Satellite-Based Estimation of Nitrous Oxide Concentration and Emission in a Large Estuary.

Estuaries are nitrous oxide (NO) emission hotspots and play an important role in the global NO budget. However, the large spatiotemporal variability of emission in complex estuary environments is challenging for large-scale monitoring and budget quantification. This study retrieved water environmental variables associated with NO cycling based on satellite imagery and developed a machine learning model for NO concentration estimations. The model was adopted in China's Pearl River Estuary to assess spatiotemporal NO dynamics as well as annual total diffusive emissions between 2003 and 2022. Results showed significant variability in spatiotemporal NO concentrations and emissions. The annual total diffusive emission ranged from 0.76 to 1.09 Gg (0.95 Gg average) over the past two decades. Additionally, results showed significant seasonal variability with the highest contribution during spring (31 ± 3%) and lowest contribution during autumn (21 ± 1%). Meanwhile, emissions peaked at river outlets and decreased in an outward direction. Spatial hotspots contributed 43% of the total emission while covering 20% of the total area. Finally, SHapley Additive exPlanations (SHAP) was adopted, which showed that temperature and salinity, followed by dissolved inorganic nitrogen, were key input features influencing estuarine NO estimations. This study demonstrates the potential of remote sensing for the estimation of estuarine emission estimations.