Atmospheric Precision: Validating Sentinel-5p and Data Against Ground-Based Measurements in Sri Lanka

This study demonstrates the accuracy of satellite-based air pollution data from Sentinel-5P (OFFL and NRTI products) by comparing it with ground-level measurements of NO₂ and SO₂ collected by the Central Environmental Authority (CEA) of Sri Lanka. This study seeks to determine how reliable and precise satellite monitoring is compared to traditional ground-based methods, especially in areas where data is limited. Satellite-derived column densities (mol/m²) were processed using Google Earth Engine (GEE) and converted to parts per billion (ppb) by considering variations in altitude and atmospheric pressure. Comparative analyses were performed for the cities of Kandy and Battaramulla, where CEA monitoring stations are currently in operation. The results were then evaluated against CEA observations using standard error metrics, including Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE). The results show a clear seasonal pattern and a significant correlation between satellite and ground-based measurements for NO₂, demonstrating that the satellite effectively captures trends in urban pollution. On the other hand, the data for SO₂ demonstrated less consistency, pointing out the challenges that satellite sensors face in accurately measuring SO₂ levels close to the ground. Such variability is probably due to the reactive characteristics of SO₂ and the localised sources associated with its emissions. Differences can be explained by factors such as spatial averaging, the integration of the atmospheric conditions, and variations in measurement methods. The analysis finds that Sentinel-5P data, especially regarding NO₂, is important for assessing air quality in developing countries. It suggests combining this data with ground-based networks to improve monitoring efforts. Future research should look into AI-enhanced fusion models and the growth of ground station networks in challenging environments.