A clustering approach based on triangular diagram to study the seasonal variability of simultaneous measurements of PM10, PM2.5 and PM1 mass concentration ratios

This paper presents a clustering approach based on triangular diagram used to study the seasonal variability of ratios between mass concentrations of PM10, PM2.5 and PM1. Simultaneous measurements of PM10, PM2.5 and PM1 mass concentrations were categorized according to the season they were sampled and the corresponding size-segregated particulate matter ratios (i.e. PM1/PM10, PM2.5/PM10, PM (1) /PM2.5 and PM2.5-PM1/PM10-PM1) were calculated and plotted using a triangular diagram. A hierarchical cluster analysis (HCA) was used to group all simultaneous particulate matter (PM) mass concentration measurements. The triangular diagram results show a general trend from low values of size-segregated particulate matter ratios related to sampling in warmer seasons (i.e. spring-summer) towards higher values of size-segregated particulate matter ratios related to sampling in colder seasons (i.e. winter-autumn). The HCA identifies three well differentiated clusters for simultaneous measurements of PM mass concentrations. These clusters divide the triangular diagram into three well distinguishable areas. These areas relate to data recorded in warmer seasons (PM2.5/PM10 < 50 %, PM1/PM10 < 40 % and PM2.5-PM1/PM10-PM1 < 30 %), data recorded in colder seasons (PM2.5/PM10 > 60 %, PM1/PM10 > 40 % and PM2.5-PM1/PM10-PM1 > 35 %) and data recorded in both warmer and colder seasons (PM2.5/PM10> 50 %, PM1/PM10> 30 % and PM2.5-PM1/PM10-PM1 < 35 %). The latter area can be regarded as a transition region, occurring where the PM intermodal size fraction is about one-half of the coarse one. The triangular diagram also emphasizes how the intermodal size fraction may not be neglected compared to both coarse and fine size fractions in some environments in colder seasons.