SCEC Award Number 11026 View PDF
Proposal Category Collaborative Proposal (Integration and Theory)
Proposal Title Correlation between seismic clustering properties and regional physical conditions
Investigator(s)
Name Organization
Ilya Zaliapin University of Nevada, Reno Yehuda Ben-Zion University of Southern California
Other Participants two graduate students
SCEC Priorities A9, A2, A4 SCEC Groups Seismology, FARM, SHRA
Report Due Date 02/29/2012 Date Report Submitted N/A
Project Abstract
 (1) Statistical methodology and software for cluster analysis of seismicity. A a non-parametric method for identifying statistically significant earthquake clusters is developed; the method is based on a bimodal distribution of earthquake distance in the space-time-magnitude domain. Accordingly, the cluster identification is not based on apriori cluster parameters or ad-hoc assumptions about the seismic flow.

(2) Classification of earthquake clusters. The existence of two dominant types of earthquake clusters in southern California is established: aftershock-dominated sequences and swarms. The identification and quantitative analysis of the cluster types is done via the network representation of clusters. The topological cluster properties are highly coupled with a dozen of conventional metric cluster statistics (duration, area, fore/aftershock number, etc.).

(3) Coupling between cluster type and physical properties of the lithosphere. The two dominant types of earthquake clusters have distinct preferred spatial locations. Swarms tend to occur in the regions with low level of effective viscosity. The aftershock-dominated sequences tend to occur in the regions with high level of effective viscosity.

(4) Asymmetric distribution of early aftershocks on bimaterial faults. The relations between spatial symmetry properties of earthquake patterns along faults in California (CA) and local velocity structure images is analyzed to test the hypothesis that ruptures on bimaterial faults have statistically preferred propagation directions. We have found strong asymmetric patterns in early-time spatially-close aftershocks along large faults with prominent bimaterial interfaces, with enhanced activity in the directions predicted for the local velocity contrasts, and absence of significant asymmetry along most other faults.
Intellectual Merit The project develops an original methodology for cluster analysis of seismicity that adapts to the regional properties (seismic level, magnitude distribution, etc.). Specific applications include declustering and identification of foreshock-mainshock-aftershock sequences and swarms. Ability to perform such analyzes in objective fashion, without ad-hoc assumptions and predefined parameters is important for various SCEC research objectives, including developing EQ forecast algorithms.

The project results establish quantitative connections between various cluster properties of seismicity and the properties of the lithosphere (heat flow, effective viscosity). The project results contribute to understanding of the behavior of bi-material faults. Overall, the results provide important modeling constraints and directly contribute to improving seismic hazard assessment.
Broader Impacts Broader impact:
Our studies have developed and applied improved statistical methods for deriving high-resolution information from earthquake catalogs that are related to specific properties of fault zones and the crust. The results provide important information for developing refined seismic hazard assessments, and constraining the physics of earthquake ruptures and dynamics of seismicity. The PIs will organize a special session at the 2012 Annual SSA meeting related to the research initiated in this project.

Student involvement:
Jennifer Bautista (BSc student at UNR -- Undergraduate Honors Thesis); Andrew Hicks (MSc student at UNR -- MSc thesis); Maggie Michalowski (MSc student at UNR); Zachary Ross (PhD student at USC)
Exemplary Figure Figure 3: Spatial distribution of different types of earthquake families. The figure shows the average leaf depth ‹d› for N = 452 clusters with maximal magnitude 4 ≤ m < 6 from the relocated catalog of Hauksson et al. (2012). Recall that the high (low) values of the average leaf depth ‹d› correspond to swarms (aftershock-dominated sequences). The high-depth families (swarms) are concentrated within the regions with low effective level of viscosity – Coso and Salton trough. The high-viscosity regions – e.g., Mojave, San Bernardino, Ventura – are characterized by the low-depth families (aftershock-dominated sequences).