走向现代数学学术报告 - 黄国文副教授（No. 954)

报告题目：Spatially Varying Multi-Pollutant Calibration of Air Quality Model Outputs

报 告 人：黄国文 副教授（Western University）

邀 请 人：陈森明 副教授

报告时间：2026年7月6日 11:00

报告地点：东海岸校区-D实209

报告摘要：Air quality monitoring networks provide reliable local measurements but are spatially sparse, while chemical transport models such as CMAQ provide complete gridded fields but can exhibit systematic local bias. This talk presents a spatially varying multi-pollutant calibration framework for integrating NAPS observations with CMAQ outputs for PM2.5, NO2, and O3 over Ontario. The proposed approach models each observed pollutant as a location-specific affine function of the CMAQ pollutant vector, allowing both pollutant-specific correction and cross-pollutant borrowing. Smooth coefficient surfaces are estimated using radial basis functions and penalized least squares, then transferred from monitoring sites to CMAQ grid cells to produce calibrated spatial fields. The talk will discuss the data construction, motivation from station-level heterogeneity, model formulation, coefficient interpretation, prediction transfer, and practical limitations. Results show that spatially varying calibration improves fitted error relative to a global multivariate calibration benchmark and provides interpretable maps of local bias, scaling, and cross-pollutant structure.

个人简介：Dr. Guowen Huang is an Assistant Professor in Statistics at Western University. He earned his PhD in Statistics from the University of Glasgow and previously held postdoctoral research positions at National Tsing Hua University and the University of Toronto. Before joining Western, Dr. Huang was an Associate Professor at Shantou University. His research specializes in spatial statistics, particularly the statistical modeling of air pollution data and its effects on human health.