Spatiotemporal Patterns of Methane and Nitrous Oxide Emissions in China's Inland Waters Identified by Machine Learning Technique

The fugitive emissions of greenhouse gases, primarilymethane (CH4) and nitrous oxide (N2O), fromwater environmentshave aroused global concern. However, there are currently limitedinformation about national-scale data of CH4 and N2O emissions from inland waters, such as lakes, rivers, andreservoirs, particularly in developing countries. This study employedmachine learning techniques, based on the literature data and nationalwater quality monitoring data, to reveal the CH4 and N2O emission patterns of China's inland waters at thethird-level basin and daily resolution. Our results show significantseasonal variations in CH4 emissions, which were influencedby total nitrogen and chemical oxygen demand concentrations. Northernwatersheds were identified as hotspots of CH4 emissions,with 57% higher CH4 flux than the other watersheds. Incontrast, N2O had a relatively lower contribution to totalcarbon emissions and showed smaller temporal and spatial variations.The estimated total emissions of CH4 and N(2)Oin China's inland waters in 2021 amounted to 80.22 Tg of carbondioxide equivalent, accounting for 9-11% of China'sterrestrial carbon sinks. This research provides valuable insightsto guide the counting and control of greenhouse gas emissions fromenvironmental water bodies. High-resolutionspatiotemporal footprint of methane andnitrous oxide emissions from China's inland waters was providedand the drivers for the spatial and seasonable variations were identified.