The homogeneity test of the annual total ozone column (OCT) and the seasonal time series was performed over the state of Alagoas, using the Bartlett test (1937) with 95% significance. The results indicated that the average annual and seasonal TCO values are homogeneous for all the series analyzed. Bartlett's test is a statistical procedure that checks the homogeneity of variances between various groups or samples, testing the null hypothesis that all variances are equal against the alternative hypothesis that at least one of them is different. This test is particularly relevant in analyses such as ANOVA, where the assumption of equal variances (homoscedasticity) is essential for the validity of the results. However, it is important to note that the Bartlett test is sensitive to the normality of the data; If the data does not follow a normal distribution, the test may erroneously indicate that the variances are different.
Table 1 shows the position and statistical dispersion of the concentrations of the Total Ozone Column (TCO) in the municipalities analyzed. The annual average of TCO in the region is 263.24 ± 9.91 DU (Dobson Units), with a maximum value of 311 DU observed in the cities of Maceió and São Luís do Quitunde, and a minimum value of 304 DU in Arapiraca. When analyzing the annual averages of OCT, it was considered essential to assess the degree of dispersion of these data. The standard deviation (σ), which measures the uniformity of a data set (Wilks, 2006), was 9.91 DU on average, with a maximum variation of 9.94 DU and a minimum of 9.87 DU between cities. The average coefficient of variation was 3.77%, with a maximum of 3.78% and a minimum of 3.75%, indicating a low variation and suggesting that the regions studied have homogeneous characteristics.
This homogeneity in the distribution of TCO can be attributed to region-specific atmospheric factors, such as the Brewer-Dobson Circulation (CBD), which, according to Hauchecorne et al. (2002), Mohanakumar (2008) and Solomon et al. (2016), does not suffer significant interference from the Antarctic Polar Vortex (VPA), whose intensity is weak in this area. This atmospheric dynamic facilitates the uniform distribution of observed ozone concentrations.
From an environmental and public health perspective, variations in TCO concentration can have significant impacts. The concentration of ozone in the stratosphere is crucial for protecting life on Earth from harmful ultraviolet (UV) rays. Therefore, the homogeneity in the distribution of ozone suggests a relatively uniform protection against UV radiation in Alagoas, which is positive for public health. On the other hand, any significant variation in TCO may indicate changes in atmospheric conditions that could lead to adverse consequences, such as increased risk of skin cancer and other health problems related to UV radiation exposure.
Additionally, TCO analysis can provide important insights into climate change and atmospheric circulation patterns, which are key to understanding large-scale environmental changes. Therefore, ongoing studies on TCO and its variations are essential to monitor environmental impacts and protect public health.
Figure 2 illustrates the average monthly TCO in DU in the six cities from 2008 to 2016. Maceió and São Luiz do Quitunde recorded the highest TCO values at 263.66 ± 9.93 DU, with São Luiz do Quitunde reaching a maximum of 311 DU. On the other hand, Pão de Açúcar de Açúcar recorded the lowest TCO at 235 DU, with a variation of 76 DU. Compared to Souza et al. (2022a), ozone concentrations in Mato Grosso do Sul from 2005 to 2020 exhibited significant variability, ranging from 260 DU in the Pantanal to 347 DU in the Cerrado, with a variation of 101 DU.
The analysis highlights temporal and seasonal fluctuations in ozone concentrations, reaching the peak on 09/13/2010 (winter) for the cities of São L do Quintude and Maceió (311 DU); 09/13/2010 (winter) for the cities of Arapiraca (304 DU), Coruripe (319) and Palmeira dos Índios (304 DU) and Pao de Açucar on 10/22/2010 (winter) with (310 DU). In the seasons of the year, the highest average TCO occurs in the spring season with an average of 271 DU. This seasonality underscores the profound influence of climatic and seasonal factors on ozone dynamics, emphasizing its critical role in interpreting atmospheric trends. These findings deepen our understanding of ozone distribution and the impact of seasonal and climate variations, for effective environmental management and strategies to mitigate potential impacts on regional air quality. The homogeneous distribution of OCT according to Hauchecorne et al. (2002), Mohanakumar (2008) and Solomon et al. (2016), can be attributed to the Brewer-Dobson Circulation (CBD), which does not suffer significant interference from the Antarctic Polar Vortex (VPA), which, in this region, has a weak intensity. This atmospheric dynamic contributes to the uniform distribution of observed ozone concentrations.
In southern Brazil, Crespo et al. (2011) analyzed the average behavior of the total ozone column and observed a decline in minimum values from 1979 to 1996. Peres et al 2024, using total ozone column (TOC) data, studied the mean reduction of 7.0 ± 2.9 DU in 62 events over southern Brazil between 2005 and 2014, using total ozone column (TOC) data. These events were more frequent in October, the positive phase of the ENSO (El Niño-Southern Oscillation) index had a significant influence on most events, while the Quasi-Biennial Oscillation (QBO) showed a balance in its influence. These findings underscore the importance of understanding the interplay between global and regional phenomena on ozone variability
In Alagoas, from 2008 to 2017, monthly averages showed a drop in minimum ozone values in May, at a rate of 5% per year (see Supplementary Material 1.1). The time series of the Total Ozone Column (TCO) analyzed by the Mann-Kendall (MK) method showed slightly similar trends, with a slight increase in tropospheric ozone, ranging from 1.75 DU for Cururipe to 2.84 DU for São Luis do Quitunde. However, trend analysis by the Mann-Kendall test revealed a slight decrease in stratospheric ozone for the summer, autumn, and spring seasons, but an increase in the winter ozone column (OCT) (MS 1.1, Table 2). The relatively low increasing trend of ozone in the ozone column (TCO) may be related to an increase in emissions of ozone precursors. It is observed that, for the annual values, the Mann-Kendall (MK) test indicates significant trends at the level of 5% of significance in the cities of Arapiraca, Palmeira dos Índios, São Luís do Quitunde and Maceió. Regarding the summer and spring seasons, the MK test did not identify any significant trend. For the autumn season, significant trends were identified in the cities of Arapiraca, Coruripe, Pão de Açúcar and Palmeira dos Índios. In winter, a significant trend was observed in the cities of Arapiraca, Palmeira dos Índios, São Luís do Quitunde and Maceió (Pohlert, T. (2016)).
As reported by Clain et al. (2009), Thompson et al. (2014), and Souza et al. (2022), ozone enhancement (OCT) may be associated with anthropogenic activities such as urbanization, transportation, industrialization, and biomass burning, as well as long-range transport of pollutants. These results agree with the findings of Kim and Newchurch (1998), who investigated the influences of biomass burning on the ozone column (TCO), noting that the annual maximum of ozone during the spring (September to November) is related to increased biomass burning activity.
The decrease in the Total Ozone Column (TOC) in the region can be explained by several factors that interact in a complex way in the atmosphere: The Brewer-Dobson Circulation (CBD) is an atmospheric process that transports ozone from the tropical region to the poles. Changes in the intensity and structure of this circulation can influence the amount of ozone that remains in tropical regions, including the region. During certain weather phenomena, such as El Niño, this circulation can be altered, reducing the amount of ozone transported to tropical areas. El Niño and OQB are phenomena that affect the distribution of ozone in the stratosphere and troposphere. During El Niño events, for example, changes in atmospheric circulation can lead to a temporary reduction in ozone concentrations in certain regions. These variations are caused by changes in wind patterns and the temperature of the stratosphere. The seasonal cycle of ozone can vary, with lower concentrations observed at certain times of the year due to natural changes in atmospheric dynamics. In tropical regions, the variability may be more pronounced, with influences from semiannual and semiannual modes of variability. Although the Montreal Protocol has significantly reduced the emission of ozone-depleting substances (such as CFCs), there are still residual effects that can temporarily impact CTO. In addition, local and regional pollutants can chemically interact with ozone, contributing to its depletion on a local scale.
These factors, when combined, may result in a temporary reduction of the Total Ozone Column in the study area, thereby increasing the risk of exposure to UV radiation.
These findings underscore the dynamic nature of ozone levels over time and emphasize the importance of ongoing monitoring and understanding for effective environmental management and policymaking.
The evaluation of the behavior of interannual and seasonal variability of the CTO for the entire region is shown in Fig. 2. The interannual variability for the period 2008-2016 is represented in Fig. 2. It is possible to observe that the CTO values in the analyzed period are typically concentrated between 304 DU and 311 DU, reinforced by Sousa et al. (2020) with some atypical years 2010 where the highest values occur (Fig. 2), in line with studies by Lima (2018) with a study period from 2005 to 2015 for the same region, similar to that observed by Lopo et al. (2013) in the period 2001 to 2009 and Sousa et al. (2020) in the period from 1978 to 2013, Lima et al 2020 in the period from 1997 to 2018.
The variation in OCT is not as pronounced as seen in Sahai et al. 2000, Lopo et al. (2013) and Sousa et al. (2020), but it presents a smooth curve with a variation of approximately 24.0 DU, its minimum value is in May (250 DU) and the maximum value in October (274 DU). In the first quarter of the year, January-February-March, the CTO values remain almost stationary between 262 DU and 263 DU, in the following months it decreases until reaching the lowest value, in May, with 250 DU. After this period, there is a constant increase until October, where it reaches the maximum annual value with 274 DU. November and December show a gradual decrease in values, 271 DU and 267 DU, respectively (Fig 5 and 6). This annual cycle is observed over higher latitudes, however with greater variations as shown in Peres et al. (2017) and Dias Nunes (2017; 2020).
Normal, Lognormal, Logistic, and Weibull probability density functions (PDFs) were applied to model the TCO time series over a 9-year period (2008-2016). These distributions are widely used in environmental studies to analyze the frequency of variables such as ozone concentrations and temperatures. The probability density [f(x)] functions and cumulative distribution functions [F(x)] of these PDFs were employed to fit the TCO data (Table 2, see Supplementary Material 1.2), as described by Sousa (2020a) and Reis et al. (2022). To identify the most appropriate distribution function to describe OCT concentrations, we performed the Kolmogorov-Smirnov (KS) statistical test (Table 3, see Supplementary Material 1.2)), which evaluates the quality of the fit between the observed data and the theoretical distributions, providing information on the adequacy of each distribution over the study period.
The analysis of the asymmetry of the empirical distributions in each city is also presented in Figures 3 (MS1.2). These figures show the distribution of the OCT data over the study period, including the estimated densities of the distributions (DM), allowing a preliminary verification of the proximity between the estimated densities and the empirical distribution of the data.
Tables 2 and 3 (SM1.2) present the results of the calculations of the shape and scale parameters, as well as the results of the Kolmogorov-Smirnov (KS) test, used to select the most appropriate model for each city studied. According to the results of the KS test, it was observed that the Normal distribution was the one that came closest to the OCT data in the period evaluated, with a p-value higher than 0.05, indicating a good fit.
In the study by Souza et al. (2020b), ozone concentrations in Campo Grande, Mato Grosso do Sul, Brazil, were evaluated for the year 2016. 15 PDFs were used to identify the best-fit distribution in different seasonal periods: the whole year, spring (September to December), summer (December to March, characterized by high solar radiation), autumn (March to June), and winter (June to September, characterized by low solar radiation). The study focused on the analysis of seasonal variations in the statistical behavior of PDFs.
The performance of these distributions was evaluated by means of three quality tests: Kolmogorov-Smirnov (KS), Anderson-Darling (AD) and Chi-Square. The comparative analysis of the results indicated that the generalized distribution of extreme values provided a good overall adjustment throughout the year. However, specific stations showed variations in the best-fit distributions. For winter, the 3-parameter Gamma distribution was the most suitable, while the 3-parameter Lognormal distribution adjusted better in the spring, and the Weibull distribution was optimal for the summer. Autumn also showed a good fit with the Gamma distribution of 3 parameters. Interestingly, winter and autumn, characterized by lower O3 concentrations, kurtosis, and asymmetry, coincided in the distribution adjustment. On the other hand, summer and spring, marked by higher O3 concentrations and different kurtosis and asymmetry values, required distinct PDFs (Figure 3, see Supplementary Material 1.2).
A noticeable negative slope from 2008 to 2009 suggests a temporary decrease in TCO during this period, followed by a gradual increase in subsequent years. It is important to note that this trend in the Alagoas region may not reflect global or regional trends. Further investigation into the causes of these TCO fluctuations, such as changes in atmospheric circulation patterns, anthropogenic emissions, or natural phenomena, would provide valuable insights into their underlying mechanisms and implications for ozone layer protection. The interannual variability of TCO is predominantly influenced by annual variations in local weather, solar activity, teleconnection patterns, and other weather modes. Solar radiation plays a critical role in modifying ozone concentrations through photochemical reactions in the upper atmosphere. In addition, the absorption of solar radiation by ozone in the stratosphere influences atmospheric thermal dynamics, thus altering atmospheric circulation and ozone distribution (Zhou et al., 2006). These interannual variations also coincide with ENSO (El Niño-Southern Oscillation) events, in which TCO tends to increase during El Niño years: 2009-2010 (+4 DU or 1.54%, 2014-2016 (3 DU or 1.13%); La Niña: 2008-2009 (-5DU or (1.89%), 2010-2011 (zero), 2011-2012 (-1 or 0.38%), 2016-2017 (-2 DU or 0.76%).
These events are cyclical and vary in intensity, with some episodes being stronger or more prolonged than others.
Several studies, including Zhou et al. (2013), Li et al. (2020), and Zou et al. (2020), have identified complex temporal and spatial variations in TCO. Zhou et al. (2013) reported significant negative trends, particularly in January, indicating a potential decline in ozone during this period over decades. Li et al. (2020) and Zou et al. (2020) provided additional insights, highlighting negative and positive trends in different seasons and latitudes, reflecting different analysis methodologies, data sources, and regional conditions.
In addition, Kuttippurath et al. (2023) introduced further complexity by examining the Hindu Kush Himalayan and Tien Shan regions, revealing significant seasonal variations with predominantly negative trends in summer and autumn, contrasting with positive trends in winter. These findings highlight the seasonal variability in ozone behavior influenced by local climatic conditions and interactions with other factors. In conclusion, these studies underscore the complexity of TCO trends and emphasize the importance of integrated, long-term monitoring and research to fully understand regional atmospheric and climate changes.
Figure 4 shows the monthly change in TCO averages for the region. The highest average monthly TCO of 275 DU was observed for the month of September in the city of Maceió, when the highest average ozone averages occur, and the lowest average monthly TCO of 250 DU was in May in the city of Coruripe. The analysis of the data presented and the comparison with the studies by Peres et al. (2017) and Sousa et al. (2020) indicate that the annual cycle dominates the seasonal variability of TCO. The annual cycle is associated with seasonal changes in solar radiation and atmospheric temperature, which affect the production and destruction of ozone. The observations of the present study, with higher concentrations of OCT in September and lower in May, corroborate the predominance of the annual cycle over the semiannual one, reinforcing the idea that seasonal variability is mainly driven by annual factors related to the seasons.
Figure 5 shows the seasonal variation of the interannual averages of TCO for the region. The highest seasonal average TCO of 283 DU was observed during the spring months (OND - October, November and December) at the Coruripe station, coinciding with the peak ozone averages. In contrast, the lowest average TCO of 249 DU was observed during the fall months (AMJ – April, May, and June) in the same season. The analysis of the monthly averages of TCO highlights the importance of evaluating the dispersion of the data. Standard deviation (SD) is a measure of data uniformity (Wilks, 2006) and helps to understand variability within the data set.
Figure 5 shows the seasonal variation of the interannual averages of the Total Ozone Column (TCO) for the region, showing a seasonal pattern characteristic of mid-latitudes. The highest TCO values occur during the spring, while the lowest values are recorded in the fall (Peres et al., 2017; Toihir et al., 2018). The months of April to June and August to September exhibit the lowest indices of TCO variability, indicating minimal variation and greater homogeneity (Peres et al., 2017; Toihir et al., 2018). In contrast, January, February, July, October, November, and December are transition months, with greater variability, with typical oscillations around ±7.6 DU (Xie et al., 2014a; 2014b; Lima et al., 2020; 2021).
This seasonal cycle also reveals that, after spring, winter and summer continue with intermediate levels of TCO, while autumn consistently maintains the lowest values (Xie et al., 2014a; 2014b; Lima et al., 2020; 2021). The regularity of this pattern, with a single annual peak and minimum, is a reflection of the patterns observed at mid- and high-latitudes (Xie et al., 2014a; 2014b; Lima et al., 2020; 2021). It is important to highlight that the equatorial regions exhibit a considerably lower TCO amplitude (~24 DU) compared to the variability of 76 DU observed in the region studied (Xie et al., 2014a; 2014b; Lima et al., 2020; 2021).
The seasonal variability of ozone in this region is influenced by factors such as atmospheric circulation, local weather conditions, and human activities (Souza et al., 2022a; 2022b). The observations of the maximum TCO values in September/October and the minimum values in May are in line with the typical seasonal behavior observed in mid-latitude regions (Souza et al., 2022a; 2022b). Differences between ozone production rates and concentrations at latitudes are attributed to large-scale atmospheric circulation patterns that redistribute air masses vertically and horizontally (Souza et al., 2022a; 2022b). Understanding these seasonal variations is crucial for assessing local air quality and implementing effective environmental policies aimed at reducing pollutant emissions and promoting the sustainable development goals (SDGs) related to health, clean energy, sustainable cities, and climate action (UN, 2015). Continuous monitoring and research are essential to further elucidate the impacts of human activities and climate change on atmospheric composition and air quality in Brazil and worldwide.
The spatial distribution of the seasonal monthly average of the Total Ozone Column (TOC) during the 9-year study period (2008-2016) of the OMI sensor on the Aura satellite is represented in Figures 6a, 6b, and 7. The largest spatial distribution is observed during spring (OND), while autumn (AMJ) exhibits the smallest, consistent with the temporal trends illustrated in Figure 5.
The average annual TCO value evaluated from 2008 to 2016 was 263.50 ± 1.51 DU, reaching a maximum of 264.04 ± 1.75 DU and a minimum of 268.89 ± 1.46 DU (Table 1). Specifically, the highest monthly average occurred in 2010 in the municipality of Arapiraca, measuring 266.07 DU, while the lowest was recorded in 2015 in the municipality of Palmeira dos Índios with 261.80 DU (Figure 6a). Over the years, the monthly average fluctuated around 263.50 ± 0.40 DU, as illustrated in Figure 6b, with a maximum value of 274.56 ± 0.45 DU and a minimum of 250.09 ± 0.35 DU. It is noteworthy that October had the highest monthly average of 275.08 DU in Coruripe, while May had the lowest value of 249.66 DU in Palmeira dos Índios (Figures 6a, 6b and 7). The lower value observed in May can be attributed to its proximity to the winter solstice in the southern hemisphere, while the higher values in October/November and December correlate with the zenith position of the Sun and the reduction of cloud cover in the study region. A strong similarity was identified in interannual OCT behaviors among the six study regions.
Seasonal analysis of the O3 data reveals distinct variations in concentration and variability at different altitudes and seasons (Figures 5 and 7). The results indicate TCO peaks during the spring (OND - October, November, and December) and lows during the fall (AMJ - April, May, and June). This cyclical behavior is closely related to the Earth's position in its orbit around the Sun, where solar radiation influences the production of ozone in the stratosphere, resulting in two maximums and two minimums throughout the year. (Figure 5).