A B S T R A C T
In recent years, Sentinel-5P satellite data have proven effective for monitoring spatiotemporal air pollution. This study analyzes four major pollutants - NO₂, CO, SO₂, AAI - over Kerman, southeastern Iran during the period 2018–2024. Annual and seasonal distribution maps were generated, and satellite data validated with ground measurements. Correlations with LST, EVI, DEM, and road density were examined, and a Composite Index of Air Pollution (CIAP) was developed. The results revealed clear seasonal variability in pollutant concentrations, with higher levels generally observed during colder seasons. A noticeable decline in CO and NO₂ concentrations was detected in 2022, highly likely associated with reduced human activities during the COVID-19 pandemic. Spatial analysis indicated that urban Zone 4 exhibited the highest mean CIAP value (0.773), alongside elevated concentrations of NO₂ and CO, whereas the lowest CIAP value (0.647) was observed in Zone 3. Validation results demonstrated acceptable agreement between satellite and ground-based observations, with correlation coefficients of 0.52 for SO₂, 0.50 for CO, 0.48 for NO₂, and 0.43 for AAI. Correlation analysis further showed strong positive relationships between NO₂ and CO (r = 0.79) and between pollutant levels and road density (r up to 0.62), while negative correlations were observed with elevation and vegetation cover. Overall, the findings confirm the capability of Sentinel-5P data for urban air quality monitoring in arid environments and highlight the combined influence of anthropogenic and environmental factors on air pollution patterns, providing valuable insights for urban air quality management and environmental policymaking.
Extended Abstract
Introduction
Air pollution has emerged as a major environmental challenge of the 21st century, affecting human health, ecosystem functionality, and urban environments. In developing countries, rapid urbanization, industrial growth, and increasing vehicular emissions have intensified air pollution. Among the major atmospheric pollutants, nitrogen dioxide (NO₂), sulfur dioxide (SO₂), carbon monoxide (CO), and aerosols have important environmental and health implications. Kerman, located in southeastern Iran, is particularly relevant for air-quality assessment because of its arid climate, expanding urban area, industrial activities, and limited vegetation cover.
Urban air pollution is influenced by anthropogenic and environmental factors, including population density, industrial activities, traffic intensity, land use, meteorological conditions, topography, and vegetation cover. These factors contribute to spatial and temporal variations in pollutant distributions. Although ground-based monitoring stations provide valuable air-quality data, their limited spatial coverage can restrict the assessment of pollution patterns across urban areas.
Satellite remote sensing offers a complementary approach to overcoming these limitations. The Sentinel-5P satellite, equipped with the TROPOspheric Monitoring Instrument (TROPOMI), provides observations of several atmospheric constituents, including NO₂, SO₂, CO, and aerosol-related indicators. These observations enable the investigation of pollutant distributions and their temporal variations over large areas. Previous studies have demonstrated the value of Sentinel-5P and geographic information systems (GIS) for identifying pollution patterns and hotspots. Research in Iran has also highlighted the influence of industrial activities, meteorological conditions, wind patterns, and geographical characteristics on atmospheric pollution. In addition to assessing individual pollutants, composite indicators can provide an integrated representation of pollution. The Combined Index of Air Pollution (CIAP) combines information from multiple pollutants to facilitate the identification and comparison of areas experiencing relatively high cumulative pollution pressure.
Despite advances in satellite-based air-quality assessment, a comprehensive spatial and temporal analysis of multiple pollutants in Kerman over an extended period remains necessary. Therefore, this study investigates the spatial and temporal patterns of NO₂, SO₂, CO, and the Aerosol Absorbing Index (AAI) in Kerman during 2018–2024. It also develops and applies the CIAP index, evaluates satellite observations against ground-based monitoring data, and examines the relationships between pollutants and selected environmental and anthropogenic variables. The findings aim to improve understanding of urban air pollution in Kerman and support air-quality monitoring and management in areas with limited ground-based monitoring infrastructure.

Methodology
This study investigated the spatial and temporal patterns of NO₂, CO, SO₂, and the Aerosol Absorbing Index (AAI) in Kerman during 2018–2024 using Sentinel-5P satellite observations. The analysis comprised three components: pollutant trend and spatial-distribution analysis, development and validation of the Combined Index of Air Pollution (CIAP), and assessment of relationships between pollutants and environmental and anthropogenic factors.
Pollution Trend Analysis and CIAP Index Development
Satellite observations from the TROPOspheric Monitoring Instrument (TROPOMI) aboard Sentinel-5P were obtained through Google Earth Engine (GEE) for January 2018–December 2024. Annual and seasonal composites were generated to examine temporal variations and spatial distributions of NO₂, SO₂, CO, and AAI. The processed datasets were subsequently transferred to ArcGIS Pro for spatial analysis and, where necessary, resampling to establish a consistent spatial framework.
The CIAP index was developed to provide an integrated assessment of relative pollution pressure. Because satellite observations represent atmospheric column measurements and are not directly equivalent to health-based permissible concentration standards, each pollutant was normalized using the minimum–maximum method to obtain dimensionless values between 0 and 1. The normalized variables were then aggregated through weighted averaging, with equal weights assigned to all pollutants because locally applicable information on their relative toxicity was unavailable. CIAP was interpreted as a relative composite indicator for comparing pollution intensity across the study area, rather than as a substitute for individual pollutant assessments or health-based air-quality standards.
Validation Using Ground-Based Data
Satellite observations were evaluated against daily measurements from the Moshtagh air-quality monitoring station for 2018–2022. Satellite and ground-based data were temporally matched, and only dates with valid observations from both sources were included. Satellite values were extracted from the pixel closest to the monitoring station and compared with corresponding ground-based measurements. Agreement was assessed using the Pearson correlation coefficient (R) and coefficient of determination (R²). Given the differences between satellite-derived atmospheric column measurements and ground-level observations, the analysis focused on temporal association rather than direct numerical equivalence.
Correlation with Environmental and Anthropogenic Factors
Environmental and anthropogenic variables included elevation, land surface temperature (LST), vegetation cover, and road density. Elevation data were obtained from the Shuttle Radar Topography Mission (SRTM) digital elevation model at 30 m resolution, LST from Landsat observations, and the Enhanced Vegetation Index (EVI) from MODIS data, using Google Earth Engine. Road density was calculated through kernel density analysis in a GIS environment.
The environmental layers, pollutant variables, and CIAP were aligned to a consistent spatial framework. A regular sampling grid was established across the study area, and values were extracted at corresponding sampling locations. Pearson correlation coefficients were calculated in RStudio to examine the direction and strength of relationships between pollutants, CIAP, and environmental and anthropogenic variables. Correlation strengths were interpreted using the Chaddock scale.

Results and Discussion
The results revealed substantial spatial, seasonal, and interannual variability in NO₂, SO₂, CO, and AAI across Kerman during 2018–2024, indicating differences in emission sources and environmental influences.
AAI varied considerably across years and seasons, with relatively high values around 2022 followed by some reduction. Higher values were frequently observed in central and peripheral areas, indicating spatial heterogeneity in aerosol-related conditions. NO₂ showed elevated values in areas associated with urban activities and transportation, with a noticeable decline in 2022. CO exhibited similar temporal variations, including a reduction around the same year, potentially associated with changes in human mobility and anthropogenic activities during the COVID-19 pandemic. These temporal patterns suggest that urban pollutant levels respond to both long-term development and short-term changes in human activity.
SO₂ displayed a distinct spatial pattern, with relatively high values in western and northwestern parts of the study area, where industrial and mining activities may contribute to emissions. Seasonal variations included relatively elevated levels during colder periods. Atmospheric stability, wind direction, and other meteorological conditions may influence the distribution and accumulation of this pollutant. Overall, areas in the western, northwestern, central, and eastern parts of Kerman frequently exhibited elevated pollutant levels. These patterns may reflect the combined effects of industrial and mining activities, traffic intensity, urban development, and atmospheric circulation.
The CIAP analysis provided an integrated assessment of relative pollution pressure across municipal areas. Zone 4 recorded the highest average CIAP value (approximately 0.773), accompanied by comparatively high NO₂ and CO values, whereas Zone 3 had the lowest average value (approximately 0.647). These differences identify areas with relatively greater and lower cumulative pollution pressure within the study area.
Validation against observations from the Moshtagh monitoring station for 2018–2022 yielded coefficients of determination (R²) ranging from approximately 0.185 to 0.279. Among the individual pollutants, SO₂ showed the highest reported correlation (R ≈ 0.52), whereas AAI showed the lowest (R ≈ 0.43). These results indicate modest agreement between satellite-derived and ground-based observations. The differences may reflect the distinct measurement characteristics of the datasets: Sentinel-5P provides atmospheric column information, whereas ground stations measure air quality near the surface. Differences in spatial representativeness, atmospheric conditions, and temporal matching may also affect the observed relationships. Therefore, satellite observations are best used alongside ground-based measurements rather than as their direct substitutes.
Correlation analysis highlighted associations between atmospheric pollutants and environmental and anthropogenic factors. Road density showed positive relationships with pollutants, particularly NO₂ and SO₂, suggesting an association between transportation intensity and pollution distribution. NO₂ and CO were also strongly positively correlated (r ≈ 0.79), indicating potentially related emission patterns. In contrast, vegetation cover and elevation generally showed negative relationships with several pollutants. Greater vegetation cover was associated with lower pollutant values, while elevation may influence pollution distribution through topographic and atmospheric processes. These associations emphasize the combined roles of transportation, industrial activities, vegetation, topography, and atmospheric circulation in shaping spatial pollution patterns across Kerman.
Overall, the findings indicate that air pollution in Kerman results from interactions among anthropogenic activities and environmental conditions. The spatial distribution of pollutants, differences in CIAP values, and observed relationships with environmental variables demonstrate the value of combining satellite observations, ground-based measurements, and GIS-based analysis for identifying relative pollution hotspots and supporting urban air-quality management.

Conclusion
This study demonstrated the value of integrating Sentinel-5P observations, ground-based measurements, GIS-based analysis, and composite indicators to assess urban air pollution in Kerman during 2018–2024. The analysis revealed substantial spatial, seasonal, and interannual variability in NO₂, SO₂, CO, and AAI. The observed reductions in NO₂ and CO around 2022 may be associated with changes in human mobility and anthropogenic activities during the COVID-19 pandemic. Spatial analysis identified areas with relatively elevated pollution levels, particularly in zones influenced by industrial and mining activities, transportation networks, and urban development. The CIAP index further distinguished areas with different levels of relative cumulative pollution pressure, with Zone 4 recording the highest average value (0.773) and Zone 3 the lowest (0.647).
Validation against ground-based observations indicated modest agreement between satellite-derived and surface measurements, highlighting the importance of interpreting satellite data in conjunction with monitoring-station records. Correlation analysis also revealed positive associations between road density and selected pollutants, as well as a strong positive relationship between NO₂ and CO (r ≈ 0.79). Conversely, vegetation cover and elevation generally showed negative associations with several pollutants. These findings suggest that air pollution patterns in Kerman reflect the combined influence of anthropogenic activities and environmental conditions. However, the results should be interpreted as relative spatial and temporal patterns rather than direct assessments against health-based air-quality standards.
From an environmental management perspective, the findings support expanding ground-based monitoring, particularly in identified pollution hotspots, and strengthening the management of industrial and mining emissions. Improving public transportation and developing appropriate urban green infrastructure may also contribute to reducing local pollution pressure. Future research should incorporate additional monitoring stations, detailed meteorological variables, and advanced statistical or machine-learning models to improve the interpretation and prediction of pollutant distributions. The integrated framework applied in this study can support air-quality assessment and management in Kerman and other cities with limited monitoring infrastructure.

Funding
There is no funding support.

Authors’ Contribution
A. Rashki was responsible for conceptualization, designing the research methodology, formal analysis of satellite data, final editing, as well as supervision and executive management of the research.
S. Taleghani contributed to writing the original draft, data acquisition, validation of findings, data management, implementation of satellite image processing codes, and map visualization.

Conflict of Interest
The authors declared no conflict of interest.

Acknowledgments
We are grateful to all the scientific consultants of this paper.