Background:
Atmospheric pollution at the surface reflects interacting effects of emissions, chemical transformation, secondary formation, and atmospheric transport. Ground-based measurements provide detailed local information, whereas remote sensing describes atmospheric conditions over broader spatial domains. However, differences between surface and column observations complicate their interpretation. A process-based framework is therefore needed to relate recurrent surface pollutant regimes to atmospheric-column variability.

Objectives:
This study aims to develop and evaluate an exploratory process-based framework linking surface pollutant regimes with column-integrated atmospheric observations through chemical transformation, secondary formation, vertical redistribution, and transport.

Methods:
Hourly surface observations from an urban-background monitoring station in Tirana for 2010–2012 were analysed using principal component analysis (PCA), non-negative matrix factorization (NMF), and k-means clustering. The dataset comprised 16,632 observations of 11 pollutant variables. PCA identified dominant covariance structures, NMF characterized non-negative pollutant co-variation patterns, and clustering identified recurrent multivariate regimes. Separately, ERA5/GIOVANNI data for 2024–2025 were used to examine ozone (O₃), temperature (T), and relative humidity (RH) relationships using Pearson and Spearman correlations, partial correlations, spatial analysis, and distributional assessment. NMF was used exclusively for exploratory pattern recognition and not quantitative source apportionment.

Results:
The first three PCA components explained 58.8% of total variance, accounting for 27.2%, 21.9%, and 9.8%, respectively. NMF and k-means identified three recurrent pollutant patterns interpreted as traffic/combustion, coarse-particle/resuspension, and photochemical/secondary regimes. In the ERA5/GIOVANNI analysis, O₃ showed a strong negative correlation with temperature (r = −0.825, p < 0.001), a positive correlation with RH (r = 0.344, p < 0.001), and temperature and RH were negatively correlated (r = −0.610, p < 0.001). These results indicate substantial coupling between ozone variability and meteorological conditions. However, the observed statistical relationships cannot independently demonstrate specific chemical mechanisms or establish direct correspondence between historical surface regimes and contemporary atmospheric-column observations.

Conclusion:
The proposed "Invisible Bridge" provides an exploratory framework for interpreting surface–column differences through atmospheric transformation and transport. Its current evidence is process-based rather than a direct validation of PMF–remote-sensing integration, providing a basis for future temporally matched and quantitatively integrated observations.