Space-time variations of crustal, fault zone, and seismicity structures

Beno Gutenberg made fundamental contributions to knowledge about large-scale earth structures and properties of moderate to large earthquakes using the seismic data available at the time. Data recorded in the last few decades by improved regional networks and dense seismic arrays provide opportunities for resolving fine details of subsurface rocks and seismicity in 4D. I review such results based primarily on recent data from Southern California. The discussed topics include multi-scale seismic imaging of the crust and fault zones, monitoring temporal changes of seismic velocities, and tracking localization of rock damage and low magnitude seismicity before large earthquakes.

The seismic imaging results reveal hierarchical rock damage structures around large fault zones with intense core damage zones and bimaterial interfaces. The fault damage zones follow overall a flower-shape structure, with significant damage in the top few km that decreases in amplitude and width with depth, and they tend to be offset from the surface trace to the side with higher seismic velocity at depth. The top 100-300 m section has generally extreme seismic properties (very low Vp, Vs, Q values; very high Vp/Vs ratios), which make it highly susceptible to failure and temporal changes. Large earthquakes produce changes of seismic velocities that decay with distance from the rupture zones, but remain significant on a regional scale in the shallow crust. The co-seismic velocity changes are followed by log(t) recovery, and can be very large (e.g. >30%) in the top 100-300 m. Appreciable changes of shallow materials are also generated by atmospheric and other non-tectonic loadings on various timescales.

The results on localization processes are based on (i) estimated production of rock damage by background seismicity, (ii) spatial localization of background events within damaged areas, and (iii) progressive coalescence of individual earthquakes into clusters. The analyses reveal generation of earthquake-induced rock damage on a decadal timescale around eventual rupture zones of large earthquakes, and progressive localization of background seismicity 2-3 yrs before M > 7 earthquakes in Southern and Baja California and M > 7.5 events in Alaska. This localization phase is followed by coalescence of earthquakes into growing clusters that precede the mainshocks. Corresponding analyses around the 2004 M6 Parkfield earthquake in the creeping section of the San Andreas fault, which is essentially always localized, show opposite tendencies to those involving faults that are locked in the interseismic periods.

Continuing efforts in these topics include merging local high-resolution imaging results within regional models, monitoring temporal changes of properties at seismogenic depth, and including geodetic data and insights from laboratory experiments in the localization analyses.