报告题目：Urban-Scale Atmospheric Component Transport Based on Physical Models and Machine Learning Methods
报 告 人：李林峰 博士
报告时间：2026年9月4日（周五） 9:40-10:20
报告地点：藕舫楼724室
主 持 人：王曰朋 教授
报告摘要：
This talk presents two recent studies on modelling atmospheric composition transport in urban environments. First, a conditioned U-Net model is developed for predicting concentrations from a single emission source. The machine learning model is then used for emission inversion within a variational framework, leveraging automatic differentiation for gradient computation. Second, I will introduce our recent development of an urban carbon dioxide transport model that considers both anthropogenic and biogenic emissions. A case study of the London Borough of Camden is used for model validation.
报告人简介：
Dr. Linfeng Li got his PhD in the Department of Earth Science and Engineering at Imperial College London and is now a Postdoctoral Research Associate at Imperial. His expertise lies in finite-element physical modelling and neural networks (NN). He was the first to introduce graph neural networks for tackling fluid-structure interaction problems. Currently, his research focuses on modelling urban carbon dispersion with AI methods—supporting sustainable city design and planning, and ultimately contributing to global atmospheric carbon management.
欢迎广大师生踊跃参加！
数学与统计学院
江苏省应用数学（南京信息工程大学）中心
江苏省系统建模与数据分析国际合作联合实验室
2026年9月3日
