*Article* **Surface Ozone Pollution: Trends, Meteorological Influences, and Chemical Precursors in Portugal**

**Rafaela C. V. Silva 1,2 and José C. M. Pires 1,2,\***


**Abstract:** Surface ozone (O<sup>3</sup> ) is a secondary air pollutant, harmful to human health and vegetation. To provide a long-term study of O<sup>3</sup> concentrations in Portugal (study period: 2009–2019), a statistical analysis of ozone trends in rural stations (where the highest concentrations can be found) was first performed. Additionally, the effect of nitrogen oxides (NOx) and meteorological variables on O<sup>3</sup> concentrations were evaluated in different environments in northern Portugal. A decreasing trend of O<sup>3</sup> concentrations was observed in almost all monitoring stations. However, several exceedances to the standard values legislated for human health and vegetation protection were recorded. Daily and seasonal O<sup>3</sup> profiles showed high concentrations in the afternoon and summer (for all inland rural stations) or spring (for Portuguese islands). The high number of groups obtained from the cluster analysis showed the difference of ozone behaviour amongst the existent rural stations, highlighting the effectiveness of the current geographical distribution of monitoring stations. Stronger correlations between O<sup>3</sup> , NO, and NO<sup>2</sup> were detected at the urban site, indicating that the O<sup>3</sup> concentration was more NOx-sensitive in urban environments. Solar radiation showed a higher correlation with O<sup>3</sup> concentration regarding the meteorological influence. The wind and pollutants transport must also be considered in air quality studies. The presented results enable the definition of air quality policies to prevent and/or mitigate unfavourable outcomes from O<sup>3</sup> pollution.

**Keywords:** cluster analysis; meteorological influence; Multiple linear regression; NO<sup>x</sup> influence; Pearson's correlation; rural trends; surface ozone

### **1. Introduction**

With the constant increase of air pollutants emissions since the Industrial Revolution, an increase of ozone (O3) near the Earth's surface has been observed [1–3]. This pollutant is a powerful oxidant, affecting human health by causing respiratory and cardiovascular diseases [4]. It also affects vegetation and ecosystems, leading to crop yield and biodiversity losses [5,6]. A total value of 769.2 billion USD loss equivalent to decreases in agricultural production and the occurrence of respiratory diseases and mortality could be ascribed to O<sup>3</sup> exposure in China [7].

O<sup>3</sup> is a secondary pollutant resultant of the reaction between nitrogen oxides (NOx) and volatile organic compounds (VOCs) released to the atmosphere from natural and (mostly) anthropogenic activities [1]. The photochemical regime for surface ozone production determines the sensitivity of O<sup>3</sup> to anthropogenic sources. Usually, urban and suburban zones are VOC-limited due to higher levels of NO<sup>x</sup> emissions, while less-populated zones (rural sites) are NOx-limited [8,9]. Domínguez-López et al. [10] analysed the effect of NO<sup>x</sup> (NO<sup>2</sup> and NO) concentrations in surface ozone at different locations through a spatial and temporal variation study. An opposite daily variance was observed between O3, NO, and NO<sup>2</sup> concentrations in urban and suburban areas. Maximum ozone concentrations

**Citation:** Silva, R.C.V.; Pires, J.C.M. Surface Ozone Pollution: Trends, Meteorological Influences, and Chemical Precursors in Portugal. *Sustainability* **2022**, *14*, 2383. https://doi.org/10.3390/su14042383

Academic Editor: Alessandra De Marco

Received: 22 December 2021 Accepted: 17 February 2022 Published: 19 February 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

were observed in the early afternoon, while NO<sup>x</sup> concentrations usually achieve two peaks (early morning and late afternoon). In rural sites, no hourly peak O<sup>3</sup> or NO<sup>x</sup> was observed. Sun et al. [11] showed that rural sites presented O<sup>3</sup> levels (approximately) 30 µg m−<sup>3</sup> higher than urban regions in the study period of 1990–2019. Therefore, rural and remote areas are the object of many studies, as the highest ozone levels are found there [12,13]. Several factors (such as higher average temperatures and usually lower NO<sup>x</sup> emissions) lead to a more significant accumulation of this pollutant in these areas. There is also evidence of the influence of other chemical air pollutants in O<sup>3</sup> tropospheric levels, such as anthropogenic aerosols, that can affect O<sup>3</sup> photolysis rates directly (through earth radiation scattering) and indirectly (due to the formation of clouds), leading to a weaker surface insolation [14,15]. Li et al. [9] found that the increase of O<sup>3</sup> concentrations (predicted due to the decrease in NO<sup>x</sup> emissions) was in fact generated by the abrupt decrease in PM2.5, resulting in a reduction of aerosols as a sink of HO2, stimulating the production of this secondary pollutant. O<sup>3</sup> is known for its complex formation, depending on many variables. In addition to chemical precursors, meteorological variables are also considered in studying and predicting ozone levels. In different parts of the globe, Fang et al. [16] and Afonso and Pires [17] reached the same results through correlation analysis: ozone shows a positive correlation with temperature, meaning it increases with the increase of this meteorological parameter, and the opposite with relative humidity, that shows a negative correlation with O3. Other variables that can lead to the formation or elimination of surface O<sup>3</sup> are air pressure, wind speed, and direction. The same authors also concluded that O<sup>3</sup> has a negative correlation with local air pressure and a positive correlation with wind speed. Being a pollutant resultant of a radiation-induced chemical reaction, O<sup>3</sup> shows higher levels with clear skies, which is also related to the presence of anticyclones. The association of O<sup>3</sup> concentrations with different weather conditions is more detailed in Domínguez-López et al. [10].

In Portugal, surface O<sup>3</sup> pollution has been characterised in many studies [17–21], especially in the northern [17,22–24] and rural regions [13,25,26]. To complete the study area with more recent data and improve O<sup>3</sup> trends' understanding in Portugal, this study aims to (i) determine the evolution of O<sup>3</sup> levels in Portuguese rural stations, focusing on exceedances to legally imposed levels for human and vegetation protection; (ii) compare the relationship between the O<sup>3</sup> concentrations and one of its precursors (NOx) for different environments; and (iii) evaluate the effect of meteorological conditions in ozone concentrations. This study presents a long-term characterisation (data from more than 10 years) of O<sup>3</sup> concentrations from different environments, including regions in which no similar study was performed yet. In addition, the statistical models were applied to explanatory variables that were selected based on the knowledge of the chemical reactions of O<sup>3</sup> production. This integration enables better performance in predicting O<sup>3</sup> concentrations, allowing, in advance, to prevent and/or mitigate possible unfavourable outcomes.

#### **2. Materials and Methods**

#### *2.1. Ozone Concentration Analysis at Rural Stations*

For a better understanding of the trends and levels of surface ozone at rural sites, the study period 2009 to 2019 was selected. Table 1 represents the geographical information of the 15 existing rural stations [27]. The minimum distance to the seashore was estimated using a tool available on Google Earth that allows measuring the distance between two points. Figure S1 presents the map with the geographical distribution of the rural sites.

The statistical analyses were performed only for O<sup>3</sup> concentrations recorded at stations with a monitoring efficiency higher than 75%. The temporal trend of O<sup>3</sup> concentrations was assessed by determining the annual average concentration for each station. The exceedances to all thresholds presented in the current legislation were determined: (i) alert (240 µm m−<sup>3</sup> ) and information (180 µm m−<sup>3</sup> ) threshold, (ii) target value for human protection (120 µm m−<sup>3</sup> ), and (iii) AOT40, representative of vegetation protection, as a target value (8000 µg m−<sup>3</sup> h) and a long-term goal (6000 µg m−<sup>3</sup> h) [28]. The data collected at

four rural sites from different regions in Portugal were then used to characterise the spatial variability of O<sup>3</sup> concentrations. Histograms with density functions and violin plots were plotted in Python.

**Table 1.** Geographical coordinates, altitude, and minimum distance to the seashore of the existing rural sites.


Cluster analysis (CA) was applied to the rural stations to evaluate the representativeness of the stations with local O<sup>3</sup> measurement. This analysis aims to group monitoring sites in the same class/cluster according to the observed behaviour (daily concentration fluctuation) of collected O<sup>3</sup> data. A hierarchical clustering method was used, generating solutions with 1 to n clusters. Ward's minimum variance method was used to determine the cluster distance. A dendrogram (or tree diagram—graphical representation of hierarchical CA) was determined for each year. Based on the obtained dendrograms, a matrix of relative frequencies was used to pair each monitoring site in clusters, using measured O<sup>3</sup> concentrations. This matrix enables the definition of the number of different O<sup>3</sup> behaviours in the studied remote areas. Daily profiles of O<sup>3</sup> concentrations were determined for each group of monitoring sites and the daily distribution of the daily maximum values using Microsoft Excel Macros developed by the authors.
