Atmospheric Factors Affecting a Decrease in the Night-Time Concentrations of Tropospheric Ozone in a Low-Polluted Urban Area

Ozone (O3) decomposition in the troposphere is a very important process which prevents excessive O3 accumulation in the air. It is particularly significant on warm summer days which are marked by a high risk of photochemical smog. We used Spearman’s rank correlation test to determine relationships between the drop in O3 concentrations over time (-ΔO3), nitrogen oxide (NO), nitrogen dioxide (NO2), and total nitrogen oxide (NOx) concentrations and meteorological factors (1-h average) in low-polluted urban area in Olsztyn (north-eastern Poland). Nitrogen oxide concentrations were measured continuously by the chemiluminescence method, and O3 concentrations were determined by the UV photometric method. The obtained results suggest that the rate of decomposition of tropospheric O3 is affected mostly by the presence of NOx, high temperature, and air humidity (positive correlation) as well as by wind speed (negative correlation). Maximum correlation coefficient values were reported between –ΔO3 and air temperature, –ΔO3 and absolute air humidity when NOx concentrations were low (below 1.0 microgram per cubic meter), reaching 0.271 and 0.243, respectively. These results indicate that O3 also reacted with air components other than NO and NO2. Precipitation at average temperature of < 0 °C did not significantly contribute to a drop in O3 concentrations at night-time. In the warm season, precipitation slowed down the rate of O3 decomposition, mostly because NOx were scrubbed by rain. An analysis of seasonal and daily –ΔO3 fluctuations revealed that –ΔO3 values were highest in the summer and shortly after sunset in the diurnal cycle.


Introduction
High ozone (O 3 ) concentrations in the tropospheric layer of the Earth's atmosphere pose a threat to the health of humans and land animals and have an adverse effect on plants (de Wit et al. 2015;Harmens et al. 2016). Episodes of high concentrations of O 3 , particulate matter (PM10, PM2.5), and nitrogen dioxide (NO 2 ) are major current problems related to air pollution. However, it should be stressed that PM10, PM2.5, and NO 2 are found mostly in the air in urban and industrial regions, whereas O 3 is present not only in urban but also in suburban and rural areas. The maximum daily 8-h mean concentration of O 3 should not exceed 120 μg m −3 (long-term objective) or 120 μg m −3 on more than 25 days per calendar year (target value) (EU 2008). In Europe, 41% of all stations showed O 3 concentrations above the target value for the protection of human health in 2015, which is considerably more stations than over the previous 5 years. In addition, only 13% of air quality monitoring stations fulfilled the longterm objective (no exceedance of the threshold level). Eighty-eight percent of the stations with values above the long-term objective were background stations (EEA 2017). Tropospheric O 3 is also a greenhouse gas which contributes to global warming (IPCC 2007). In addition to its adverse effect on human health and the natural environment, tropospheric O 3 also contributes to significant economic losses by damaging crops, lowering crop yield and causing damage to certain materials, in particular rubber. The annual losses resulting from the above have been estimated at around €6.7 billion in 47 countries in Europe (Holland et al. 2006).
Large quantities of O 3 are formed in the stratosphere. A part of this naturally formed stratospheric O 3 goes down to troposphere. However, photochemical processes involving nitrogen oxides (NO x ), volatile organic compounds (VOCs), and carbon monoxide (CO) are a more important source of O 3 in ground-level ambient air (Sharma et al. 2017). Photochemical smog is the most dangerous phenomenon related to the formation of O 3 as a result of the above chemical reactions. In addition to O 3 , a smoggy atmosphere comprises other characteristic components such as organic oxidizers (including peroxyacetyl nitrate, PAN), aldehydes, NO x , hydrogen peroxide, and free radicals (Kanaya et al. 2007). Photochemical smog is one of the most adverse contributors to the quality of urban air (EEA 2017). Its formation is supported by the following meteorological conditions: high insolation, high temperature, temperature inversion and dry, stagnant air masses (Manahan 2007;Hosoi et al. 2011). This environment triggers intense reactions with sunlight irradiation which produce O 3 and other secondary air pollutants. The most important phase of those changes is the NO-NO 2 -O 3 inter-conversion cycle which takes place in line with the below equations (Crutzen et al. 1999;Warmiński and Bęś 2009): where M molecules which absorb excess energy and determine the stability of O 3 molecules (usually N 2 or O 2 molecules) hν photon energy (with electromagnetic wave length of λ = 290-430 nm) R1-R3 reactions establish a photostationary state equilibrium which is described by the following equation (Leighton 1961): where J NO 2 nitrogen dioxide (NO 2 ) photolysis rate constant (rate constant of R1 reaction) k 3 nitrogen oxide (NO) oxidation rate constant (rate constant of R3 reaction) The above equation demonstrates that all other reactions which reduce NO concentrations in favor of NO 2 increase O 3 concentrations. Those reactions are inclusive of photochemical oxidation of CO, methane (CH 4 ), and non-methane VOC (Warmiński and Rogalski 2007;Kalbarczyk and Kalbarczyk 2017), namely, Atmospheric O 3 is a relatively unstable gas. It participates in a variety of redox reactions in the atmosphere. Atmospheric O 3 reacts with both organic and inorganic compounds (mainly with NO, SO 2 , CO, H 2 O) which are in plentiful supply in ground-level air (Crutzen et al. 1999;Hosoi et al. 2011). For this reason, cyclic, diurnal changes are observed in tropospheric O 3 levels where maximum concentration values are reported in the afternoon and minimum concentrations are noted in the hours directly preceding sunrise (Mazzeo et al. 2005;Kanaya et al. 2007;Warmiński and Rogalski 2007;Kalbarczyk et al. 2016). It has been suggested that the most important night-time reaction takes place between O 3 and NO. As demonstrated by Nicholson et al. (2001), the drop in O 3 concentrations in urban air which is characterized by high NO levels is more pronounced in the summer than in the winter. According to our previous research, the NO oxidation rate constant (k 3 ) in the city of Olsztyn (north-eastern Poland) ranged from 22.5 to 27.7 ppm −1 min −1 in July and from 14.1 to 15.5 ppm −1 min −1 in January (Warmiński and Rogalski 2007). According to Mazzeo et al. (2005), the k 3 constant in Buenos Aires was 21.5 to 23.0 ppm −1 min −1 in wintertime. The reported differences are due to various thermal conditions because k 3 is a function of air temperature as demonstrated by the equation (Seinfeld and Pandis 2016): where T-air temperature (K). Other redox reactions involving O 3 may be predominant at low NO x concentrations, subject to the locally occurring gaseous pollutants, aerosols, and water vapor. Therefore, we hypothesized that the rate of decrease in tropospheric O 3 is also influenced by factors other than nitrogen oxide concentrations and temperature. The main motivation for the present study was to assess the sensitivity of night-time ozone concentrations to meteorological parameters at different NO x levels. The results will be useful for modeling tropospheric ozone concentrations. According to Wałaszek et al. (2018), modeling errors are predominantly systematic and result from chemical mechanisms, meteorological fields, and initial and boundary conditions, including background ozone levels. Reliable estimation of initial ozone concentrations is crucial for reducing ozone modeling errors. Ozone decomposition in the lower atmospheric layer is a very important phenomenon which prevents excessive accumulation of this gas in ambient air, in particular on warm summer days.

Location of the Monitoring Site
The study was conducted at the air quality monitoring station of the University of Warmia and Mazury in Olsztyn. The station's location and the mode of collecting representative air samples are consistent with the requirements set forth in the Regulation of the Polish Minister of Environment of September 3, 2012 and Directive 2008/50/ EC of the European Parliament and of the Council of May 21, 2008(EU 2008. The city of Olsztyn is situated in north-eastern Poland, in a region which is popularly known as the Green Lungs of Poland (Fig. 1). The region is characterized by a low degree of industrialization and by exceptional environmental values. The main sources of air pollution in Olsztyn are road transport and household gas emissions. The air quality monitoring station is situated in the south-western part of Olsztyn, on Lake Kortowskie. The monitoring site has the following geographic coordinates: 53°45′ 34.3″ N, 20°27′ 09.5″ E. The nearest sources of pollution are Warszawska street which is a part of national road No. 16 (east-bound, at a distance of approx. 500 m), residential estates in Słoneczny Stok composed of single-family houses equipped with coalfired boilers (1000 m), Dajtki (1400 m), Brzeziny (1600 m), and the city center (2200 m).
The results of daily measurements were compared with data from selected air quality monitoring stations of the Chief Inspectorate for Environmental Protection (GIOŚ) to determine spatiotemporal trends in tropospheric O 3 and NO x levels. The selected monitoring stations are situated in the Region of Warmia and Mazury (northeastern Poland), the Warsaw Metropolitan Area (capital of Poland, central Poland), and the Silesian Metropolitan Area (highly industrialized region in southern Poland). The stations in north-eastern Poland are situated in the cities of Olsztyn, Elbląg, Ostróda, Mrągowo, and Gołdap. Data were also obtained from a rural background monitoring station in the Borecka Forest (PL0005R Diabla Gora) which is a part of the European Monitoring and Evaluation Program (EMEP). In the Warsaw Metropolitan Area, O 3 concentrations were presented as the average value from five monitoring stations in Warsaw, and the results from the Silesian Metropolitan Area were averaged based on the data from the cities of Katowice, Zabrze, Tychy, and Dąbrowa Górnicza. The concentrations of NO x were measured in nine monitoring stations in each region. Detailed information about monitoring stations and measurements is provided by GIOŚ at http://powietrze.gios.gov.pl.

Measurements and Instrumentation
NO and NO 2 concentrations were measured continuously by the chemiluminescence method, and O 3 levels were determined by the UV photometric method (API 200E and API 400E analyzers, Teledyne Advanced Pollution Instrumentation, Inc. USA). The applied methods are consistent with the ISO 7996:1985and ISO 13964:1998standards (ISO 1985ISO 1998), and the analyzers have US EPA certification (USEPA RFNA-1194-099 and USEPA EQOA-0992-087). All analyzers met the requirements of minimum data capture (DC), stipulated by Directive 2008/50/EC (EU 2008). Minimum DC for O 3 and NO x is 90% in summer and 75% in winter. In our study, conducted in 2006, minimum DC was 98.6 and 76.1% for O 3 and 98.5 and 91.1% for NO x in summer and winter, respectively. Air temperature and humidity were measured with the DMA580 digital thermo-hygrometer, wind direction and speed measurements were performed using the DNA502 gonio-anemometer and the DNA511 tacho-anemometer, and precipitation was measured with the DQA031 automated rain gauge (LSI s.p.a., Italy). Instantaneous data (5-s) were registered, validated, and averaged to 1 h in a datalogger with the CS5 v. 5.3 system (CSMS, Poland). Temporal trends in ground-level O 3 and NO x concentrations in ambient air in low-polluted urban area of the city of Olsztyn are presented in Fig. 2. The concentrations of O 3 , NO 2 , and NO and meteorological parameters were measured throughout 2006, but this study accounts only for data collected at night-time, when photochemical processes cease and when only reactions leading to O 3 decomposition take place.

Data Analysis
We determined the drop in O 3 concentrations (-ΔO 3 ) in hour C with the use of the equation: where x AB , x BC -average O 3 concentration (μg m −3 ) based on instantaneous measurements between hour A to hour B and hour B to hour C; t C , t B -time (h) at hour C and B from the beginning of the measurement series. Validated NO, NO 2 , NO x , and O 3 concentrations and meteorological parameters (ambient air temperature, relative and absolute air humidity, wind speed and direction, total atmospheric precipitation) were subjected to a preliminary statistical analysis to verify the hypothesis on the goodness of fit of value distribution to normal distributions. Kolmogorov-Smirnov and chisquare tests invalidated the above hypothesis. Therefore Spearman's rank correlation test was employed to determine relationships between -ΔO 3 and the investigated factors.
Calculations were performed in several time series: (i) throughout the year (all night-time hours); (ii) hours without atmospheric precipitation; (iii) cool season (period with average weekly temperature of ambient air at night-time < 0°C; in this study, this criterion was fulfilled over a period of 12 weeks, from 01 Jan. 2006 to 25 Mar. 2006); (iv) warm season (12 warmest weeks in the year, from week 25 to 36, i.e. from 11 Jun. 2006 to 09 Sept. 2006; the average night-time temperature of ambient air in the above period was higher than 13.9°C); (v) only hours during which specific NO x concentrations were recorded (≤ 1.0, ≤ 2.0, ≤ 3.0, and > 3.0 μg m −3 ).
Coefficients of correlation were calculated, and seasonal and diurnal variations in -ΔO 3 were analyzed. The average weekly value of -ΔO 3 was adopted as the year as the independent variable. Diurnal variation was analyzed separately for the cool and warm season, and the time of the day (independent variable, x) was transformed in line with the following rules: (i) from sunset to midnight, the values of x for 4 p.m., 5 p.m., 6 p.m., etc. were set at 4, 5, 6, etc.; (ii) at midnight, the value of × was set at 12; (iii) from midnight to sunrise, the values of x for 1 a.m., 2 a.m., 3 a.m., etc. were set at 13, 14, 15, etc. The above transformation was performed to ensure the continuity of x value (hours) from sunset to sunrise. The obtained data were verified statistically in STATISTICA 10 software (StatSoft, Inc.).

Spatiotemporal Trends in Tropospheric Ozone and Nitrogen Oxide Concentrations in Poland
The average concentration of tropospheric O 3 in all air quality monitoring stations in the Region of Warmia and Mazury (north-east Poland) was lowest in 2016 and highest in 2006 at 48.7 and 56.9 μg m −3 , respectively (Fig. 3a). A clear decreasing trend in the above parameter was observed in north-east Poland. In our study, the concentration of tropospheric ozone on Lake Kortowskie in Olsztyn decreased from 61. Olsztyn than on Lake Kortowskie. The above was related to differences in the concentration of NO x which was around 8 μg m −3 higher in central Olsztyn (Fig. 3b). O 3 and NO x concentrations are measured simultaneously in many locations in Poland and around the world. A negative correlation is observed between ozone and NO x levels due to the reaction between O 3 and NO (Nicholson et al. 2001;Kalbarczyk et al. 2016). For this reason, ozone levels in industrialized urban areas characterized by higher NO x concentrations, such as Warsaw and the Silesian Metropolitan Area, are approximately 10-11 μg m −3 lower than in north-eastern Poland. However, a rising trend in the concentrations of both ozone and nitrogen  The key factors which determined the rate of drop in O 3 concentrations at night-time were wind speed, total atmospheric precipitation, ambient air temperature, absolute humidity of ambient air, NO 2 , and NO x concentrations (Table 1). Except for wind speed and total atmospheric precipitation, the remaining factors were positively correlated with -ΔO 3 . If data from hours with atmospheric precipitation are omitted in the analysis, the influence of chemical factors on the determination of the dependent variable (-ΔO 3 ) decreases, while the impact of meteorological factors increases. The obtained results supported our assumptions that temperature significantly contributes to a decrease in O 3 concentrations. However, no statistically significant correlation was found between the rate of O 3 concentration drop and input O 3 and NO concentrations. It could probably be explained by the multifactorial nature of the phenomenon. Moreover, NO concentrations determined at air quality monitoring stations are not input concentrations but the concentrations of NO which has not yet been oxidized to NO 2 . At measuring stations situated at least several hundred meters away from NO x sources, NO is largely oxidized to NO 2 before it reaches the monitoring station with the air masses (Nicholson et al. 2001;Warmiński and Rogalski 2006). Therefore, NO x concentrations are proportional to initial NO concentration levels. For this reason, no statistically significant dependencies were found between -ΔO 3 and NO levels, but only between -ΔO 3 and NO 2 and NO x concentrations ( Table 1). The formation of NO 3 radicals in dark hours is a well-described reaction which leads to a drop in O 3 concentrations (Zheng et al. 2017): NO 3 radicals are determined practically only at nighttime because they rapidly photolyze to NO 2 and O (Atkinson and Arey 2003, Stutz et al. 2010, Wang et al. 2013, Akimoto 2016. The highest correlation was found between wind speed and -ΔO 3 , and the determined dependency was negative. Strong winds cause intense mixing of masses of polluted and clean air; therefore, the higher the wind speed, the greater the drop in NO x levels ( Table 2). The negative correlation between NO x concentrations and wind speed has been documented in the works of, among others, Delaney and Dowding (1998), Tang et al. (2011), Karagiannidis et al. (2015), and Valotto and Varin (2016). For this reason, we determined lower NO x concentrations and -ΔO 3 values under strong wind conditions (Tables 1  and 2). A negative correlation between O 3 concentrations and wind speed was also reported by Banta et al. (2011) and Kalbarczyk et al. (2016).
A positive correlation between -ΔO 3 and absolute humidity could be due to the reaction between O 3 and water in the gaseous state. However, this mechanism of O 3 breakdown at night-time is doubtful due to a very low value of the correlation coefficient at 0.133 (Table 1). Other mechanisms could be involved in the decomposition of ground-level ozone, but the reaction with NO (R3) appears to be predominant.
Ozone and particles adsorbed on aerosol surface also react at a faster rate in high humidity air. Chughtai et al. (2003) proposed the following relationship between the drop in O 3 concentrations (−d [O 3 ]/dt) caused by its reaction with soot aerosol and absolute humidity of air: indicates that the higher the air humidity, the higher the rate of O 3 reaction with soot components. Sakamoto et al. (2004) also noted that the presence of water vapor in air speeds up the oxidation of SO 2 through O 3 on the surface of yellow sand particles. Buckley and Birks (1995) also pointed to the possibility of visible-light photolysis of O 3 •H 2 O cluster molecules (Van der Waals complexes) as well as reaction without the involvement of solar radiation: Reaction R7 is energetically possible for light with a wavelength under 665 nm, while reaction R8 is exothermic and it takes place even without light absorption, i.e., at night. Those reactions could also be responsible for higher -ΔO 3 values observed at night-time when air humidity is high.
Air pollutants, in particular the highly reactive O 3 , can also undergo dry deposition on the surface of soil, plants, buildings, etc. During the day, in vegetated areas, both stomatal fluxes and non-stomatal fluxes play an important role in the dry deposition of O 3 . At night, the dry deposition of O 3 is dominated by non-stomatal fluxes (Fares et al. 2010;El-Madany et al. 2017). The total dry deposition of O 3 is several-fold or 10-to 20fold lower at night than during the day (Padro 1996;El-Madany et al. 2017). According to the literature, stomatal flux is proportional to air humidity (Zapletal et al. 2012;Fares et al. 2013). The night-time concentrations of O 3 decrease steadily due to the absence of sunlight and redox reactions.
Total atmospheric precipitation was negatively correlated with -ΔO 3 (Table 1). Although the Spearman's rank correlation coefficient was statistically significant (p < 0.01), the precipitation factor in the analysis of annual data did not produce satisfactory results (R S = − 0.058). We propose a hypothesis that this is due to the types of precipitation at various times of the year. The analysis of annual data accounts for total precipitation Cool season-a period with average weekly temperature of ambient air at night-time of < 0°C; warm season-12 warmest weeks in the year AH absolute humidity, T temperature, WD wind direction, WS wind speed, P total precipitation *Spearman's correlation coefficients marked with an asterisk are statistically significant at p = 0.01 which is not broken into different types of precipitation, i.e., rain, snow, or hail. For this reason, we performed an additional analysis of correlation covering two seasons: (i) cool (marked by ambient air temperature below 0°C; coldest 12 weeks in 2006) and (ii) warm (12 warmest weeks in 2006). The correlation between -ΔO 3 and total atmospheric precipitation was higher in the warm period (R S = − 0.113, p < 0.001) in comparison with data covering the entire year (Table 3). A negative correlation rank coefficient may seem puzzling in this context. It indicates that O 3 concentrations decrease at a slower rate as precipitation intensity grows. An analysis of data showed that at times of precipitation over the warm season, -ΔO 3 reaches 1.43 μg m −3 h −1 on average, while it is threefold higher in hours without precipitation (Fig. 4). This is probably due to lower NO x concentrations at times of precipitation and, as indicated earlier, NO significantly contributes to a reduction in tropospheric O 3 levels. It should be noted that the value of R s was very low, although statistically significant. A stronger correlation between -ΔO 3 and NO 2 (NO x ) concentrations and wind speed was determined in the warm season than in the cool season (Table 3). However, statistically significant correlations between -ΔO 3 and air temperature and between -ΔO 3 and air humidity were not found in either of the analyzed periods.

Influence of NO x Concentrations on the Results of Correlation Analysis
Nitrogen oxide concentrations proved to be one of the most important factors determining the rate of the drop in O 3 concentrations at night-time. In view of the above, an additional Spearman's rank correlation test was performed for night-time data, and the obtained average NO x concentrations were in the range of ≤ 1.0, ≤ 2.0, ≤ 3.0, and > 3.0 μg m −3 . An analysis of correlation test results in Table 4 indicates that selected meteorological factors have a growing impact on -ΔO 3 values as NO x concentrations decrease from > 3.0 to ≤ 1.0 μg m −3 . It applies primarily to temperature and absolute humidity and, to a lesser extent, to wind speed. A reverse dependency was reported for NO 2 and NO x concentrations.  Tables 1, 3, and 4 present Spearman's rank correlation coefficients (R S ) and significance levels (p) determined not only between -ΔO 3 and chemical and meteorological factors, but also between -ΔO 3 and the hour of the day and the week of the year. In most cases, those coefficients were statistically significant. Seasonal variations in -ΔO 3 (weekly data) are presented in Fig. 5. Maximum -ΔO 3 values were noted in the summer, whereas minimum ΔO 3 values (near zero) were observed at the beginning and end of the year. The noted difference is due mostly to seasonal variations in air temperature.
In the cool period with temperatures of < 0°C, diurnal changes in -ΔO 3 values ranged from around − 1.0 to 4.0 μg m −3 h −1 (Fig. 6a). Higher fluctuations in -ΔO 3 values were observed in the warm period, from around 1.0 to more than 8.0 μg m −3 h −1 (Fig. 6b). The highest -ΔO 3 Table 4 Effect of NO x concentrations on the results of Spearman's correlation analysis between -ΔO 3 (dependent variable) and the investigated factors (independent variables) Independent variables Spearman's correlation coefficient (R S ) for night-time hours with NO x concentration within the following ranges *Spearman's correlation coefficients marked with an asterisk are statistically significant at p = 0.01 a Threshold concentration level above which significant correlations between -ΔO 3 and NO x concentrations were determined values were recorded between 4 p.m. and 9 p.m. in the cool season and between 9 p.m. and 10 p.m. in the warm season. The highest drop in O 3 concentrations is observed in the hours directly after sunset, when air temperature is still relatively high but photochemical processes have already ceased. In the following hours, -ΔO 3 values gradually decrease. OH and HO 2 radical concentrations also begin to decrease as of sunset (Arias and Hastie 1996;Cantrell et al. 1996;Commane et al. 2010).

Conclusions
Ozone decomposition reactions in the troposphere are an important stage of the O 3 cycle. The higher the rate of those reactions, the lower the risk of photochemical smog. The analysis of changes in concentration levels of O 3 and other air parameters in the lower parts of the troposphere at night-time supported the preliminary identification of factors which significantly contribute to the rate of those reactions in low-polluted urban area. We conclude that the rate of decomposition of tropospheric ozone (-ΔO 3 ) was affected mainly by the presence of NO x and wind speed. It should be noted, however, that NO x concentrations were largely determined by wind speed. High wind speed causes intense mixing of air masses, leading to a drop in the concentration levels of pollutants in urban areas, including NO x . Therefore, wind speed indirectly influenced -ΔO 3 . Other significant determinants included air temperature and absolute humidity which were positively correlated with -ΔO 3 . Maximum correlation coefficients for these parameters were observed at low NO x concentrations in the air (≤ 1.0 μg m −3 ), which reached 0.271 and 0.243, respectively. Both values of R S were similar to the value of R S for NO x . It should be noted that correlation coefficients were Fig. 6 Diurnal variation of -ΔO 3 and air temperature at night-time in the cool (a) and warm season (b). x-value is the transformed night-time hour (x = 0 for 12 PM and x = 12 for 12 AM) low, but statistically significant. The above could suggest that ozone also reacts with air components other than NO x . The observation that the decrease in ozone concentrations was not correlated with relative humidity, but was correlated with absolute humidity, in particular when NO x levels were very low, is a crucial finding. The effect of absolute humidity on -ΔO 3 decreased with a rise in NO x levels. Precipitation in the cool season with temperatures of < 0°C did not contribute to a drop in O 3 concentrations at night-time. In the warm season, rainfalls slowed down the rate at which O 3 concentrations decreased, compared with dry spells. This could result from effective removal of NO x , in particular the highly watersoluble NO 2 , by rain. O 3 is sparingly water-soluble, and its wet deposition takes longer relative to NO 2 . Seasonal and daily fluctuations in temperature were responsible for -ΔO 3 variations. The highest -ΔO 3 values were reported in the warmest period of the year and shortly after sunset in the daily cycle. The presence of high concentration levels of OH and HO 2 radicals directly after sunset could also be a cause of the observed variations in -ΔO 3 values at that time.
To conclude, it can be postulated that some factors which significantly affect O 3 formation during day-time (nitrogen oxides, high air temperature, low wind speed) also contribute to the rate of O 3 decomposition at nighttime. For this reason, daily O 3 concentration amplitudes are much higher in the summer than at other times of the year. We are also of the opinion that low air humidity during the summer continental high slows down the drop in O 3 concentrations at night-time. Day-time weather in the same period is marked by high temperatures and insolation, which supports O 3 formation in photochemical processes. Due to the above, O 3 concentrations may exceed target value for protection of human health after only several hot summer days and nights with low air humidity.
Our results can be useful for improving tropospheric ozone models. Such models account for ozone forming parameters as well as initial ozone levels (Wałaszek et al. 2018). As demonstrated by the study, selected meteorological parameters and NO x levels have a small, but statistically significant effect on initial ozone levels.