资源环境科技发展态势分析平台

Investigating the association between weather circulation patterns and high PM2.5 episodes is useful for interpreting the connection between physical weather and chemical weather. Principal component analysis (PCA) is often applied to decompose circulation modes but has limitations for studying high PM2.5 events related circulation patterns. This study describes an improved circulation classification integrated with PCA and k-means algorithm oriented to high local PM2.5. The classification scheme was applied in Xiamen, southeastern China, when local PM2.5 exceeded 75 mu g m(-3) (the 24-hour limit Chinese Ambient Air Quality Grade II standard) during the winter 2013. Nine typical circulation patterns were classified. Circulation patterns related to the highest PM2.5 concentrations were associated with a negative pressure anomaly at 850 hPa over the Sea of Japan which yielded a strong transport of PM2.5 from northern China during windy days. The improved classification methodology links large-scale circulation to local PM2.5 in target city and is able to distinguish possibly different circulation patterns over continual haze episodes. This clustering method can be applied in any cities and would be useful for predicting chemical weather and serving local environmental policymakers.