主题：Synergizing Modeling, Data Analytics and Statistical Approaches: Advancing Pollutant Retrieval and Monitoring from Model Development to Spatiotemporal Assessment
主讲人：香港科技大学和香港中文大学Hugo Wai Leung MAK麦伟樑博士
主持人：统计与数据科学学院周岭教授
时间：2026年6月5日（周五）下午4：00-5：00
地点：柳林校区弘远楼408会议室
主办单位：统计与数据科学学院和统计研究中心 科研处
主讲人简介：
Hugo Wai Leung MAK (麦伟樑博士) 现为香港科技大学(HKUST)数学系Visiting Faculty及卓越教育学院(AEE)青年会士，并任职香港中文大学(CUHK)数学系。他同时也是香港人工智能研究院(HKRIAI)荣誉学者、国际数理统计学会(IMS)、香港教育局-课程发展议会、香港考试及评核局及香港科技园(HKSTP) 孵化计划等委员会的成员。他的研究兴趣涵盖计算数学与数据科学、空间数据分析、遥感建模、应用统计与机器学习算法、图像分析，以及可持续和智慧城市发展。他在相关领域和跨学科范畴发表多篇论文(含Q1区多篇)、Google Scholar引用逾930次，H指数为12。他近年来担任多本国际学术期刊的编委、客座编辑和审稿人，及多个国际会议的程序主席和技术委员会成员，并正参与韩国静止轨道环境监测卫星(GEMS)和荷兰皇家气象研究所的CINDI-3国际研究项目。他近年来获得多项荣誉，包括IMS ICSDS 会议差旅奖(2025)、香港中文大学教师流动计划(出境)奖 (2025) 、遥感杰出审稿奖(2024)、香港桂冠论坛青年科学家荣誉 (2023, 2025)、香港科技大学理学院的卓越研究奖(2019)、Epsilon Fund Award (2019)、全球青年科学家高峰会(GYSS)10大海报奖(2017)，以及多个国际会议的最佳报告奖。
内容提要：
Air pollution represents a pressing global environmental threat closely linked to public health and multiple Sustainable Development Goals (SDGs) due to its non-orthogonality. This talk will explore how various high-resolution satellite products, numerical models and data analytic approaches can be effectively synergized to build a multi-step algorithmic framework, for retrieving spatial maps and temporal trends of tropospheric column densities and ground level pollutant concentrations within prescribed spatial domains. With the identification of optimized machine learning approaches from statistical assessments of relevant case studies, missions on pollutant retrieval, real-time monitoring and spatiotemporal assessments across the full modeling pipeline could be advanced. Despite its scientific advancement, traditional air quality models often failed to fully capture statistical fluctuations in pollution trends, extreme climatic events, and cross-regional transport dynamics, while suffering from inherent biases and uncertainties. Therefore, we introduced the bias-adjusted Hybrid Statistical-Dynamic Model (HSDM) that integrates mechanistic relationships between local meteorology, atmospheric chemical transformations, and regional/trans-boundary emission sources. HYSPLIT trajectory analysis and k-mean clustering were also adopted to trace pollutant transport pathways and quantitatively distinguished local versus regional contributions based on wind field characteristics. As for monitoring pollution episodes and extreme events, bias adjustment techniques were further embedded to reconcile numerical model outputs with in-situ observations, with the aim of improving prediction accuracy for historical and next-day pollution profiles. Based on outputs from The Geostationary Environment Monitoring Spectrometer (GEMS) mission, the talk will also explore the statistical assessments that reveal strong linkages between NO₂ column density and XCO₂ anomalies, which enabled the identification of regions with co-occurring CO₂ and NOₓ emissions.
These research advancements could effectively resolve key limitations of conventional models, support evidence-based trans-boundary pollution control policies, and provide technical support for regional environmental management and citizen science applications via data analytic means. This interdisciplinary integration also bridges modeling, observation, and data science, offering a scalable paradigm for continental-scale air quality management; while highlighting future directions for data synergy and uncertainty reduction with advanced statistical algorithms and models.
