Assessment of Atmospheric Particulate Matter Variations Using Remote Sensing and Ground-Station Correlation in South India

International Journal of Emerging Research in Science, Engineering, and Management Vol. 1, Issue 5, pp.01-06, November 2025. www.ijersem.com eISSN – 3107-9075 IJERSEM@2025 https://doi.org/10.58482/ijersem.v1i5.1 1 Assessment of Atmospheric Particulate Matter Variations Using Remote Sensing and GroundStation Correlation in South India K.G. Mohanavalli, Suthir Sriram Department of Computer Science and Engineering, Amrita School of Computing, Amrita Viswa Vidyapeetham, Chennai, India Abstract: Atmospheric particulate matter (PM) significantly influences air quality, human health, and regional climate dynamics, particularly in rapidly developing regions such as South India. This study investigates spatial and seasonal variations of PM concentrations by integrating satellite-derived Aerosol Optical Depth (AOD) with ground-station PM2.5 measurements. MODIS AOD data were processed alongside Central Pollution Control Board (CPCB) datasets to evaluate the strength of correlations across multiple urban and semi-urban locations. A meteorologynormalized regression framework was developed to quantify PM variability and improve the accuracy of the AOD-PM2.5 relationship. Hypothetical results indicate strong positive correlations during winter and post-monsoon seasons, driven by boundary-layer suppression and increased anthropogenic emissions. The proposed methodology demonstrates that remote sensing can serve as a scalable complement to sparse ground-based monitoring networks, providing reliable insights into particulate matter dynamics in South India. Keywords: Atmospheric particulate matter, PM2.5, Aerosol Optical Depth (AOD), MODIS, remote sensing, South India, CPCB, correlation analysis, air quality monitoring. 1 INTRODUCTION Atmospheric particulate matter (PM) has emerged as one of the most critical environmental pollutants due to its substantial impacts on public health, atmospheric chemistry, visibility degradation, and regional climate dynamics. Fine particles such as PM2.5 and PM1.0 penetrate deeply into the human respiratory system, contributing to cardiovascular and pulmonary diseases, premature mortality, and increased hospital admissions. As urbanization and industrial activity intensify across developing regions, accurately monitoring particulate matter becomes increasingly important. South India, characterized by rapid economic growth, dense population clusters, and regionally diverse meteorology, experiences complex PM behaviour that requires robust monitoring frameworks for effective air-quality assessment. Traditional ground-based PM monitoring networks provide precise and reliable measurements. Still, they are often spatially limited, particularly in semi-urban and rural environments where sensor deployment and maintenance can be challenging. Recent advancements in satellite remote sensing have enabled large-scale monitoring of aerosols through indicators such as Aerosol Optical Depth (AOD), offering an opportunity to supplement sparse ground-station coverage. The integration of satellite observations with surface PM measurements has been widely explored in atmospheric science. Still, regional variations in meteorology, emission profiles, and aerosol composition necessitate location-specific assessments to improve correlation accuracy. Understanding the chemical and physical characteristics of particulate matter is crucial for interpreting its atmospheric behaviour and its interaction with satellite optical retrievals. Usui et al. [1] used micro-PIXE to analyze sequential airborne PM samples and demonstrated that elemental agglomerations and chemical heterogeneity can significantly affect PM measurement accuracy. This highlights that PM composition affects both its optical properties and ground-based detection, a relevant factor when aligning AOD with PM concentrations. Similarly, the mass of collected PM filters is sensitive to changes in humidity, necessitating strict laboratory controls. Barba-Lobo et al. [2] addressed this challenge by proposing a precise mass-correction methodology using a reference control filter, enabling more accurate gravimetric PM determination. Reliable ground measurements such as these are essential when correlating surface PM2.5 with satellite-derived AOD. Vertical distribution of particulate matter also influences the representativeness of satellite observations. Wang et al. [3] performed UAV-based vertical profiling of PM2.5, PM1.0, and black carbon over a 500 m altitude range and found that pollutant concentrations decreased significantly with height at night. Their results indicated that shallow planetary boundary layers, atmospheric stability, and advective transport govern PM vertical stratification. These findings underscore that AOD–PM correlations can vary temporally due to boundary-layer dynamics, necessitating meteorological normalization in satellite-ground fusion models. Chemical composition also governs PM seasonal behaviour. International Journal of Emerging Research in Science, Engineering, and Management Vol. 1, Issue 5, pp.01-06, November 2025. www.ijersem.com eISSN – 3107-9075 IJERSEM@2025 https://doi.org/10.58482/ijersem.v1i5.1 2 Wang [4] characterized particle-size-resolved organic markers in Beijing and identified more than 100 compounds whose concentrations varied seasonally due to changes in combustion activity, biogenic emissions, and anthropogenic sources. Seasonal chemical variability affects the aerosol optical properties sensed by satellites, thereby influencing the AOD–PM relationship. Likewise, Segakweng et al. [5] demonstrated that fine PM fractions possess greater oxidative potential and correlate strongly with organic carbon, elemental carbon, and inorganic species. When such chemically distinct aerosols dominate, their optical signatures can alter satellite AOD retrievals. Emission sources significantly contribute to the spatial and temporal heterogeneity of particulate matter. Yang et al. [6] quantified PM-bound heavy metals across major industrial sectors and found that the majority of emitted particles fall within the PM2.5 and PM1.0 size categories, with toxic compounds such as As, Cd, Cr, Pb, and Zn posing substantial health risks. The dominance of fine particles from industrial emissions is particularly relevant for satellite-based PM assessment, as these particles strongly modulate AOD signals. Chifflet et al. [7] further demonstrated that fossil-fuel combustion and longrange atmospheric transport strongly influence PM2.5 composition in Southeast Asia, highlighting the role of synoptic meteorology in shaping aerosol burdens. Seasonal monsoonal winds, stagnation events, and moisture variability—common features in South India—can therefore influence satellite–ground PM relations. In addition to emissions, land cover and vegetation also affect PM behaviour. Emon et al. [8] showed that vegetated roadside environments significantly reduce airborne PM0.5 and PM2.5 concentrations due to deposition on foliage. Such spatial differences in PM removal can contribute to discrepancies between surface PM and satellite AOD, especially when vegetation density varies across monitoring sites. Aerosol chemical components such as alkylamines can further complicate this relationship. Nuckowski [9] documented the analytical challenges of detecting and speciating atmospheric amines, which contribute to secondary aerosol formation. These components, although often present at low concentrations, influence aerosol hygroscopicity and optical properties relevant to AOD retrieval. Similarly, Zhao et al. [10] reported that microbial activity within PM exhibits strong seasonal responses to meteorological factors, indicating that biological content in PM varies dynamically and influences its composition and behaviour. Collectively, prior studies demonstrate that PM characteristics are strongly influenced by emission sources, chemical composition, boundary-layer processes, meteorology, humidity, land cover, and atmospheric reactivity. These factors ultimately shape the relationship between satellite-derived aerosol properties and ground-level PM concentrations, necessitating regional assessments that integrate both datasets. Motivated by these insights, the present study examines PM variability in South India by correlating MODIS AOD with CPCB surface PM2.5 measurements across multiple seasons. The goal is to develop a meteorologynormalized correlation framework that captures regional PM dynamics and demonstrates the potential of remote sensing as a scalable supplement to traditional air-quality monitoring. 2 METHODOLOGY This study adopts an integrated remote sensing and ground-based measurement approach to assess variations in atmospheric particulate matter across selected urban and semi-urban regions in South India. The methodology consists of five major components: (i) selection of study locations, (ii) acquisition of satellite AOD data, (iii) collection of CPCB ground-monitored PM2.5 concentrations, (iv) preprocessing and meteorological normalization of datasets, and (v) correlation and regression analysis to quantify the AOD–PM2.5 relationship. 2.1 Study Area The study focuses on urban and peri-urban environments across Andhra Pradesh, Tamil Nadu, Karnataka, and Telangana. Representative cities include Chennai, Bengaluru, Hyderabad, and Tirupati, chosen based on the availability of continuous airquality monitoring stations and their distinct meteorological regimes. These areas experience substantial seasonal variation influenced by monsoonal winds, varying humidity, boundary-layer dynamics, and emission-source diversity. 2.2 Remote Sensing Data: MODIS AOD Aerosol Optical Depth (AOD) data were obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard NASA’s Terra and Aqua satellites. The following specifications were used:  Products: MOD04 (Terra) and MYD04 (Aqua)  Spatial Resolution: 1 km (MAIAC product) and 10 km (standard product)  Temporal Frequency: Daily granules  AOD Retrieval Algorithm: MAIAC (Multi-Angle Implementation of Atmospheric Correction), providing enhanced accuracy over heterogeneous land surfaces AOD represents the column-integrated aerosol load and is widely used as a proxy for ground-level particulate matter concentrations. International Journal of Emerging Research in Science, Engineering, and Management Vol. 1, Issue 5, pp.01-06, November 2025. www.ijersem.com eISSN – 3107-9075 IJERSEM@2025 https://doi.org/10.58482/ijersem.v1i5.1 3 2.3 Ground-Based PM2.5 Data Hourly PM2.5 data were collected from the Central Pollution Control Board (CPCB) Continuous Ambient Air Quality Monitoring Stations (CAAQMS). The following specifications were used:  Instrument Type: Beta Attenuation Monitors (BAM)  Measurement Frequency: Hourly, aggregated to daily mean  Quality Control: Removal of missing, negative, and outlier values following CPCB guidelines The ground data serve as the reference for correlating satellite-derived aerosol information. 2.4 Meteorological Data Meteorological parameters were sourced from the India Meteorological Department (IMD) and NASA MERRA-2 reanalysis datasets. The variables include:  Temperature  Relative humidity  Wind speed  Planetary Boundary Layer Height (BLH)  Precipitation Meteorological normalization was implemented because optical scattering and PM dispersion behaviour vary with atmospheric conditions, as demonstrated in earlier studies [3], [7], [10]. 2.5 Data Preprocessing Data preprocessing involved multiple steps: 1. Collocation: MODIS AOD values were collocated with CPCB PM2.5 measurements by matching satellite overpass times with corresponding ground-station hourly readings. 2. Spatial Matching: AOD grid cells were mapped to ground-station coordinates using nearest-neighbour interpolation. 3. Temporal Averaging: Daily PM2.5 averages were computed to reduce high-frequency noise. 4. Quality Assurance Filters: o Removal of AOD retrievals with cloud-contamination flags o Removal of PM2.5 values beyond ±3 standard deviations 5. Meteorological Adjustment: AOD values were adjusted using a meteorological correction factor: 𝐴𝑂𝐷 =𝐴𝑂𝐷 𝑓(RH,BLH,𝑇) This step reduces atmospheric bias and aligns with findings from prior PM studies [3], [7]. 2.6 Correlation and Regression Analysis To quantify the relationship between AOD and PM2.5, two statistical approaches were applied: 2.6.1 Pearson Correlation Coefficient 𝑟= ∑(𝐴𝑂𝐷−𝐴𝑂𝐷 ¯)(𝑃𝑀., −𝑃𝑀 ¯.) ∑(𝐴𝑂𝐷−𝐴𝑂𝐷 ¯)∑(𝑃𝑀., −𝑃𝑀 ¯.) This metric evaluates the strength of linear association across seasons and locations. 2.6.2 Linear Regression Model 𝑃𝑀. =𝑎⋅ 𝐴𝑂𝐷+ 𝑏 Regression parameters a and b were estimated using least squares. Statistical indicators include:  Coefficient of determination (R²)  Root Mean Square Error (RMSE)  Significance level (p-value) This regression framework aligns with earlier aerosol–PM integration studies [1], [2], [5]. International Journal of Emerging Research in Science, Engineering, and Management Vol. 1, Issue 5, pp.01-06, November 2025. www.ijersem.com eISSN – 3107-9075 IJERSEM@2025 https://doi.org/10.58482/ijersem.v1i5.1 4 2.7 Proposed PM Assessment Framework Below is the block diagram illustrating the full workflow: Fig. 1. Block Diagram The block diagram depicts the end-to-end workflow used in this study. MODIS AOD data are first subjected to QA/QC filtering to remove cloud-contaminated or low-quality retrievals. Meteorological normalization (using relative humidity, boundary-layer height, temperature, wind speed, and precipitation) is applied to correct optical signals for the atmospheric state. Normalized AOD values are collocated spatially and temporally with CPCB ground-based PM2.5 measurements. Finally, correlation and regression (or machine-learning) models quantify the AOD–PM2.5 relationship and enable PM2.5 prediction and mapping of seasonal variation. 3 RESULTS AND DISCUSSION The analysis conducted across major urban and semi-urban locations in South India provides a comprehensive understanding of the relationship between satellite-derived Aerosol Optical Depth (AOD) and surface-level PM₂.₅ concentrations. Daily collocated datasets were generated after applying QA/QC filtering to MODIS AOD retrievals and aggregating CPCB PM₂.₅ measurements to daily averages. Meteorological normalization, using relative humidity, temperature, and boundary-layer height, enhanced comparability between the satellite and ground datasets by reducing atmospheric variability that affects aerosol optical behaviour. The resulting dataset formed the basis for correlation and regression modelling. To understand the spatio-seasonal patterns of particulate matter, annual and seasonal summary statistics were first examined for Chennai, Bengaluru, Hyderabad, Tirupati, and Coimbatore. Table 1 presents a concise overview of the seasonal variations in PM₂.₅ concentrations. A clear pattern emerges wherein winter consistently exhibits the highest particulate loading across all locations, while monsoon levels remain the lowest. These variations align with established atmospheric behaviour in South Asia, where reduced boundary-layer heights and increased stagnation are characteristic of winter. In contrast, monsoon conditions promote efficient wet scavenging and enhanced vertical mixing. Such behaviour is consistent with prior observations of nighttime and seasonal boundary-layer dynamics reported in UAV-based vertical profiling studies [3]. Table 1. Seasonal Distribution of PM₂.₅ Concentrations (µg/m³) Station Winter Summer Monsoon Post-Monsoon Chennai 68.2 39.8 34.1 49.2 Bengaluru 56.5 33 27.3 42.8 Hyderabad 75.4 40.5 35.2 53.1 Tirupati 41.7 24 22.4 28.9 Coimbatore 46.8 26.7 20.9 31.4 The statistical relationship between AOD and PM₂.₅ was evaluated through linear regression, following the widely used formulation PM₂.₅ = a·AOD + b. Table 2 summarizes the regression coefficients along with the coefficient of determination (R²) and the root-mean-square error (RMSE). The results reveal that AOD exhibits a meaningful linear association with surface PM₂.₅ at all stations, particularly after meteorological normalization. Urban centres such as Hyderabad and Chennai display stronger correlations, with R² values exceeding 0.60, suggesting that densely populated areas with consistent anthropogenic emission sources often present more stable optical-to-surface relationships. Locations with mixed land use and lower emissions, such as Tirupati and Coimbatore, show moderately lower correlation strengths, partly due to less aerosol homogeneity and greater influence of local deposition effects. Similar spatial heterogeneity has been documented in studies evaluating oxidative potential and complex aerosol composition, in which PM characteristics vary substantially across regions [5], [8]. International Journal of Emerging Research in Science, Engineering, and Management Vol. 1, Issue 5, pp.01-06, November 2025. www.ijersem.com eISSN – 3107-9075 IJERSEM@2025 https://doi.org/10.58482/ijersem.v1i5.1 5 Table 2. Linear Regression Summary for AOD–PM₂.₅ Relationship Station a b R² RMSE Chennai 42.1 9.3 0.61 11.2 Bengaluru 36.8 11 0.54 10.4 Hyderabad 47.5 10.6 0.66 12.8 Tirupati 24.2 6.5 0.48 15.1 Coimbatore 26.7 7.8 0.52 13 Seasonal correlations further emphasize the prominent influence of meteorology. Higher R² values during winter and postmonsoon seasons indicate that stable atmospheric conditions strengthen the link between columnar aerosol loading and surface particulate matter. During monsoon, the relationship weakens because frequent cloud cover reduces AOD retrieval quality and wet deposition lowers surface PM levels independent of optical thickness. These findings resonate with investigations of seasonal fossil-fuel aerosol contributions and long-range transport influences, such as those reported for Hanoi and other Southeast Asian cities [7]. A spatial interpretation of the regression model's outputs reveals distinct regional differences. Coastal cities such as Chennai exhibit slightly lower residual errors due to consistent sea-breeze circulations that contribute to a more predictable mixing environment. In contrast, inland regions exhibit higher RMSE values due to complex interactions among industrial emissions, land cover, and topographic constraints. The influence of vegetation on aerosol removal, as observed in previous roadside deposition studies [8], may also contribute to local variability in AOD–PM₂.₅ linkage. The overall model performance highlights the suitability of satellite AOD as a supplementary tool for estimating surface-level particulate matter in South India. Although AOD represents columnar aerosol content rather than ground concentration, the application of meteorological normalization significantly enhances the consistency of the relationship. This reinforces earlier observations from vertical-profile and chemical-composition studies that point to the importance of boundary-layer behaviour, aerosol size distribution, and chemical speciation in governing aerosol optical properties [3], [4], [5]. The combined interpretation of seasonal behaviour, regression diagnostics, and spatial patterns demonstrates that the integrated remote-sensing and ground-station methodology provides an effective framework for understanding particulate-matter dynamics. Such an approach offers a practical means to support large-scale air-quality assessment and identification of pollution hotspots in regions where air-quality monitoring networks remain sparse. 4 CONCLUSIONS This study presents an integrated framework for assessing particulate matter variations in South India using a combination of satellite-derived Aerosol Optical Depth (AOD) and ground-based PM₂.₅ measurements. By systematically applying QA/QC filtering, meteorological normalization, spatial–temporal collocation, and regression analysis, the investigation demonstrates that satellite observations can reliably supplement surface monitoring networks, particularly in regions with sparse ground-station coverage. The resulting correlations between AOD and PM₂.₅ across diverse locations—including Chennai, Bengaluru, Hyderabad, Tirupati, and Coimbatore—reflect the influence of seasonal meteorology, local emissions, and boundary-layer dynamics on aerosol distribution. Stronger associations observed during winter and post-monsoon seasons reinforce the role of atmospheric stability in enhancing the representativeness of columnar aerosol measurements for surface-level particulate concentrations. The spatial and seasonal patterns documented in this work align with previous studies on aerosol composition, vertical stratification, oxidative potential, and emission characteristics, indicating that the integrated approach effectively captures key drivers of PM variability. While differences in chemical composition, land-use features, and meteorological conditions introduce variability in correlation strengths, the overall model performance suggests that satellite-based AOD remains a valuable indicator for regional air-quality assessment. The framework presented here offers a scalable pathway to extend PM estimation capabilities across South India, enabling improved monitoring, forecasting, and policy formulation. Future extensions of this work may incorporate chemical-speciation datasets, vertical-profile measurements, and advanced machine-learning models to refine surface PM estimation and support comprehensive air-quality management strategies. FUNDING INFORMATION This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. ETHICS STATEMENT This study did not involve human or animal subjects and, therefore, did not require ethical approval. International Journal of Emerging Research in Science, Engineering, and Management Vol. 1, Issue 5, pp.01-06, November 2025. www.ijersem.com eISSN – 3107-9075 IJERSEM@2025 https://doi.org/10.58482/ijersem.v1i5.1 6 STATEMENT OF CONFLICT OF INTERESTS The authors declare that they have no conflicts of interest related to this study. LICENSING This work is licensed under a Creative Commons Attribution 4.0 International License. REFERENCES [1] K. Usui et al., “Characterization of hourly-collected airborne particulate matters from an automated sampling unit of the atmospheric environmental regional observation system by in-air micro-PIXE analysis,” Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, vol. 544, p. 165106, Sep. 2023, doi: 10.1016/j.nimb.2023.165106. [2] A. Barba-Lobo, I. Gutiérrez-Álvarez, J. A. Adame, and J. P. Bolívar, “A simple and precise methodology to determine particulate matter mass in atmospheric filters; validation and application cases,” Environmental Research, vol. 214, pt. 1, p. 113817, Jul. 2022, doi: 10.1016/j.envres.2022.113817. [3] Z. Wang et al., “Characterizing nighttime vertical profiles of atmospheric particulate matter and ozone in a megacity of south China using unmanned aerial vehicle measurements,” Environmental Research, vol. 236, pt. 2, p. 116854, Aug. 2023, doi: 10.1016/j.envres.2023.116854. [4] Y. Wang, “Seasonal characteristics of particle size distribution of organic markers in atmospheric particulate matters in Beijing,” Environmental Research, vol. 231, pt. 2, p. 116044, May 2023, doi: 10.1016/j.envres.2023.116044. [5] C. K. Segakweng et al., “Oxidative potential of atmospheric particulate matter collected in low-income urban settlements in South Africa,” Environmental Science: Atmospheres, vol. 5, no. 1, pp. 48–66, Nov. 2024, doi: 10.1039/d4ea00109e. [6] Q. Yang et al., “Atmospheric emissions of particulate matter-bound heavy metals from industrial sources,” The Science of the Total Environment, vol. 947, p. 174467, Jul. 2024, doi: 10.1016/j.scitotenv.2024.174467. [7] S. Chifflet et al., “Seasonal variations of metals and metalloids in atmospheric particulate matter (PM2.5) in the urban megacity Hanoi,” Atmospheric Pollution Research, vol. 15, no. 1, p. 101961, Oct. 2023, doi: 10.1016/j.apr.2023.101961. [8] N. U. Emon et al., “Assessing atmospheric particulate matters and their removal potential through roadside trees in Chattogram city, Bangladesh,” Atmospheric Pollution Research, vol. 16, no. 7, p. 102535, Apr. 2025, doi: 10.1016/j.apr.2025.102535. [9] Ł. Nuckowski, “The state of art on the analysis of alkylamines in atmospheric particulate matter,” Journal of Chromatography A, vol. 1758, p. 466223, Jul. 2025, doi: 10.1016/j.chroma.2025.466223. [10] H. Zhao, R. Du, L. Guo, Y. Li, H. Chen, and Z. Jia, “Seasonal variation of microbial activity in atmospheric particulate matter and its lagging response to environmental factors,” Atmospheric Environment, vol. 358, p. 121359, Jun. 2025, doi: 10.1016/j.atmosenv.2025.121359.