We introduce a clustering method for time series based on tail dependence. Such a
method considers spatial constraints by means of a suitable dissimilarity index that merges temporal and spatial dependence via extreme-value copulas. The proposed approach is applied to the study of rainfall extremes.
