Report Title

A Neural Network-Based Scale-Adaptive Cloud-Fraction Parameterization Scheme

Report Time

July 20, 2023 (Thursday) 9:00-11:30 a.m

Report Content

The report mainly introduces a novel neural network-based scale-adaptive cloud-fraction parameterization scheme. Contrasted with traditional cloud-fraction parameterization schemes, the novel scheme presents three distinct advantages: 1) By employing neural networks, it sidesteps unwarranted assumptions regarding the functional structure of parameterization scheme functions; 2) Constructed using CloudSat data, it effectively reduces errors stemming from foundational data; 3) It accommodates the influence of model horizontal and vertical resolutions, making it adaptable across different resolutions and variable-resolution models. According to the findings, this scheme can more accurately depict sub-grid cloud-fraction variations with cloud water content and relative humidity. This leads to a marked enhancement in the simulation of cloud-fraction vertical structure. There is substantial promise for enhancing the simulation performance of cloud radiation effects within global climate models.

Guest Information

Chen Guoxing is a professor at Fudan University who obtained his Bachelor's and Ph.D. degrees from Peking University in 2008 and 2013 respectively. From 2013 to 2020, he studied and worked at the State University of New York at Albany. Since September 2020, he has been teaching in the Department of Atmospheric and Oceanic Sciences at Fudan University. He has long been engaged in research on aerosol-cloud-climate interaction and regional climate change based on numerical models and deep learning, with his research accomplishments possessing profound scientific foresight and application value.