the increase in the PM10-2.5 fraction is dominant, the increase in PM1 concentration has a more negative impact on human health. The following graph shows the number of resuspended particles that can settle in different parts of the respiratory system. Comparing Fig. 4 and Fig. 5, it is evident that although the increase in concentration is dominated by the PM10-2,5 fraction, more PM1 particles are trapped in the human body. The PM1 fraction is a minority fraction in the concentration increase in most cases. Considering the total amount of trapped particles in the human body, the fraction of PM1 is not negligible, especially in head airways region. Similarly, the PM2.5-1 fraction is too significant due to its concentration increase. However, due to health risks, the proportion of trapped particles has increased, especially in the alveolar region. In tracheobronchial region, the fraction of PM10-2,5 particles captured is dominant, as in the overall concentration increase in Fig. 5. Thus, the results show that PM1 particles are trapped in the head airways despite the low total concentration. PM10-2,5 particles are predominantly trapped in the tracheobronchial region. The alveolar region is dominated by PM1 capture, and it should be noted that PM2.5-1 particles also make a non-negligible contribution in this region. Gender and age don't influence the results.

Conclusion
The present study focuses on the identification of the wind-induced resuspension process in cities. Emission monitoring data are used. The data are divided into four annual periods. Times when the resuspension process is suppressed and humidity reaches values above 60 % are not evaluated [10]. The remaining data were separated into size fractions PM10-2,5, PM2.5-1 and PM1 and expressed as a mathematical dependence on wind speed. From the resulting functions, it was determined which fraction was responsible for the increase in PM10 concentration due to increasing wind speed. It was found that particles with the size fraction PM10-2,5 resuspended at a dominant rate, which is in agreement with previous studies. The original assumption of a decrease in PM2.5 concentration with increasing wind speed was not evident. This may be due to the fact that the PM2.5 fraction has split into PM2.5-1 and PM1. The methodology used identified a concentration increase in the PM2.5-1 and PM1 fractions. However, this increase cannot be attributed with certainty to resuspension alone. The source of these particles may be traffic or other sources [4][5][6][7][8], which the wind carries. A more detailed data analysis would be needed to confirm or disprove this. It would be necessary to link the particle concentration information to the NOx concentration.
With increasing wind speed, the concentration of all size fractions increases in each season. Whatever the source of the particles, this increase has an impact on human health. Although the increase in PM10-2,5 fraction has been dominant, the increase in PM1 concentration is more serious for health. These particles are inhaled more by humans compared to PM10-2,5. Within the respiratory system, different fractions are trapped in different parts. The PM1 fraction is trapped in the head airways. The PM10-2,5 fraction in the tracheobronchial region and the PM2.5-1 fraction in the alveolar region.