Calculating air mass trajectories is common in atmospheric analyses. However, if explainable results are to be achieved, several procedures are needed to process the vast amount of information handled. Clustering methods are statistical tools usually considered for such a purpose. Although they are based on rigorous algorithms, certain questions still remain when these methods are applied. The current review is organised in sections according to the sequence followed by such procedures. First, the types of clustering methods are described, with their core being the distance used. One key point is the stopping rule, which determines the final number of clusters. A simple classification based on this number is then suggested. Finally, the graphical presentation of the results is examined and the main drawbacks are commented on. A range of applications and results are considered to illustrate each section, and certain caveats and recommendations are also presented.
