.png){kind=small}
25 NOV 2025 (TUE) 10:00-10:45 | 11:15-12:00 | 14:00-14:45
- GEOG HKU
- 4 days ago
- 5 min read
Updated: 3 days ago
Departmental Research Seminars Series
GeoAI, Remote Sensing, Land-Atmosphere/Ocean or Human-Environment Interactions
Date: 25 NOV 2025 (Tuesday)
Time: 10:00-10:45 | 11:15-12:00 | 14:00-14:45 (HKT)
Venue: CLL, Department of Geography, 10/F, The Jockey Club Tower, Centennial Campus, HKU
Mode: Hybrid
Meeting ID: 977 0595 0258
Password: 939815
Registration link: https://hkuems1.hku.hk/hkuems/ec_hdetail.aspx?guest=Y&ueid=104140
10:00-10:45 | Via Zoom
A Time Series Foundation Model for Land Surface Dynamics Monitoring
Abstract:
Existing remote sensing foundation models rely on cloud-free image patches or patch-based time series inspired by computer vision architectures, which are unable to effectively utilize partially cloudy observations. However, near real-time monitoring of surface dynamics and the retrieval of quantitative biophysical and geophysical variables (e.g., soil moisture and carbon fluxes) fundamentally rely on leveraging all available cloud-free satellite observations to provide timely, frequently updated information and to account for memory effects in ecosystem processes. This is challenging as satellite time series data are inherently irregular due to cloud contamination with variations in record length and acquisition dates across both locations and years.
This study presents a novel method to address the challenge and demonstrates its application for land cover monitoring, within-season crop type mapping, and time series reflectance modeling and reconstruction across the conterminous United States using the Harmonized Landsat and Sentinel-2 (HLS) data. The proposed approach directly processes HLS raw irregular time series data as input using the Transformer model with the masked attention mechanism. Consequently, this method maximizes the use of temporal information to deliver more accurate and timely results compared to traditional machine learning or patch-based deep learning approaches. Additionally, the study develops a strategy for building a robust time-series generative pre-trained foundation model, named HLS-GPT, which demonstrates superior performance compared to the NASA–IBM Prithvi model. This advancement provides a transformative pathway for more effective utilization of irregular time series data in operational large-scale geospatial analysis and land surface dynamics monitoring.
Dr Hankui ZHANG
Associate Professor, Geospatial Sciences Center of Excellence, Department of Geography and Geospatial Sciences, South Dakota State University, USA
Hankui Zhang (he/him) is an Associate Professor at the Geospatial Sciences Center of Excellence, confounded by South Dakota State University (SDSU) and the USGS Earth Resources Observation and Science (EROS) Center. He completed his Ph.D. at the Chinese University of Hong Kong (CUHK) in 2013. Dr. Zhang's research focuses on developing quantitative remote sensing algorithms to enhance the usability of satellite data. Notably, his BRDF correction method has been adopted by NASA to produce Harmonized Landsat and Sentinel-2 (HLS) products. He has also advanced the application of deep learning in remote sensing for land cover mapping and the retrieval of biophysical and geophysical variables. He has organized a special issue for the leading journal Remote Sensing of Environment and has published over 70 peer-reviewed SCI papers, garnering more than 7,000 citations. Dr. Zhang is also a member of the USGS-NASA Landsat Science Team.
11:15-12:00 | In Person
Food for Thought: Mapping Sustainable Diets from Global Patterns to Everyday Practices
Abstract:
Food consumption sits at the centre of contemporary sustainable human–environment interactions, shaping environmental impacts, health outcomes, and societal resilience. In this talk, I present my multi-scale research—from global patterns to household and individual practices—to examine how food consumption evolves across diverse contexts and generates differentiated environmental, health, and socio-economic outcomes. Drawing on mixed-method approaches that integrate life-cycle assessment, behavioural modelling, ethnographic insights, and advanced data analytics, I investigate how environmental pressures, cultural norms, affordability, and ambient environments jointly influence food-related behaviours in China, the United States, Brazil, the United Kingdom, and other regions. This work uncovers the socio-technical mechanisms through which sustainability challenges are produced and mitigated, revealing cross-country contrasts in the environmental and nutritional trade-offs of everyday choices. Building on these insights, I explore opportunities for leveraging digital tools to support behavioural change, and applying refined geographical data including remote sensing and GIS to examine biodiversity loss, land-use pressures, and food environment associated with food system transitions. By integrating cultural, behavioural, and human geographical perspectives, this research highlights pathways for advancing sustainable and equitable diets across countries and ethnic group.
Dr Pan HE
Senior Lecturer, School of Earth and Environmental Sciences, Cardiff University, United Kingdom
Pan He is a Lecturer in Environmental Science and Sustainability at Cardiff University. Her research focuses on environmental and health outcomes of human consumption behaviour, with particular emphasis on sustainable food systems, climate-related risks, and socio-technical practices. She integrates environmental impact modelling with behavioural scientific methods such as experiments, questionnaires, and interviews to inform policy and systemic interventions. Her research has been published in leading journals such as Nature Climate Change, Nature Food, Nature Sustainability, and One Earth. She has led and co-led internationally collaborative projects funded by major UK and Chinese agencies, including the Royal Society International Exchanges, the National Natural Science Foundation of China, the China Postdoctoral Science Foundation, GW4 Generator Funding, and Nanjing University’s International Fellowship for Young Scholars. Her work spans multi-country, interdisciplinary partnerships across the US, the UK, China, and Europe, advancing global understanding of how everyday behaviours shape environmental and health outcomes.
14:00-14:45 | In Person
Atmospheric Chemical Transport Modeling: Linking Chemicals in the Sky and Earth’s Environment
Abstract:
Atmospheric chemical transport models utilize mathematical frameworks to simulate atmospheric chemicals. These models are essential for understanding physical and chemical processes that contribute to significant environmental issues such as air pollution, climate change, and ecosystem degradation. In this talk, I will discuss our efforts aimed at developing models to deepen our understanding of the interactions between human activities, atmosphere, and climate. The focus will be on two storylines: light-absorbing aerosols and halogens.
Light-absorbing aerosols, particularly black carbon (BC) and brown carbon (BrC), play critical uncertain role in climate change. I will share a decade of my research journey on understanding these compounds. While our early studies indicate policies aimed at reducing light-absorbing aerosol emissions, as previously recommended by IPCC, might have limited efficacy in mitigating global warming; our latest work unveils the hidden role of dark BrC, an underappreciated warming agent from biomass burning.
Halogen radicals (Cl, Br, I), along with their precursors, are predominantly generated through land-ocean-atmosphere interactions. These radicals affect the tropospheric environment in multiple ways, such as altering the lifetime of greenhouse gases, stimulating ozone pollution, and enhancing cloud formation. Despite their significance, the mechanisms regulating halogen concentrations remain poorly understood. In this part, I will introduce our modeling efforts regarding the global cycling of Cl, Br, and I in the atmosphere and elucidate their effects on atmospheric oxidants, particles, and air quality.
Dr Xuan WANG
Assistant Professor, School of Energy and Environment, City University of Hong Kong, Hong Kong SAR
Prof. Xuan Wang is an Assistant Professor at City University of Hong Kong. He obtained his Ph.D. in Environmental Chemistry from MIT in 2017, after which he served as a postdoctoral fellow at Harvard University before joining CityUHK. His research focuses on the chemical composition of the atmosphere and its environmental impacts. His group develops models and integrates them with observations to study the complex interactions among human activities, natural processes, and the earth’s climate. He has contributed over fifty publications on leading journals such as One Earth, Nature Geoscience, and ES&T, with related works highlighted by Nature and One Earth for providing scientific insights into climate mitigation policy. He is an early career editorial board member of the journal ACS ES&T Air, reviewer for dozens of journals including Science and various Nature series publications, and the advisor for Vision Carbon Limited, a start-up based in Hong Kong.
Comments