Application of k-means and hierarchical clustering techniques for analysis of air pollution: A review (1980–2019)

Clustering is an explorative data analysis technique used for investigating the underlying structure in the data. It described as the grouping of objects, where the objects share similar characteristics. Over the past 50 years, clustering has been widely applied to atmospheric science data in particular, climate and meteorological data. Since the 1980's, air pollution studies began employing clustering techniques, and has since been successful, and the aim of this paper is to provide a review of such studies. In particular, two well known and commonly used clustering methods i.e. k-means and hierarchical agglomerative, that have been applied in air pollution studies have been reviewed. Air pollution data from two sources i.e. ground-based monitoring stations and air mass trajectories depicting pollutant pathways, have been included. Research works that have focused on spatio-temporal characteristics of air pollutants, pollutant behavior in terms of source, transport pathways, apportionment and links to meteorological conditions, comprise much of the research works reviewed. A total of 100 research articles were included during the period of 1980–2019. The purpose of the clustering approach, the specific technique used and the data to which it was applied constitute much of the discussion presented in this review. Overall, the k-means technique has been extensively used among the studies, while average and Ward linkages were the most frequently applied hierarchical clustering techniques. Reviews of clustering techniques applied in air pollution studies are currently lacking and this paper aims to fill that gap. In addition, and to the best of the authors' knowledge, this is the first review dedicated to clustering applications in air pollution studies, and the first that covers the longest time span (1980–2019).