Hierarchical Clustering for Optimizing Air Quality Monitoring Networks

Abstract

Hierarchical clustering (HC) analysis groups datasets into clusters based on their (dis)similarity, and can be used to assess air-quality monitoring networks representability. The methodology describe here is a new approach to designing optimized air-quality monitoring networks by combining Kolmogorov-Zurbenko filtering (KZ) and HC of observed and modelled time series. Here we present the optimization of the air quality network in the province of Alberta, Canada, for NO2, SO2, PM2.5 and O3. The study suggests that network optimization will vary depending on chemical species due to different emissions sources and/or the results of secondary chemistry. Making use of hourly and time-filtered time series allows identifying emission sources, with much of the signal identifying sources emissions residing in shorter time scales (hourly to daily) due to short-term variation of concentrations, and background concentrations can be identified by larger time scales (monthly or over). The methodology is also capable of generating maps of sub-regions within which a single station will represent the entire sub-region, to a given level of dissimilarity, when applied to gridded datasets such as chemical transport modelling output.