Machine-Learning Representation of Aerosol Surface Tension Improves Simulations of Aerosol-Cloud Interactions in Global Climate Models
编号:703
访问权限:仅限参会人
更新:2026-08-31 18:29:34 浏览:0次
张贴报告
摘要
Surface tension has usually entered global climate models as a fixed property of pure water or as a bulk-solution correction. This treatment makes interfacial thermodynamics a secondary parameter and obscures a pathway by which organic aerosol phase state and mixing state can regulate cloud activation. Here we develop a machine-learning parameterization trained on particle-resolved simulations of composition-dependent effective surface tension and implement it in the Community Atmosphere Model. Rather than prescribing a single macroscopic surface tension, the framework translates aerosol size, composition, humidity, and organic enrichment into an activation-relevant surface tension that remains compatible with modal cloud activation schemes.
The resulting simulations show that dynamic surface tension is not simply a uniform modifier of cloud condensation nuclei. Its largest impact emerges where the default critical supersaturation lies close to the cloud maximum supersaturation. In these susceptible regimes, a moderate reduction of the Kelvin barrier can shift particles across the activation threshold. This reframes organic surface enrichment from a passive chemical property into a threshold-controlling variable for aerosol-cloud interactions.
The surface-tension pathway propagates beyond aerosol activation. Sensitivity experiments indicate coherent responses in cloud droplet number, liquid water path, low-cloud properties, and shortwave cloud radiative effect, especially in regions where organic-rich particles coexist with clouds that are sensitive to additional activated droplets. The response is therefore controlled not only by the abundance of organic aerosol, but also by the joint distribution of particle size, hygroscopicity, interfacial state, and cloud dynamical supersaturation.
This work extends earlier bulk-surfactant and parcel-model studies by embedding particle-resolved interfacial physics into a global climate model. It provides a process-based route to connect aerosol mixing state and organic emissions with cloud-radiative uncertainty, and establishes testable diagnostics for future CCN closure and satellite-based cloud evaluation. Dynamic aerosol surface tension may therefore represent a missing pathway linking microscopic particle structure to macroscopic aerosol-cloud-radiation interactions.
稿件作者
Wenxiang Shen
Nanjing University
发表评论