The Effect of Drought on PM Concentrations in the Czech Republic

Received: March 14, 2022
Revised: June 24, 2022
Accepted: August 1, 2022

 Copyright The Author's institutions. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited. 

Cite this article:

Šmejkalová, A.H., Brzezina, J. (2022). The Effect of Drought on PM Concentrations in the Czech Republic. Aerosol Air Qual. Res. 22, 220130. https://doi.org/10.4209/aaqr.220130

HIGHLIGHTS

• A difference was found in PM_x concentrations depending on the extent of drought.
• A relation was found between the number of consecutive dry days and PM₁₀ conc.
• A correlation was found between PM_x conc. on dry days and soil temperature and moisture.

Keywords: Air pollution, Drought, Particulate matter, Soil erosion

1 INTRODUCTION

The air quality in Europe has improved significantly over the last few decades (Butt et al., 2017; Colette et al., 2016). Implementing a new EU directive to member state laws (EEA, 2020) forced major emission emitters to modernize technologies and decrease total emissions. The results of the regulations reflecting EU directives are apparent in anthropogenic emission inventories over the last few decades (Colette et al., 2016; EEA, 2018). However, air quality depends not only on the actual amount of emitted pollutants, but also on meteorological conditions (Ding et al., 2021; Zhang et al., 2021) and several other factors, for example, reactions of chemical entities in the atmosphere (the so-called secondary reactions). With regard to meteorological conditions, for example low temperatures in combination with poor dispersion conditions during wintertime represent a very unfavorable condition often leading to significant increase in pollution levels—low temperatures increasing heating intensity and thus emissions from heating, poor dispersion conditions leading to accumulation of pollutants close to their sources and near ground level. (Cichowicz et al., 2017; Lv et al., 2020; Shi et al., 2015).

One of the consequences of the current ongoing climate change is a more frequent occurrence and intensity of heat and drought waves (more detailed is described e.g., in Perkins-Kirkpatrick and Lewis (2020)). These changes can also affect the concentrations of certain air pollutants. Several major heatwaves associated with dry periods were observed in Europe in the past decades, most recently in 2015 and 2018 (Lin et al., 2020). With regard to air quality, the most significant effect is associated with meteorological and agricultural drought. A meteorological drought can be quantified based on the precipitation amount. Agricultural drought is defined by the low availability of soil moisture which adversely limits the crop yield (Mannochi et al., 2004). An agricultural drought may cause damage to vegetation that presumably influences soil erosion.

Drought can affect air quality directly by increasing dust concentrations or changing the atmosphere's chemistry. Lack of precipitation reduces the ability of the atmosphere to decrease pollution levels by wet scavenging (Vicente-Serrano et al., 2020; Wang et al., 2017). In the past, many times drought resulted in large dust storms. Changes in the chemistry of the atmosphere caused by drought usually manifest themselves by changes in ground-level ozone concentrations, which is formed via photochemical reactions.

Drought can, however, also have an indirect effect on the concentrations of air pollutants. An example of an indirect effect of drought on air quality are forest fires, during which a significant amount of naturally produced emissions are released into the atmosphere (Vicente-Serrano et al., 2020). The impact of forest fires on air quality was reported in Bo et al. (2020). The sensitivity of fine dust to drought was demonstrated by Achakulwisut et al. (2018). This study suggests that drought, driven by climate change, is a potential risk for public health due to increased dust concentrations.

The relationship between drought and PM_x concentrations was studied during the winter period (Hu et al., 2019; Liao et al., 2020). Our study, however, focuses on the effect of drought on PM_x concentrations during the vegetation season to exclude the effect of increased PM_x concentrations from heating emissions and because in the warm half-year the soil is not potentially frozen or covered with snow. The connection between soil temperature and moisture and PM_x concentrations was also investigated as the soil is a potential source of aerosol particles. It was assumed that soil temperature influences PM_x concentrations more significantly compared to air temperature in the non-heating season. Despite the fact that air temperature correlates with soil temperature (Islam et al., 2015; Yeşilirmak, 2014), they are not equal and the deeper in the soil one measures, the greater the difference can be. In general, soil temperature shows much less variability during the day as well as during the year than air temperature.

The presented study focuses on study influence of drought to air quality (namely PM₁₀ and PM_2.5 concentrations) in the Czech Republic. Since the other studies from our region aimed at drought and heat waves from a meteorological point of view (e.g., Trnka et al., 2016; Vautard et al., 2007), this kind of evaluation is unique in the Czech Republic. Main goal is to investigate some patterns of the relation between drought and PM₁₀ concentrations to understand more deeply, what influence on air quality can be expected during dry periods. Based on theoretical knowledge the authors stated a hypothesis that drought could have potentially negative effects on air quality during the vegetation season.

 
2 METHODS

 
2.1 Station Selection

There is a well spatially distributed network of ambient air quality monitoring stations (National Air Quality Monitoring Network (NAQMN)) and meteorological and climatological stations operated by the Czech Hydrometeorological Institute in the Czech Republic. The high number of ambient air quality monitoring stations (approximately 200 stations) was filtered for the purposes of this study, and only stations with PM₁₀ concentrations data over the entire period of analysis (between 2010 and 2019) were selected. The second criterion for station selection was that precipitation data from a professional meteorological station located within a radius of 200 m from the air quality monitoring point (ambient air quality stations themselves are not equipped with a rain gauge) is available. Given the very high spatial and temporal variability of precipitation, it is crucial only to select stations that are relatively close to each other. Given that precipitation data was taken from a location at most 200 m away from the air quality monitoring site, one can assume that the precipitation amount is equal or very similar to the precipitation amount that would be measured directly at the air quality measuring point. The above described selection criteria were satisfied by a total of 21 stations (Fig. 1).

Fig. 1. Spatial distribution of the ambient air quality monitoring stations data of which were used in this study. Circles show the position of stations with PM₁₀ measurement with colour representing that particular station type. Rural stations are coloured green, suburban blue and urban stations red. Stations that in addition to PM₁₀ also monitor PM_2.5 concentrations are distinguished by a black dot in the middle of the circle. The map data were processed in ArcGis programme (ESRI, 2018).

Ambient air quality monitoring stations within the NAQMN are classified based on the classification scheme for the purposes of Exchange of information (EoI) listed in the Council Decision 97/101/EC. This classification was implemented into the Air Quality Information System (AQIS) database, which collects and stores data from all the NAQMN stations. In the above table (Table 1) station type is represented by two letters separated by a slash. First letter indicates the type of station—traffic (T), industrial (I) or background (B). Second letter specifies the type of zone Urban (U), Suburban (S) or Rural (R) (CHMI, 2019a) (Table 1).

 
2.2 Drought Specification

In meteorology there is currently no exact consensus as to how a dry day or rain episode should be defined. However, for a day to be classified as a “wet day” (rain), commonly used threshold values for daily precipitation amount are 0.1, 0.2 and 1.0 mm measured between 06–06 UTC (ČMeS, 2017). The actual threshold value may depend on rain gauge measurements resolution (WMO, 2018), or the purpose of the study, also taking into account the particular studied area. This amount can, therefore, vary from 0.1 to 5.0 mm (Reiser and Kutiel, 2009). The calculation of the long-term meteorological normals considers number of days with rainfall of ≥ 1.0 mm (WMO, 2017). This study uses a threshold value of 0.2 mm (rain day is a day with precipitation amount > 0.2 mm, dry day ≤ 0.2 mm). The threshold value of 0.2 mm was chosen to avoid rain gauge measurements resolution errors and experience; the same threshold value was used for example in the study Scafetta and Mazzarella (2021).

Apart from having an exact definition of a dry day, also an exact definition of a dry episode is needed. Since studying the effect of drought on PM₁₀ concentration is the goal of this study, and the resuspension effect during drought is expected, the influence of moisturizing the ground by rain must be minimized. A dry episode was defined as five consecutive dry days (five consecutive days with a total precipitation amount of ≤ 0.2 mm on each particular day). A wet episode was defined as day or number of days with precipitation amount > 0.2 mm.

 
2.3 Data Measurement and Data Quality

PM₁₀ and PM_2.5 concentration data used in this study were measured within the NAQMN by both automated stations and samplers. Automated stations are equipped with analysers, which provide data in real-time. In contrast, samplers sample particles on a filter and these are subsequently transported to a laboratory, where a gravimetric determination is performed to determine PM concentrations (Table 2). The determined PM₁₀ and PM_2.5 concentration data fulfils the criteria according to the Annex 1 of Directive 2008/50/EC and Annex IV of Directive 2004/107/EC for minimum data availability of 90% (CHMI, 2019b). Meteorological data (in this case, precipitation, wind speed, soil temperature and soil moisture) was measured according to the WMO specifications.

 
2.4 Period of Analysis

Level of air pollution (in this case, PM₁₀ and PM_2.5 concentrations) has gradually been decreasing over the last three decades in the Czech Republic. The amount of emissions released into the ambient atmosphere began to drop in 1990s in response to the new Act No. 309/1991 Coll. on air protection, coming into force. During the next several years, air quality protection was improved by adopting a new legislation (Act No. 76/2002 Coll and Act No. 201/2012 Coll), reflecting EU directives (CHMI, 2019a). Similarly to other gaseous pollutants emissions, the aerosol particle emissions also significantly decreased by tens of percent between 1990 and 2018 (Fig. 2). However, the actual measured concentrations are not directly proportional to the emission levels and instead also reflect other factors, such as meteorological conditions. The effect of meteorological conditions is manifested in the course of PM₁₀ concentrations at stations in the Czech Republic between 2001 and 2016. After 1999, the PM₁₀ and PM_2.5 emissions were more or less similar. However, PM₁₀ concentrations began to increase in 2001, and higher concentrations were observed until 2006; a similar situation was observed in 2010 (Fig. 3). Year 2010 was characterized by unfavourable dispersion conditions, especially in the winter. Unfavourable meteorological situation can be quantified by a characteristic referred to as a degree day (D21- defined as the sum of differences in the indoor temperatures (21°C) and the average daily outdoor temperatures on heating days). D21 reflects the effect of meteorological conditions on emissions; this is described in detail in CHMI (2020), (Fig. 3). The period of analysis in this study is the ten-year long interval between 2010–2019, where always a 6-month period between April and September was analysed in each year. The period of analysis of this study saw no major changes in legislation related to air quality that could have a significant effect on emissions.

Fig. 2. The course of PM_2.5, PM₁₀ and total suspended particle emissions (TSP) in the Czech Republic, 1990–2018. Source: CHMI (2019a).

Fig. 3. Course of PM₁₀ average annual concentrations and annual heating season characteristic expressed as degree days (D21) in the Czech Republic between 2001–2017. Shaded regions represent years with increased PM₁₀ concentrations. The individual series are shown for various station types specified in the legend. Adapted from: CHMI (2018).

 
2.5 Drought and Wetness Ratio

Based on the criteria described in Chapter 2.2, a parameter R_d/w representing the ratio between mean PM₁₀ concentrations in dry and wet period was calculated as:

 
2.6 Long-range Transport

Long-range transport was explored through backward trajectories calculated by the HYbrid Single-Particle Lagrangian Integrated Trajectory HYSPLIT_4 model (Draxler and Rolph, 2013). 72-hour backward trajectories were recalculated every 24 hours with Global Data Assimilation System (GDAS) meteorological data with 1° × 1° grid resolution used. The start time at 6:00 UTC was set to coordinate with the time when the daily PM_x concentrations averages are counted. Since the resolution of the HYSPLIT model is smaller than the geographical distance of the stations, only one receptor site was selected - the station in Košetice, which represents the centre of the Czech Republic, height of the receptor was set to 500 m AGL. The total amount of 1748 backward trajectories were clustered into 6 clusters representing air masses with different origins.

 
2.7 Soil Temperature and Moisture

Soil temperature and moisture are only monitored at some meteorological stations. For the purposes of this study, a rural station in Košetice was used as it has a complete data series of soil temperature and moisture measurements as well as a complete series of PM₁₀ and PM_2.5 concentrations data.

Soil temperatures are taken at depths of 5, 10, 20, 50 and 100 cm. Soil moisture is measured for three different levels of 0–10 cm, 10–50 cm and 50–100 cm deep. It can be assumed that resuspension from the surface will mostly be affected by the moisture and temperature of the upper-most layer of the soil. The analysis therefore focused on the 0–10 cm depth data for soil moisture and 5 cm data for soil temperature.

The dataset used consisted of data from the exact same period, i.e., vegetation seasons (April–September) between 2010 and 2019. 3 RESULTS PM₁₀ concentrations were generally higher in the dry period during the whole period of analysis. Differences between various station types did not exceed 16.0%. Rural and urban stations differed by 15.6%. The average R_d/w at rural stations was 141.2%, at suburban 134.5% and urban 125.6%. PM₁₀ concentrations median at rural stations was 19.8 µg m^–3 and 14.0 µg m^–3, at suburban 21.4 µg m^–3 and 15.9 µg m^–3 and at urban 21.1 µg m^–3 and 16.8 µg m^–3 in dry and wet periods respectively. The lowest median at rural stations was measured at Churáňov station 12.0 µg m^–3 and 8.0 µg m^–3 in dry and wet periods. In contrast 26.8 µg m^–3 and 19.4 µg m^–3 was recorded at Tušimice station in dry and wet periods. The most manifested variability was at Stankov station (suburban) where interquartile span was 15 µg m^–3 in dry period. Reuslt from urban station were very similar. Highest median was measured at Tábor station 23.1 µg m^–3 in dry period and 17.5 µg m^–3 in wet period (Table S1, Fig. 4).

Fig. 4. Overview of PM₁₀ concentrations during wet and dry periods April–September 2010–2019. Light green – rural stations, dry period; dark green – rural stations, wet period; light blue – suburban stations, dry period; dark blue – suburban stations, wet period; light red – urban stations, dry period; dark red - urban stations, wet period.

Relationship between the data measured in dry and wet periods at different types of stations was tested by the nonparametric Dunn test with Holm method adjustment (Dinno, 2017, 2015; Dunn, 1964), with the null hypothesis being that there is no difference between the groups. Overall, there were statistically significant differences in grouped data of PM₁₀ concentrations for each station type between a dry and wet period (Table S2).

 
3 RESULTS

 
3.1 Monthly Variability

PM₁₀ concentrations were highest in April at all station types. This is most likely due to occasional occurrence of cold days resulting in various extent of effect of domestic heating in this spring month. The effect of growing plants and crops on fields can be observed at rural stations.

The beginning of vegetation growth and low cover at fields was pronounced from April to May by a similar R_d/w 132.0 ± 3.0%. The lowest R_d/w in July (full vegetation cover) was reflected in the highest R_d/w 161.2, and 158.9% in August and September (Fig. 5). Suburban stations R_d/w ratio is more or less the same from April to May 124.4 ± 1.2%. Highest R_d/w was recorded in August at both rural and urban stations. Urban stations are characterized by a very small difference between PM₁₀ concentrations during dry and wet period in May and June, also observed at suburban stations. Highest R_d/w was observed in August—regardless of the station type. The second-highest R_d/w was found in September.

Fig. 5. Monthly variability in PM₁₀ concentrations for different station types. Light green - rural stations, dry period; dark green – rural stations, wet period; light blue – suburban stations, dry period; dark blue – suburban stations, wet period; light red – urban stations, dry period; dark red - urban stations, wet period.

Detailed comparison of the differences between PM₁₀ concentrations in dry and wet periods in the individual months and at the individual stations shows statistically significant differences. The highest percentage of statistically significant differences between dry/wet PM₁₀ concentrations was observed in August and September at all types of stations; most cases were recorded at the suburban station type. Which months had the lowest number of statistically significant differences depends on the type of station—rural and urban stations showed no statistically significant differences in May and July, and only one significant result in April. For suburban stations there is one statistically significant dry/wet difference in April and May (Table S3).

 
3.2 Effect of Cumulative Drought and Wet

Effect of long-term drought or wet period on PM₁₀ concentrations was analysed by looking at the dependency of PM₁₀ concentrations on the number of consecutive days marked as either dry or wet. Data was evaluated only to the threshold value, which was represented by the 95^th percentile of the total number of cumulative number of days with drought or with precipitation (see Chapter 2.2). Periods with consecutive number of days above this 95^th percentile value were not evaluated. The cumulative threshold (95^th percentile) value for the dry period was 10 days, and for the wet period 70 days. Time interval 0–10 or 0–70 day was divided into several categories in 2-day steps. Rest of the values above the threshold were not taken into account due to a low number of cases (not exceeding 14%) and to limit the influence of the extreme values. PM₁₀ concentrations generally increased as the cumulative number of days with drought also increased. The increase is more apparent in the interquartile range, especially after the 5^th day at rural and urban stations where a gradual growth of concentrations was recorded. Some fluctuations were observed at suburban stations, however, an increasing trend was also observed. Statistic significance was proved only at rural and suburban stations between groups of 1–5 and 5–10 cumulative days of drought (Table S4).

In contrast, an increasing number of consecutive wet days resulted in a general decrease in PM₁₀ concentrations. This effect of precipitation was apparent at all station types by a decrease of PM₁₀ concentrations at the beginning of the wet period (first three days). The dependency of PM₁₀ concentrations on cumulative days with precipitation was most pronounced at rural stations. Suburban and urban stations showed some variability in the increasing number of consecutive wet days. Close to the threshold of 70 days, one can observe some small increase of PM₁₀ concentrations. Another mechanism that probably influenced these values, were increasing local emissions or regional transport, which can overcome the wet scavenging effect of precipitation (Fig. 6). Statistic significance between groups of 1–35 and 35–70 cumulative days of wetness was manifested at all types of stations (Table S4).

Fig. 6. PM₁₀ concentration dependency on cumulative number of dry and wet days. Station indication: light green – rural stations, dry period; dark green – rural stations, wet period; light blue – suburban stations, dry period; dark blue – suburban stations, wet period; light red – urban stations, dry period; dark red - urban stations, wet period.

 
3.3 Meteorological Conditions

Some of the stations of analysis are also equipped with meteorological measurements, these data were evaluated to find out how meteorological conditions influence the PM₁₀ concentrations during dry and wet periods. Data of wind speed and wind direction were available for 13 stations. PM₁₀ concentrations and wind speed were compared by Spearman coefficient R_s. Results confirmed a weak relationship between these variables in both wet and dry period. Prevailing correlation was negative and R_s values were very similar for both these periods. Highest R_s was found at the Brno-Tuřany station, where R_s in dry period was –0.25 and in wet period –0.48 (Fig. S1). Dependence of PM₁₀ concentrations on wind speed and wind direction was analysed for only 3 stations (Košetice – rural, Prague-Libuš – suburban and Tábor – urban)—one representative station for each station type. The Conditional Probability Function (CPF) was calculated and visualized by polar plots (Carslaw and Ropkins, 2012), a critical value was set to 75^th percentile of PM₁₀ concentrations. Results confirm major differences between dry and wet period at Košetice station, not at the Prague-Libuš and Tábor (Fig. S2). Comparison of PM₁₀ concentrations and temperature measured at 2-meter height was available for 14 stations. Similar R_s results were found as in the case of wind (Fig. S3). These results are probably influenced by relatively low PM₁₀ concentrations during the studied period compared to high concentrations in cold part of year when the relationship between PM₁₀ concentrations and temperature can be expected to be much stronger. For example, R_s of PM₁₀ concentration and temperature in winter (2010–2019) at the station in Košetice is –0.76.

 
3.4 PM_2.5/PM₁₀ Ratio

The measurement of PM_2.5 concentrations is not performed at all stations included in this study. The ratio between PM_2.5 and PM₁₀ was thus calculated only for 13 stations where PM_2.5 data were available. Since various pollution sources produce various ratios of fine (PM_2.5) and coarse particles (PM_10-2.5), the PM_2.5/PM₁₀ ratio can provide information about potential particle origin and its source. Anthropogenic sources, including primary and secondary particles, result in a higher PM_2.5/PM₁₀ ratio; in contrast, a relatively high ratio of coarse particles (primary particles) is produced by natural sources (for example soil erosion), manifested by a lower PM_2.5/PM₁₀ ratio (Spandana et al., 2021; Xu et al., 2017; Zhao et al., 2019). The PM_2.5/PM₁₀ ratio varies at the individual types of stations. There is no particular ratio typical for a specific station type (Munir, 2017; Putaud et al., 2010) for example, the ratio in the range 0.60 to 0.70 is characteristic for both rural and urban stations (Munir, 2017). In case of this study, in general, the PM_2.5/PM₁₀ ratio was lowest at rural stations and highest at urban stations (0.66 and 0.74 respectively). This means that a rural station had nearly 10% higher ratio of coarse particles than an urban station. Surprisingly the difference between dry and wet period PM_2.5/PM₁₀ ratios at individual station types was very small (< 1%), no statistically significant results was found (Fig. 7, Table S5).

Fig. 7. PM₁₀ and PM_2.5 concentrations and their ratio for a dry and wet period. Station indication: light green – rural station, dry period, dark green – rural station, wet period; light blue – suburban station, dry period; dark blue – suburban station, wet period; light red – urban station, dry period; dark red – urban station, wet period. White circles indicate PM_2.5/PM₁₀ ratio for a dry and wet period. Error bars show the standard deviation.

The PM_2.5/PM₁₀ ratio during the individual months shows similar patterns at all stations only in April and May when the dry period ratio is lower compared to the wet period. At rural stations, an apparent change occurred in June; after June, the dry period ratio indicates a higher amount of PM_2.5 in the atmosphere than during the wet period. Analogous behaviour was observed at suburban and urban stations, but the change occurred one month later, in July. This change in the PM_2.5/PM₁₀ ratio between dry and wet period in the individual months may be due to the contribution of secondary particles in the air. While concentrations of coarse particles may be elevated or resuspended in April and May (in June at suburban and urban stations) during dry period because of the lack of vegetation cover on fields, secondary particles formation and other anthropogenic activities can contribute to fine particles in the summer months (Fig. 8). The PM_2.5/PM₁₀ ratio is very similar in September when the difference in the mean air temperature between dry and wet period is minimal (Fig. S4). Overall, the PM_2.5/PM₁₀ ratio observed in this study shows similar pattern as was reported in Gehrig and Buchmann (2003) where secondary increase of this ratio in summer months was recorded.

Fig. 8. PM_2.5/PM₁₀ ratio for the individual months during the dry and wet period. Lines represent median, bars indicate 5^th and 95^th percentile (bars with borders show results of dry period). Station indication: light green – rural stations, dry period; dark green – rural stations, wet period; light blue – suburban stations, dry period; dark blue – suburban stations, wet period, light red – urban stations, dry period, dark red - urban wet season.

 
3.5 Effect of Long-range Transport

PM₁₀ concentrations during dry and wet periods were analysed also in relation to long-range transport (PM_2.5 analyses were excluded for lack of data and small representativeness). Each of the six clusters was of continental origin. Clusters number 2 and 4 were aged air masses, cluster number 3 was fresh air mass (Fig. 9(a)). The highest PM₁₀ concentrations were measured in cluster number 6, during both periods. The increased contribution of pollution was probably caused by emissions from the region around Austria's capital. Lower concentrations in cluster number 3 prove the freshness of this air mass (Fig. 9, Table 3). The influence of air masses of different origins on the concentration of PM₁₀ in dry and wet periods has not been proven. There are no statistically significant differences between PM₁₀ concentrations in the individual clusters in dry and wet periods (Table S6). Wilcox-Mann-Whitney test was used for this comparison (Bauer, 1972; Hollander et al., 2013).

Fig. 9. Statistical cluster analysis of air mass backward trajectories showing dependency of PM₁₀ concentrations in dry and wet periods on different air mass origins at rural, suburban and urban stations. Receptor site Košetice station was selected as a representative of the centre of the Czech Republic. Each cluster has its own color, black horizontal line represents the median, borders of boxes show 25^th and 75^th percentile, error bars indicate the overall minimum and maximum values.

 
3.6 Soil Temperature and Moisture

The analysis looked at the PM₁₀ and PM_2.5 concentrations and their relationship with soil temperature and moisture as such, regardless of wet or dry period as it can be assumed that soil moisture on its own is an indicator of dryness or wetness. The relationship between air temperature and soil temperature at the individual depths was evaluated by R_s (Table S7) and is visualized in Fig. S5. Further evaluation is focused only on the effect of soil moisture (H_s) and soil temperature (T_s) on PM_x concentrations. The first part was devoted to their individual relations. Overall soil moisture at the Košetice station ranged from 13 to 74% Distribution of soil moisture and soil temperature values is plotted in Fig. S6.

The soil temperature (T_s) values ranged from 0.9 to 25.0°C. T_s data were rounded to an integer, and data below 5.0°C were excluded from further analyses as there is a probability that high PM₁₀ and PM_2.5 concentrations at very low temperatures (T_s < 5.0°C) are associated with domestic heating. The impact of moistured soil on reduced particle resuspension is apparent in a decrease of both PM₁₀ and PM_2.5 concentrations when H_s exceeded 30.0% (more pronounced in case of PM_2.5 concentrations) (Figs. 10(a) and 10(c)). PM₁₀ average concentration was 17.0 µg m^–3 (H_s < 30.0%) and 15.6 µg m^–3 (H_s ≥ 30.0%), PM_2.5 average concentration 11.3 µg m^–3 (H_s < 30.0%) and 10.4 µg m^–3 (H_s ≥ 30.0%). Statistically significant differences were proved between the data series (Table S8). PM₁₀ and PM_2.5 concentrations stop fluctuating and start increasing when T_s exceeds 15.0°C (Figs. 10(b) and 10(d)). PM₁₀ average concentration was 16.0 µg m^–3 (T_s < 15.0°C) and 16.3 µg m^–3 (T_s ≥ 15.0°C). Surprisingly, opposite behaviour was observed for PM_2.5 with an average concentration of 11.1 µg m^–3 (T_s < 15.0°C) and 10.6 µg m^–3 (T_s ≥ 15.0°C). Only PM_2.5 concentrations were statistically significantly different under diverse T_s conditions (Table S8). The situation is clearer when the conditions are focused on only part of the data where PM₁₀ and PM_2.5 concentrations start to decrease (in case of H_s ≥ 36.0%) or increase (in case of T_s ≥ 20.0°C) (Fig. 10). Average concentrations are listed in (Table 4).

Fig. 10. Boxplots of soil moisture (H_s0–10cm) and soil temperature (T_s0–5cm) during different levels of PM₁₀ and PM_2.5 concentrations. Black horizontal line is median, borders of boxes show 25^th and 75^th percentile, error bars indicate minimum and maximum values.

Since the results of PM₁₀ and PM_2.5 under set soil conditions were statistically significant different (Table S9), data series were split into a combination of soil conditions (H_s ≥ 36.0% T_s ≥ 20.0°C, H_s ≥ 36.0% T_s < 20.0°C, H_s < 36.0% T_s < 20.0°C, H_s < 36.0% T_s ≥ 20.0°C) to investigate which soil conditions influence the PM_x concentration the most. PM_2.5 concentrations have insufficient data available for this analysis, thus only PM₁₀ results were used. High PM₁₀ concentrations at high temperatures can be explained as the result of drought and resuspension. Data series of PM₁₀ concentration at soil moistures H_s < 36.0% T_s ≥ 20.0°C differs from the rest of the analysed H_s and T_s combinations on a statistically significant level (Table S10).

This is a clear indicator that high PM₁₀ concentrations are the result of dry conditions and the associated resuspension or soil erosion Moreover, results were further proved by the fact that at high temperatures, the concentrations are lower at higher soil moisture values (e.g., when temperatures are above 20°C and soil moisture is higher than 36%, average PM₁₀ concentration is only 16.3 µg m^–3) (Fig. 11).

Fig. 11. Boxplots of PM₁₀ concentrations during different levels of soil moisture (H_s0–10cm) and soil temperature (T_s0–5cm). Black horizontal line is median, borders of boxes show 25^th and 75^th percentile, error bars indicate minimum and maximum values.

4 DISCUSSION

The relationship between drought and air pollution has already been studied in other scientific papers. For example, a study by Achakulwisut et al. (2018) focused on the relationship between fine dust concentration and the 2-month Standardized Precipitation Evapotranspiration Index (SPEI02) anomalies. Similarly to the presented study, a correlation and a negative relationship between the two parameters was found. The study talks about the effects of soil moisture on wind erosion in a controlled experiment in a wind tunnel and it was found that the wind speed threshold increases with soil moisture—i.e., at a higher soil moisture value, a higher wind speed value (threshold) is required for erosion to occur. The presented study did not look at the correlation between wind speed and soil moisture directly, however, when looking at the results of this study, one can see that at higher temperatures (> 20°C), PM_2.5 concentrations were indeed higher at lower soil moisture values as well.

Another study published in 2018 (Doede and Davis, 2018) and conducted in California, found similar results to the ones in this paper, showing a correlation between PM₁₀ concentrations and drought effects. In this particular study, this was partly due to the drying out of lakes, where the exposed lakebed, which is normally submerged under water, can be a source of PM₁₀ particles.

Charusombat and Niyogi (2011) focused on the topic of drought vs air pollution using satellite observations, looking particularly at the variability between aerosol optical depth (AOD) retrieved from the Moderate Resolution Imaging Spectroradiometer (MODIS) and in situ particular matter (PM_2.5 and PM₁₀) concentrations. Level of drought was represented by the Standardized Precipitation Index (SPI). The results clearly indicated higher AOD values under drought conditions during summer periods. This study also looked at the differences between urban and rural sites. It was found that in agricultural areas the correlation between PM₁₀ concentrations and drought is higher than in urban areas during drought periods, which is in accordance with the results of this study, where the most profound difference was observed at rather remote, agricultural sites than in urban locations.

Javadian et al. (2019) looked at the extreme case of dust storms, which are common in semi-arid and arid regions. Such storms have an extreme effect on particulate matter concentrations and a correlation between drought events and extreme dust events was therefore found. Such storms in such an extent do not occur in the Czech Republic, but the results of the paper clearly demonstrate the possible effects of wind and resuspension that is more profound during drought times.

Exploration of the link between droughts and atmospheric aerosols was also studied by Charusombat and Niyogi (2011) using data from Indiana, U.S. Authors of this study also divided the stations based on their type (urban, suburban, rural) and a potentially more significant link during drought conditions was found in case of the rural stations, which is in accordance with the results found in this study based on data from the Czech Republic.

Overall, the results found in this study are in concordance with other papers that concentrated on this topic. This is despite the fact these studies were performed in other parts of the world, where the effect of local conditions such as vegetation type, climate type, could play a role.

The original hypothesis that assumed a potential negative effect of drought on suspended particle concentrations during the vegetation season was confirmed.

Further analyses of the effects of drought on air pollution could for example focus on the effects of drought on ground-level ozone pollution. This phenomenon was studied for example by Demetillo et al. (2019), indicating that drought conditions potentially alter multiple terms in the ozone mass balance equation.

Another possible extension of the study in this field would be to perform special-purpose measurements of air pollution and meteorological conditions at sites, where a significant effect of drought could be expected (agricultural, open areas). The presented study only uses data from long-term stationary ambient air quality monitoring stations within the Czech ambient air quality monitoring stations network.

 
5 CONCLUSION

The presented study focuses on the comparison between PM₁₀ and PM_2.5 concentrations during dry and wet periods at different types of stations (rural, suburban, urban) across the Czech Republic. Data were analysed from 2010 to 2019 for vegetation season (April–September) to avoid the major influence of emissions from domestic heating during the heating season. The parameter R_d/w (ratio of PM₁₀ concentrations between dry and wet period) varies from 126.7 to 146.7%, which confirms higher PM₁₀ concentrations during dry periods. Dunn test proved there is a statistically significant difference between the PM₁₀ concentrations at each station type for dry and wet period.

The effect of vegetation cover could be observed by the lowest R_d/w in July (full vegetation cover) and highest R_d/w in August. August is a month when both harvest and temperatures peak, so the resuspension of soil particles is most intense. Data from the individual stations in August proved a statistically significant difference between concentrations in dry and wet periods. The effect of consecutive dry or wet days on increasing/decreasing PM₁₀ concentrations was also apparent. PM₁₀ concentrations gradually increased after 5^th day of consecutive drought at rural and urban stations. Significant decrease in PM₁₀ concentrations was observed especially at the beginning of a wet period (0–2 day). The effect of cumulative wet days on concentration decrease was pronounced most at rural stations.

The PM_2.5/PM₁₀ concentration ratio in both dry and wet periods was very similar, the difference being less than 1%. In the individual months PM_2.5/PM₁₀ ratio varies between different station types. While at the beginning of a vegetation season, a smaller ratio indicates a prevailing amount of PM₁₀ particles at all station types, in the following months the situation was different. A higher PM_2.5/PM₁₀ ratio (indicating higher ratio of PM_2.5 particles) was observed during the dry period from June at rural, and July at suburban and urban stations. This fact is probably caused by secondary particles formation and other anthropogenic activities (e.g., increased traffic, barbecue) that can contribute to fine particles in the summer months, thus changing the PM_2.5/PM₁₀ ratio.

A significant effect of soil condition on PM_x concentrations was proved by analysing soil moisture (at a depth of 0–10 cm) and soil temperature at a depth of 5 cm underground. Soil moisture itself is a parameter that well represents the actual dryness of the soil. The effect of dried out soil was seen in a correlation between soil moisture and resuspension intensity or soil erosion. Soil with moisture above 36% suppressed resuspension even if soil temperature was higher than 20°C.