Urban Air Quality: Aerosol Pollution Exploration and Impact Assessment

1. Introduction

Air quality plays a vital role in both human and ecosystem health, with poor air quality causing numerous health problems and approximately 3.3 million deaths annually since 2000 [1,2,3,4,5]. This underscores the necessity for clear strategies and policies to address air pollution. Globally, countries have adopted legislative measures to reduce pollution and establish limits for key pollutants [6,7,8]. As an EU member state, Romania is also making efforts to improve air quality, including promoting sustainable clean energy technologies and implementing strategies such as the Green Deal, which aims to reduce greenhouse gas emissions to zero by 2050 [9].

Approximately 65% of urban areas worldwide have experienced an increase in PM2.5 concentrations since 2000 [1,2,3,4]. This pollutant, generated by specific sources, has a varied composition and different toxicological, physical, and chemical properties. Elemental carbon (EC) and organic carbon (OC) are key components of particulate matter with distinct sources and behaviours. Biomass burning is a major contributor to both, generally producing a higher proportion of organic carbon (OC), resulting in a low elemental carbon to organic carbon (EC/OC) ratio. Desert dust, primarily composed of minerals, can interact with anthropogenic emissions, incorporating EC and OC into complex aerosol mixtures. In such cases, the EC/OC ratio serves as a valuable indicator of source contribution and mixing processes, with implications for both air quality assessment and source apportionment [10,11,12]. These aerosol types (biomass burning, desert dust, and mixtures) are associated with a wide range of health effects due to their varying chemical composition, particle sizes, and toxicological properties, as summarized in Table 1. Biomass burning aerosols and mixtures are the most hazardous to health, as they contain toxic chemicals, polycyclic aromatic hydrocarbons and black carbon. These fine particles enter the bloodstream and deposit in organs, causing serious health effects. Although desert dust can affect health, especially in vulnerable people (children, the elderly, the chronically ill), it is considered less hazardous than aerosols from biomass burning, which are strongly linked to chronic diseases and mortality [13,14,15,16,17,18,19,20,21,22,23,24,25]. The sources of elemental carbon in Romania include both local (vehicular traffic) and long-range transport (Eastern European High and regional influences, as well as agricultural activities) [26,27,28,29].

Given these risks, accurately assessing the presence and distribution of PM2.5 is essential. A critical parameter for evaluating atmospheric aerosol loadings is Aerosol Optical Depth (AOD), which measures the attenuation of solar radiation due to scattering and absorption by particles suspended in the atmosphere. Although AOD does not directly represent surface-level PM2.5 concentrations, it serves as a valuable proxy when integrated with other datasets. The relationship between AOD and PM2.5 is complex and context-dependent, affected by local meteorology, aerosol type, and geographical factors. Nonetheless, several studies have demonstrated the utility of AOD for tracking pollution patterns and identifying aerosol sources, particularly when supported by the Ångström exponent (AE), precipitable water (PW), and atmospheric models like HYSPLIT [30,31,32,33,34,35].

The primary objective of this paper is to propose a new integrated approach for evaluating air quality and predicting the impact and risk of specific pollutants based on AOD analysis. The new approach was applied based on monitoring air quality data for two cities in Romania for 13 and 12 years, respectively (for the period 2011-2023 and 2012-2023). By integrating AOD into environmental impact and risk assessment, this research provides a novel toolset for monitoring pollution dynamics, identifying high-risk periods, and supporting policy making. The framework enhances traditional monitoring by enabling broader spatial analysis and improved forecasting capabilities. Furthermore, it emphasises the importance of aerosol classification and source apportionment in understanding local air quality trends, ultimately contributing to more informed and effective mitigation strategies.

2. Materials and Methods

2.1. Site Description

According to European regulations, Romania monitors air quality through 148 stations: 30 traffic-type (assessing the impact of traffic), 58 industrial-type (monitoring industrial activity), 37 urban and 13 suburban background (assessing the influence of settlement and human activity), 7 regional reference stations, and 3 EMEP stations (for monitoring cross-border pollution). To enhance aerosol property analysis, Romania also hosts 8 AERONET stations in Timisoara, Cluj-Napoca, Iasi, and Bucharest, which monitor local and long-range transported aerosol events. This study considers Iasi and Cluj-Napoca for integrated air quality assessment. Iasi (northeast Romania) faces major air pollution issues, with PM2.5 levels 2.3 times above the legal limit. In Cluj-Napoca (northwest), PM2.5 is also the main pollutant. In both cities, the primary sources of PM2.5 pollution include industrial activity, traffic, and residential heating. Iasi, with about 271,692 residents and 60,000 students, is a major cultural and academic centre. It has six air quality stations and one AERONET station (Iasi_LOASL) (Figure 1). Cluj-Napoca, known as ”the heart of Transylvania”, has about 286,598 inhabitants, five air quality stations, and one AERONET station at a university lab (Cluj_UBB) [36,37,38,39,40].

2.2. Experimental Data

The values used in this study come from the AERONET (Aerosol Robotic Network) database - a global aerosol monitoring system operated by NASA and partners (agencies, universities, institutions, researchers, etc.). Unlike models limited to ground-level or full-column data via remote sensing, this study focuses on optical aerosol parameters that indicate aerosol type, such as fine-mode particles (EC/OC mixtures) signalling biomass burning or soot; coated large particles that absorb solar radiation (black carbon, polluted dust); and mixtures like OC/dust, dust/EC, or dust-dominated aerosols [26,41,42,43,44].

Adapting the Integrated Impact and Risk Assessment method to aerosol-type characterisation via photometric data enables impact analysis on atmospheric components. AERONET data are categorized into three levels: unscreened (Level 1.0), cloud-screened (Level 1.5), and quality-assured cloud-screened (Level 2.0). Aerosol Optical Depth (AOD) - a key parameter - serves diverse applications in climate and weather modelling, visibility, aviation safety, and air quality due to its indication of total aerosol load [45,46,47,48]. Higher AOD values indicate a greater presence of aerosols, which negatively affects visibility and air quality [49,50]. AOD helps fill gaps where ground monitoring stations are lacking and correlates with PM2.5, a major health-impacting pollutant. Studies have shown that PM2.5 estimates improve when AOD is integrated into machine learning models [51,52,53].

For this study, AERONET photometric data from Iasi (LOASL: 47.19°N, 27.55°E) and Cluj-Napoca (UBB: 46.77°N, 23.55°E) stations were used. Both stations have improved understanding of local air quality by contributing aerosol optical property data and validating satellite-based observations [54,55,56].

We used inversion products from Level 1.5 data (version 3), selected using Level 2.0 criteria: solar zenith angle >50° and retrieval error <5% [57]. AOD at 400 nm was used to cover both polluted and unpolluted events. Level 1.5 was preferred due to its near-real-time availability – within one hour of measurement [58,59,60,61,62] – important for timely risk and impact assessment [63,64,65,66]. Note that Level 2.0 data require annual instrument calibration and verification.

Following established clustering methods, absorption Ångström exponent (AAE) and scattering Ångström exponent (SAE) were calculated at 440 nm and 675 nm, respectively, to classify aerosol types [42,54,62,67].

2.3. Integrated Assessment Methodology

Integrated Impact and Risk Assessment is an approach for quantifying environmental impact. The first step consists of establishing the environmental components, which are given a degree of importance (values from 0 to 1), depending on the specifics of the assessed activity and the opinions of the environmental specialist. Based on matrix calculations, normalised scores and importance units (IU) are obtained, which are subsequently used in calculations. The next step is represented by environmental quality quantification (Q), as shown in Equation (1). Finally, considering the units of importance (IU) and environmental quality (Q), environmental impact quantification (EI) and risk quantification (ER) are completed using Equations (2) and (3) [63,64,65,66]. In this case, the method was adjusted, and the importance of the environmental component under assessment (AOD) was considered at its maximum value (1). Thus, the IU were 1000, while the probability of pollution occurrence (p) was also considered at its maximum value (1). It results that ER is similar to EI, having the same values.

$$Q={\displaystyle \frac{{MAC}_i}{{MC}_i}}$$

(1)

Equation 1. Environmental quality quantification (Q)

Where:

Q is the environmental quality;

MAC is the maximum allowed concentration (in the absence of clear maximum allowed concentrations established by legislation for aerosols, it was necessary to adapt the integrated assessment method; in this case, we considered for the MAC the highest AOD value measured in our data set);

MC is the measured concentration.

$${EI}_i={\displaystyle \frac{IU}{Q}}$$

(2)

Equation 2. Impact on environmental quality quantification (EI)

Where:

EIᵢ is the environmental impact considering the quality indicator i;

IU is the importance unit assigned to each environmental component;

Q is the quality of the environmental component considering the quality indicator i.

$${ER}_i={EI}_i\times p$$

(3)

Equation 3. Environmental risk quantification (ER)

Where:

ER_i is the environmental risk;

EIᵢ is the environmental impact considering the quality indicator i;

P is the probability of impact/pollution occurrence.

After quantifying the environmental impact and risk, their classification follows, depending on the scenario in which the obtained EI values fall, according to Table 2.

2.4. Rescaling Method

The scores resulting from the initial impact and risk assessment method exceeded the threshold of 700, which, according to the interpretation in Table 2, indicates a severely affected environment, unsuitable for life. However, the measured AOD values don’t reflect a critical or catastrophic risk. For this reason, it was deemed necessary to develop a rescaling method that would more accurately reflect the actual implications of different types of aerosols. The applied calculation formula is shown in Equation (4).

$$y={\displaystyle \frac{\left(x-x_{min}\right)·\left(y_{max}-y_{min}\right)}{\left(x_{max}-x_{min}\right)}}+y_{min}$$

(4)

Equation 4. Rescaling formula

Where:

x is the value to be rescaled;

x_max, x_min are the limits of the original interval (EI calculated);

y_max, y_min are the limits of the new interval (we stopped at 700 because no measurements were made under catastrophic conditions);

y is the rescaled value.

Since the scientific literature demonstrates the distinct health and environmental effects of different aerosol categories, we performed a classification, selecting EC (elemental carbon) and DD (desert dust) for analysis and applying differentiated rescaling for these categories. While biomass burning – the main source of EC – is associated with severe diseases and increased mortality rates [13,14,15,16,17,18,19,20,21,22,23,24,25], DD aerosols tend to have less severe effects (Table 1). The environmental impacts of these two aerosol types also differ significantly. Although DD pollution poses certain risks, especially in arid regions, scientific studies consistently show that its environmental impact is substantially lower compared to that of EC from biomass burning. The latter releases complex toxic compounds and contributes more significantly to global warming, whereas DD increases suspended particle concentrations with more localized and less severe impacts [68,69,70,71,72,73]. Furthermore, considering that the intensity of health and environmental effects depends on the distance from the emission source, the negative impact of DD diminishes significantly as it travels farther from its origin and approaches the study areas considered here (Table 3) [13,14,15,16,17,18,19,20,21,22,23,24,25,68,69,70,71,72,73]. Since aerosol mixtures originate from both natural and anthropogenic sources, including this category in the risk score assessment complicates the linkage between emission sources and policy development. Based on these differences and the risk categories presented in Table 2, the EI (environmental impact) values for EC were rescaled between 0 and 700, while those for DD were rescaled between 0 and 350, thus reflecting the different magnitude of the impact on the environment and health. Rescaling was applied starting from 0, as scenarios without significant pollution were also included, allowing for rescaling from the minimum value.

Moreover, to validate the rescaling method, the standard assessment method was applied again, this time assigning differentiated scores to aerosol categories. EC was considered of utmost importance (assigning a score of 1 in impact assessment), while for the DD category, scores ranging from 0.1 to 1 were assigned progressively.

Furthermore, to validate both the proposed rescaling method and the risk-level-based scoring system, two standard statistical indicators were calculated: Root Mean Square Error (RMSE) and Mean Absolute Error (MAE), which are presented in the following section.

2.5. Validation With Statistical Indicators – Error Calculation: Root Mean Square Error (RMSE) and Mean Absolute Error (MAE)

The performance of the proposed rescaling method and the risk-level-based scoring system was evaluated using two widely recognized statistical indicators: Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). These parameters allow quantifying the difference between the rescaled values and the values obtained from applying the integrated method for EC and DD, thereby providing an objective basis for evaluating the method’s performance. RMSE is sensitive to large deviations and penalizes them more heavily, making it useful for identifying score combinations that generate significant discrepancies – an essential aspect in establishing the most suitable pair of scores for EC and DD. At the same time, MAE indicates the average absolute deviation, offering a clear picture of the average error across the entire dataset. The use of both indicators allows for solid validation and a comprehensive understanding of the method: RMSE helps detect situations with large errors, while MAE provides a general benchmark of the method’s overall accuracy. Although there is no international standard for classifying RMSE and MAE values, applied literature in the fields of climate, energy, and air quality model evaluation frequently uses percentage thresholds to interpret performance. Based on this, we have structured a guideline table (Table 4) that classifies RMSE and MAE values according to their degree of agreement with the reference. In principle, the smaller the error, the more accurate the model [74,75,76,77,78,79,80].

3. Results and Discussion

3.1. Overview of Aerosol Type Distribution (Iasi and Cluj-Napoca)

3.1.1. Iasi AERONET Monitoring Site

For the city of Iasi, 1209 measurements were selected from the period 2012-2023, of which 975 correspond to the elemental carbon category, and 234 to the desert dust category. The aerosols measured by AERONET over Iasi for 12 years (2012 – 2023), in general, were dominated by the occurrence of an elemental carbon (EC) and organic carbon (OC) mixture at 55% (see “EC/OC mixture” in Figure 2). The fine mode particles, grouped as “EC dominated” in Figure 2, indicate that 18% of the aerosol came from combustion/biomass burning or soot particle types. The aerosol types: “Dust/EC mix” (1%), “Mix” (2%), “OC dominated” (3%), “OC/Dust Mix” (6%) and “Coated large particles” (11%) were not discussed here, since they were present in relatively small percentages compared to “EC/OC mixture” and “EC dominated”.

3.1.2. Cluj-Napoca AERONET Monitoring Site

For the city of Cluj-Napoca, 1380 measurements were selected from the period 2011-2023, of which 1222 correspond to the elemental carbon category, and 158 to the desert dust category. Also, the aerosols measured by AERONET over Cluj-Napoca for 13 years (2011 – 2023), in general, were dominated by the occurrence of an elemental carbon (EC) and organic carbon (OC) mixture at 46% (see “EC/OC mixture” in Figure 3). The “EC dominated” in Figure 3 indicates that 22% of the aerosol came from combustion/biomass burning or soot particle types. Similarly, the aerosol types: “Mix” (2%), “Dust/EC mix” (3%), “OC/Dust Mix” (8%), “Coated large particles” (8%) and “OC dominated” (10%), were not discussed here.

3.2. Performance Evaluation Using RMSE and MAE

To validate the rescaling method and identify the most appropriate pair of scores for EC and DD, we based our analysis on the error classification presented in Table 4.

The results showed that, for both cities, the lowest values of both error metrics (RMSE and MAE), located in the percentage range 0-5% (which means a very small error and an excellent performance, according to Table 4) were achieved with importance scores of 0.4 and 0.5 for DD and 1 for EC (Figure 4). Therefore, the error analysis confirms the validity of the selected rescaling thresholds and importance scores, demonstrating their effectiveness in accurately capturing the real differences between aerosol categories.

3.3. Environmental Impact Quantification Calculated Using AERONET Data

Figure 5, Figure 6, Figure 7 and Figure 8 show the rescaled EI, calculated using AOD values from the Iasi and Cluj-Napoca monitoring stations, considering only the ”dust-dominated” and ”EC-dominated” categories. The ”EC-dominated” category falls within the subset of fine particulate matter associated with pollution from anthropogenic sources, such as fossil fuel combustion, industrial emissions, and road traffic. The methodology employed to quantify the impact is described in Section 2 – Materials and Methods.

The categories "EC/OC mixture" and "OC dominated" are also specific to fine particles; however, in terms of health effects, elemental carbon (carbon black) is considered more toxic than organic carbon [29]. The "EC/OC mixture" combines natural and anthropogenic sources, making it complex to consider this category when calculating a risk score and correlating pollution sources with policymaking.

The distribution of the rescaled environmental impact values for the elemental carbon category (EC) in Iasi (Figure 5) shows a predominance in the minor to moderate risk range. More than half of the recorded values (53.3%) fall between 100 and 350, suggesting an acceptable level of impact that still requires monitoring. Only 0.2% of the values exceed the threshold of 500, indicating isolated episodes of severe pollution, likely attributed to meteorological conditions or intense anthropogenic sources. The AOD associated with these episodes ranges from 0.88 to 1.05, reflecting a significant aerosol load on the atmospheric column. Therefore, EC remains a major pollutant of concern for local public health and environmental strategies.

For the desert dust category (DD) in Iasi, the impact distribution is significantly more attenuated (Figure 6). The majority of rescaled values (65.4%) fall below the threshold of 100, reflecting minor or even negligible risks. This trend is consistent with the hypothesis that DD effects diminish significantly with increasing distance from the emissions source. Even during episodes with higher AOD, the rescaled impact rarely exceeds 300, supporting the validity of the differentiated impact assessment method. Thus, DD’s contribution to air quality degradation in Iasi appears to be marginal compared to that of EC.

For Cluj-Napoca, the impact distribution for the EC category (Figure 7) is similar to that observed in Iasi, with 55.9% of values falling between 100 and 350, indicating moderate but acceptable risks. As in Iasi, only 0.2% of the values exceed the threshold of 500, suggesting a low incidence of critical pollution events. The maximum recorded AOD value (0.88) is slightly lower than in Iasi, which may reflect either lower combustion source intensity or more efficient pollutant dispersion. These results confirm that EC remains a dominant factor in atmospheric impact within the urban environment.

For DD in Cluj-Napoca (Figure 8), the impact distribution follows the same pattern as in Iasi, with generally low values; however, the majority (58.9%) falls within the 100-350 range, characteristic of minor risks where monitoring actions are required. The AOD associated with this interval ranges from 0.24 to 0.67. This behaviour confirms the secondary role of DD in the pollution profile of Cluj-Napoca.

Because the traditional approach to air quality assessment does not correlate with ground data, this study's results underscore the urgent need to utilise an AOD risk index to effectively highlight the impact of pollution on the entire atmospheric column. Thus, in addition to the climate change already highlighted by scientists, the radiative budget (influenced by aerosol intrusion) can also negatively impact life quality and may be of interest.

4. Conclusions

This research demonstrated the applicability and relevance of an innovative integrated approach for air quality evaluation and the assessment of impact and risk related to atmospheric pollutants, using AOD (Aerosol Optical Depth) as a key indicator. The study was conducted for two representative Romanian cities – Iasi and Cluj-Napoca – using AERONET data collected over extended periods (12 and 13 years, respectively), ensuring the robustness of the findings.

The results revealed a predominance of elemental carbon (EC) and organic carbon (OC) mixtures in both cities, primarily sourced from traffic emissions, biomass burning, and industrial activities. Specifically, Cluj-Napoca exhibited an EC/OC mixture of 46%, while Iasi had 55%. The impacts and risks calculated as the AOD index for EC show a few major ones, with an AOD index range from 0.88 to 1.05 for Iasi city and 0.73-0.88 for Cluj-Napoca.

By adapting the Integrated Impact and Risk Assessment method to the specific context of aerosol characterization, the study offered a differentiated classification of aerosol types, emphasizing the higher risks associated with EC compared to desert dust. The proposed method was statistically validated using RMSE and MAE indicators, achieving excellent performance (<5%).

Therefore, this study offers an effective and replicable tool to support environmental authorities in monitoring, forecasting, and decision-making processes regarding air quality – especially in urban areas where PM2.5 pollution is persistent and often insufficiently quantified by ground-based monitoring networks.

Given the high performance of the proposed method, a future research direction involves its application at a more detailed temporal scale – enabling the analysis of aerosol pollution risks based on seasonality, specific days, or annual variations – as well as extending the study to other urban or rural regions to allow for comparative assessments and the development of regional pollution control strategies.