We propose a strategy to identify seismic signal classes in continuous single-station seismograms in an unsupervised fashion. Our strategy relies on extracting meaningful waveform features based on a deep scattering network combined with an independent component analysis. We then identify signal classes from these relevant features with agglomerative clustering, which allows us to explore the data in a hierarchical way. To test our strategy, we investigate a two-day long seismogram collected in the vicinity of the North Anatolian fault in Turkey. We interpret the automatically inferred clusters by analyzing their occurrence rate, spectral characteristics, cluster size, and waveform and envelope characteristics. At a low level in the cluster hierarchy, we obtain three clusters related to anthropogenic and ambient seismic noise and one cluster related to earthquake activity. At a high level in the cluster hierarchy, we identify a seismic crisis with more than 200 repeating events and high-frequent signals with correlating envelopes and an anthropogenic origin. The application shows that the cluster hierarchy can be used to draw the focus on a certain class of signals and extract subclusters for further analysis. This is interesting, when certain types of signals such as earthquakes are under-represented in the data. The proposed method can be also used to discover new types of signals since it is entirely data-driven.