Quantifying source apportionment for ambient haze: An image haze extraction approach with air quality monitoring data

Abstract

The annual average concentration of fine particles (PM2.5) in Taiwan has been decreasing yearly. However, ambient haze has not abated and visibility is still poor. This study proposes a method for quantifying the source apportionment of ambient haze from various constituents, namely measured PM2.5, moisture, and secondary organic aerosols (SOA, represented by the sum of measured O3 and NO2 levels). This study's model-based demisting technology integrated image haze extracting technology with air quality monitoring data. The images for haze extraction were from the instant cameras of the Taiwan Environmental Protection Agency's air quality monitoring stations (AQMSs). Results indicate that haze constituents with the product of light extinction and scene depth have very high correlation (R2 ≥ 0.8452), which demonstrate that this quantification approach was successful. The contributions of ambient haze from various constituents had quarterly and regional variations in the eastern, northern, central, and southern regions of Taiwan, which are slightly, lowly, moderately, and heavily contaminated areas, respectively. Each area was assigned a selected representative AQMS to represent its ambient haze characteristics. Cases where the dominant contributors to haze were PM2.5, moisture, and SOA were studied, in addition to the event days of PM2.5 and ozone. We detail the limits of this approach, especially with regard to the use of images. This approach can help administrators understand the composition of ambient haze to generate effective haze strategies and policies. Residents can use this method to determine the health effects of daily haze.