Towards a New Zealand model for short-term earthquake probability: Aftershock productivity and parameters from global catalogue analysis

Abstract

The project 'Towards a New Zealand model for short-term earthquake probability: Aftershock productivity and parameters from global catalogue analysis ("Catalogue Analysis")' had two main objectives: First to study the occurrence of foreshocks to extend an existing aftershock model into a model for short-term earthquake probabilities and second to include the new model into the Californian 'Regional Likelihood model' (RELM) testing.

The question whether foreshocks behave like mainshocks that happen to have larger aftershocks is important for the understanding of earthquake nucleation and the modelling of earthquake clustering. A commonly used model for short-term earthquake occurrence, the Reasenberg-Jones model, combines empirical observations of aftershock occurrence and extends them to potential foreshocks assuming a single triggering mechanism for earthquake clusters. However, observed foreshock probabilities have been reported significantly lower than predicted by this kind of model. To investigate the discrepancy between the model prediction and the observations we reviewed an earlier description of aftershock rates. We derived a variation of the Reasenberg-Jones model which was based on the mean abundance, i.e. the average number of aftershocks per sequence in a chosen time and mainshock magnitude interval.

We calculated model parameters from the Californian catalogue by first defining earthquake sequences and secondly analyzing aftershock occurrence. The new model formulation predicted foreshock numbers that were consistent with the data. We concluded that no separate triggering mechanism was required to derive foreshock occurrence from aftershock models and thus foreshocks could be regarded as mainshocks with larger aftershocks.

Finally, we investigated the spatial and temporal separation between foreshocks and their mainshocks and found that the distribution of foreshock-mainshock pairs had similar spatial and temporal distribution as mainshock-aftershocks. We set up a model for the occurrence of aftershocks in a chosen time and magnitude interval that can easily be incorporated into the existing model for short-term earthquake probabilities (STEP) in California. The model is ready to be incorporated in a model testing centre that will soon start for California.

Overall the objectives of the project have been achieved.