Spatiotemporal Analysis of LULC Transitions and NO₂ Patterns in Bali Using Satellite Data and Machine Learning Techniques

This research demonstrates that the historical trajectory of urbanization is a more critical determinant of localized nitrogen dioxide (NO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf>) pollution than static land use/land cover (LULC) classifications. Focusing on Bali, Indonesia a region facing rapid, tourism-driven land change this research applies a fully satellite-based framework guided by the Knowledge Discovery in Databases (KDD) methodology to move beyond simple before-and-after comparisons. The study addresses a key methodological gap by implementing a novel LULC trajectory analysis and a robust spatial aggregation method to avoid the false precision common in multi-scale data integration. High-accuracy LULC maps were generated using a Random Forest algorithm (Kappa> 0.81), and their trajectories were aligned with Sentinel-5P TROPOMI NO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> data. An explanatory Land Use Regression model, validated with 10-Fold Cross-Validation (NRMSE ≈ 5.99%), explained 41.9% of the variance in NO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> and proved that the conversion of vegetation to urban land is the most impactful driver. A subsequent Getis-Ord Gi* analysis identified statistically significant hot spots in southern development corridors, whose land use composition was found to be consistently defined by a high proportion of recent urbanization (often >50<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>) and critically low stable vegetation (<15%). This trajectory informed approach provides a scalable solution for monitoring urban environmental risk in data-scarce regions and offers actionable insight to support vegetation-preserving and spatially targeted air quality policies.