作者：Z. Wang, M. Feng, M. F. Johnson, A. Lipani and F. Chan

Reservoirs play a crucial role as sources of methane (CH₄) emissions, with emission rates and quantities varying widely depending on reservoir size due to factors such as surface area, water depth, usage, operational methods, and spatial distribution. Gaining insights into emission characteristics across different reservoir sizes can aid in designing and managing reservoirs to mitigate CH₄ emissions effectively. In this study, machine learning models were applied to estimate both diffusive and ebullitive CH₄ emissions across 97,435 reservoirs in China, spanning five categories of storage capacity. This comprehensive assessment covers nearly all reservoirs within the country, revealing total CH₄ emissions of approximately 5,414 Gg. Reservoirs > 0.01 km(3) are responsible for about 90 % of these emissions, primarily due to high diffusive flux rates and extensive surface areas. Elevated CH₄ diffusion in reservoirs > 0.01 km(3) is largely influenced by their thermal stratification and capacity for organic matter accumulation. Furthermore, these reservoirs are particularly vulnerable to climate warming, which could accelerate CH₄ emission rates more rapidly in larger reservoirs than in smaller ones (below 0.01 km(3)). Consequently, prioritising CH₄ management in reservoirs > 0.01 km(3) is imperative. Nevertheless, the high ebullitive flux of CH₄ in reservoirs < 0.01 km(3), linked to their shallow depth, highlighting the potential for significant CH₄ ebullition from smaller aquatic systems. Given large and small-ranged reservoirs' distinct spatial distribution patterns, targeted management strategies are recommended: project-level management for large reservoirs and basin-level approaches for smaller reservoirs.

（来源：Water Research 2025 Vol. 279  DOI: 10.1016/j.watres.2025.123441）