On the declustering methods of seismic catalogue — an application over Indian subcontinent

Seismic catalogue preparation is one of the important steps of probabilistic seismic hazard assessment (PSHA). Catalogue preparation includes homogenization and declustering as the main stem of pre-processing followed by a completeness check to arrive at the recurrence relation. Homogenization is the process of converting the events reported with different magnitude scales into a uniform scale whereas declustering is the process of removing dependent events, i.e. foreshocks and aftershocks from an earthquake catalogue. Accepting the subjectivity involved in the adopted process of homogenization and completeness check, the selection of declustering method apparently results in significantly different declustered catalogues owing to the variation in respective spatial and temporal interaction zones. This paper is aimed to investigate sensitivity of the selection of declustering method on the resulting recurrence relations. Indian catalogue (2474 BC to July 2020 AD) and that in three smaller regions (Delhi, North-East, and Andaman) of 300-km radius are considered for the purpose of investigation. Standard approach is used for homogenization. Two widely used procedures of declustering are considered: one requires a set of specific model parameters whereas the other does not. A uniform space–time window is recommended to be used with the former regardless of the selection of model parameters followed by Stepp’s method for the completeness check. Proposed modification does not only make the former approach less sensitive (in terms of number of events and resulting recurrence relation parameters) to the model parameters but also leads to the consistent results when compared with the latter approach.