Apart from the several natural sources of methane (CH4) and ozone (O3) emissions, several anthropogenic sources such as industrial exhausts and agriculture emit these gases directly into the atmosphere. There has been a lack of research that analyses vegetation as the primary source of CH4 and O3 in the atmosphere. This study seeks to integrate the Sentinel-5P datasets with the different parameters of vegetation growth and gaseous dispersal in the troposphere using remote sensing datasets from other remote sensing satellite sensors such as thermal and multi-spectral. The multi-sensor remote sensing data integration approach is important to establish sensor interoperability to understand the variations in the concentrations of CH4 and O3 in a better and more logical manner, besides making the study more reliable. A high correlation was observed between the Normalized Differential Vegetation Index (NDVI) derived from the multi-spectral satellite data with the tropospheric concentrations of CH4 and O3 with an R2 value of 0.732 and 0.668 respectively. The Land Surface Temperature (LST), wind speed, tropospheric concentration values of CH4 and O3, and NDVI values were fed into a Support Vector Regressor model, and different kernel performances were analysed based on the estimated concentrations of CH4 and O3. It was observed that the SVR with Radial Basis Function (RBF) kernel performed better with R2-statistics of 0.646 and 0.557 in the training and testing phases respectively. The study is a first-of-its-kind approach to identifying the agricultural environment as a potential source of O3 and CH4 emissions in the atmosphere.
Keywords: Agriculture; Data integration; Multi-sensor remote sensing; SVR.
© 2025. The Author(s), under exclusive licence to Springer Nature Switzerland AG.