Identification of source regions of PM10 with backward trajectory-based statistical models during PM10 episodes

This study applies backward trajectory-based statistical techniques, residence time, conditional probability and emission attraction to evaluate potential source regions of PM10 over a coastal region. PM10 episodes were selected by principal component analysis for 1998-2005 over the Kaoping air quality basin. Residence time was applied to identify potential regions in which air parcels would remain over their 6- and 12-h trajectories. Emission attraction and conditional probability were used to analyze contribution ratios of distinct emission sources to air quality stations. The PM10 episodes screen 175 days (6 % of total days) and 35.9 % of total station numbers. Residence time and emission attraction clearly identified potential areas in which backward trajectories remained during PM10 episodes and high PM10 events. Emission attraction evaluated relative contributions of various sources (stationary, line, and area) from specific jurisdictions, and provided information on specific sources for high-priority PM10 emissions reduction. The conditional probabilities of emission attraction during high PM10 events show that high values concentrated near stationary and area sources in the city of Kaohsiung.