Abstract

Excessive ground pollutant concentrations, in particular fine particulate matter (PM2.5) cause devastating impacts towards human health and environmental conditions within neighborhood scales. In acquiring accurate PM2.5 retrieval in temporal and spatial manners, scientists and modelers attempt to describe the statistical relationship between Aerosol Optical Depth (AOD) and available PM measurements obtained from various instrument and campaigns, which can be derived from observation-based or simulation-based approaches. This presentation first identifies the potential deficiencies of existing data analytic and physical mechanisms, then proceeds with the introduction of a systematic framework that could better reveal the seasonal and spatial discrepancies of ground PM2.5 concentrations, especially in highly polluted geographical clusters and developing cities. Such framework combines the use of observational, satellite and remotely sensed datasets, the understanding of intrinsic data structures and meteorological effects, the consideration of vertical aerosol structure and humidity correction, as well as relevant numerical modeling strategies, and is proven to result in promising statistical performance in several selected case studies of China and Siberian urban areas. After validating the capability of this developed framework in detecting overall trends of PM2.5 within prescribed geospatial set-ups, the talk will extend to the discussion of extreme pollution events and the methodological modifications needed, which involve the use of the HYSPLIT backward trajectory model in identifying air mass trajectory, thus local and regional pollutant sources can be more effectively separated; the application of bias-adjustment approaches for conducting environmental forecasts, with focus on capturing extreme pollution events and spikes, as well as the abrupt changes of PM2.5 concentrations that took place within a short time period. Finally, the PM2.5 retrieval algorithm in the international Geostationary Environment Monitoring Spectrometer (GEMS) mission will be briefly described, which can provide potential room for improvement in relevant algorithmic development and data synergies for larger-scale environmental monitoring in the future.