Large Fault Zones Activated at Short Time Distance: Clues From Instrumental Seismicity of Italy

Stress transfer models have been developed in simple media in order to easily describe the change of seismicity-rate after a large event. These models have been tested by many authors, also in order to improve time dependent hazard estimates. A more complicated situation arises when the medium is highly heterogeneous, like in the Apennines, Italy. In such cases details in seismicity changes as well as material property variations should be considered. We use Italy as a test case, and in doing so we benefit from the existence of a database of instrumental seismicity for the past 20 years and of a recently published national database of potential earthquake sources. The scope of this work is to look for details in seismicity behavior which are often overlooked when the magnitude is very low. We reanalyzed the database of Italian seismicity (1981-2000) to extract sequences of events that satisfy specific spatial and temporal criteria. The parameters used are 1) distance less than 20-30 km and 2) time distance shorter than 10-20 days. The final database includes 285 sequences. We evaluated possible associations between seismic sequences and known/supposed earthquake sources, as inferred from seismological, geological and geophysical data. We then inspected the database of sequences to select areas in which fault activation occurred at short time distance. For such areas, we also highlighted sequences of large historical events. We present examples from several Italian regions whose faults have been activated at short time distance by low magnitude instrumental sequences and/or by large magnitude historical sequences. We observed that small magnitude seismicity occurs at the edges of a subsequent large rupture for several sources associated with large historical earthquakes. In one case (Umbria-Marche sequence, 1997), this circumstance was investigated using instrumental data exclusively. Although in the highly heterogeneous region of the Apennines stress transfer models have already well explained several cases of segment-to-segment interaction, our observations suggest that at a larger scale (i.e., a region encompassing several segments), the presence of low yield-stress patches and/or the possibility of a different mechanism is required.