2019 SCEC Report

Developing a Geologic Framework Model and Enhancements to the CVM

in the Unified Structural Representation Framework

SCEC Award 19104

Principal Investigators: John H. Shaw

Professor of Structural & Economic Geology Andreas Plesch (co-Investigator)

Senior Research Scientist Harvard University

Dept. of Earth and Planetary Sciences 20 Oxford St., Cambridge, MA 02138

shaw@eps.harvard.edu // (617) 495-8008

Thomas Jordan University Professor and W. M. Keck Professor of Earth Sciences

Southern California Earthquake Center University of Southern California

Los Angeles, CA 90089-0740 Phone: 213-821-1237; Fax: 213-740-0011

Proposal Categories: A: Data Gathering and Products; Collaborative Proposals Science Objectives: P3a, P1a Duration: 1 February 2019 to 31 January 2020

Summary

We developed an initial version of the SCEC Geologic Framework Model (GFM) that describes crust and upper mantle structure in Southern California. The model consists of crust, upper mantle and asthenosphere layers with the crustal layer being subdivided into 23 lithotectonic units. The model grid houses cells parametrized with a region number linked to a name and lithology type, as well as temperature with placeholders for a variety of other geophysical properties. The GFM is intended to support SCEC efforts to develop a Community Rheology Model (CRM) and support a range of other science activities.

We also applied a k-means algorithm to the SCEC Community Velocity Models (CVMs) to define various regionalizations that we compared with geologic and other geophysical constraints on structure of the crust. This analysis suggests robust, low-order regionalizations are geologically valid for both versions of the SCEC CVM’s (CVM-S 4.26 and CVM-H 15.1), while higher order regionalizations for the CVM-S show greater consistency with more detailed aspects of inferred crustal composition (see Eymold & Jordan, 2019). The analysis highlighted deficiencies in the models, as well as identified ways that the method may be used to improve future iterations of the Geologic Framework Model (GFM).

Introduction

Our efforts this past year focused on: 1) developing an initial version of the SCEC Geologic Framework Model (GFM) that describes crust and upper mantle structure in southern California (Fig. 1), and 2) regionalization of the crust based on cluster analysis of its wave-speed structure to evaluate ways to improve the GFM using geophysical constraints. The GFM is designed to aid in the development of a Community Rheologic Model (CRM) by providing lithologic descriptions and thermal properties that can be used to assign reasonable constitutive laws to model deformation in the region.

Results

Our GFM is based on definition of 23 lithotectonic units (Hearn et al., 2018; Thatcher et al., 2019) separated by major faults extracted from the Community Fault Model (CFM, Plesch et al., 2007) or by contrasts in basement lithology and tectonic affinity where such contrasts are seismically quiescent. In addition, several horizon interfaces are represented, including topography/bathymetry, the base of sedimentary basins, the base of the seismogenic crust, the Moho, and the lithosphere-asthenosphere boundary (LAB).

In order to achieve maximum compatibility with the CFM, we first identified all CFM fault representations which define the lithologic block boundaries of Thatcher et al., (2018) (Fig. 1). From surface traces we constructed model boundaries that smoothly fit the detailed CFM fault representations. This procedure was applied to several boundaries that have complex geometries, including the northern and southern boundary of the Western Transverse Ranges, and the southern boundary of the San Gabriel block. Other boundaries are currently modeled with a vertical orientation.

All horizon interfaces were modeled from data and extended were necessary to the full model domain. The base of sedimentary basins, the base of the seismogenic crust, and the Moho were adopted from the CFM and CVM-H (Tape et al., 2010; Shaw et al., 2015). The LAB was constructed from data provided by Lekic et al. (2011) and smoothly extrapolated to the full model domain. From these bounding surfaces, a gridded volume was subdivided into regions. The grid is

10km x 10km x 1km and has about 900K cells. The GFM currently has three layers: crust, upper mantle, and asthenosphere. The crustal layer contains all 23 lithotectonic units. Each cell is parameterized with a region identification number from 1 to 41 which is mapped to region name and lithology type (Fig. 1). Each cell also has a temperature value derived from a simple 1D geotherm as an example material property, which will be updated with values from the Community Thermal Model (Thatcher et al., 2019) when it is available.

Figure 1: Left: Perspective view of Southern California with GFM unit boundaries (yellow) and horizons (color contoured Moho and reflective LAB). Black outlines show CFM faults corresponding to unit boundaries. Right: Identical perspective view of GFM grid showing crustal, lithotectonic units (regions of color scale), the upper mantle layer (dotted), and the asthenosphere layer (solid green). Three crustal units are dotted to provide a sense of the volumetric nature of the grid. 3x vertical exaggeration.

We anticipate that the new GFM will be evaluated by the CXM working group to define areas for refinement. This may include incorporating additional lithologic regions, as well as representing sedimentary basins. The GFM will be made available to the SCEC Community through and extension of the UCVM framework that is used to support the SCEC Community Velocity Models (CVMs). Furthermore, the unique region ID within the GFM will be to develop associations with database framework that identifies appropriate rheologic properties for various types of deformation modeling. Together, the GFM and this database framework will comprise the Community Rheologic Model (CRM). The main benefit of keeping the database framework as a distinct element from the GFM will be that ongoing changes in the definition of rheologic properties will not require updates to the GFM.

The USC group worked with the Harvard team to map crustal regions in Southern California that have similar depth variations in seismic velocities by applying cluster analysis to the velocity profiles extracted from the three-dimensional tomographic models CVM-S4.26 and CVM-H15.1 (Fig. 2). The CVM-S4.26 results were recently published in JGR (Eymold & Jordan 2019).

Eymold & Jordan (2019) applied a K-means algorithm to partition the 1.5 million P and S velocity profiles of CVM-S4.26 into K sets that minimized the inter-cluster variance. The regionalizations for K ≤ 10 generate a coherent sequence of structural refinements; each increment of K introduces a new region typically by partitioning a larger region into two smaller regions or by occupying a transition zone between two regions. The results for K ≤ 7 are insensitive to initialization and trimming of the model periphery; nearly identical results are obtained if the P and S velocity profiles are treated separately or jointly.

The regions for K = 7 (Fig. 2, left panels) can be associated with major physiographic provinces and geologic areas with recognized tectonic affinities, including the Continental Borderland, Great

Valley, Salton Trough, and Mojave Desert. The regionalization splits the Sierra Nevada and Peninsular Range batholiths into the western and eastern zones consistent with geological, geochemical, and potential-field mapping. Three of the regions define a geographic domain comprising almost all of the upper crust derived from continental lithosphere.

Well-resolved regional boundaries of the CVM-S4.26 regionalization coincide with major faults, topographic fronts, and/or geochemical transitions mapped at the surface [see Eymold & Jordan (2019) for a description of 13 examples]. The consistent alignment of these surface features with deeper transitions in the crustal velocity profiles indicates that regional boundaries are typically narrow, high-angle structures separating regions with characteristic crustal columns that reflect different compositions and tectonic histories.

We applied the same K-means clustering algorithm to CVM-H15.1, and the K = 7 results are compared in (Fig. 2). One complication is that CVM-H15.1 has surface topography, whereas CVM-S4.26 does not. To facilitate an apples-to-apples comparison, we sampled a “topography squashed” version of CVM-H15.1.

Figure 2: Map of K-means regionalizations of CVM-S4.26 (left, from Eymold & Jordan, 2019) and CVM-H 15.1 (right). The mean velocity profiles (a and b) for the regions in each model are shown below. White dashed line on left panel is a nominal resolution contour; i.e., velocity profiles outside this contour are poorly resolved by the CVM-S4.26 inversion.

Several aspects of the CVM-H15.1 results support the geophysical validity of the low-order regionalizations and statistical robustness of the cluster analysis: 1. As in the case of CVM-S4.26, the clustering results are insensitive to the cluster initialization.

Initializations using randomly selected profiles as cluster means converged to the same regionalization.

2. The agreement scores between the a and b regionalizations are consistently good (> 80%) for K ≤ 10, which is higher than the scores obtained by CVM-S4.26. This may be due to an enforced scaling between the P and S velocities rather than an agreement between independent estimates.

3. The regions defined by the clusters are multiply-connected, but each can be associated with plausible combinations of recognized physiographic provinces.

4. The progression of regionalizations with K generates a coherent set of structural refinements.

On the other hand, the two regionalizations show substantial differences. In some cases, they reflect model deficiencies; e.g., the deep sedimentary basins of Great Valley are poorly represented in CVM-H15.1 and don’t incorporate the structural refinements to the Great Valley structure recently released by the Harvard group.

More significant are the differences in the mean regional profiles plotted in the lower panels of Fig. 2. Both models show similar variations in the near-surface profiles and in the depth of the Moho discontinuity. However, in the mid-crust (10-20 km depth), the a and b variations in the CVM-S4.26 profiles are a factor of two or three larger in amplitude than those of CVM-H15.1. This difference may reflect the lack of mid-crustal resolution in the dataset inverted in the tomographic refinement of CVM-H15.1 (Tape et al. 2010).

Our analysis demonstrates that a cluster analysis of velocity profiles provides structural regionalizations of the 3D tomographic models that are useful for model intercomparisons, and they define structural features that can be evaluated using geological and geochemical constraints. Conversely, sorting out which features of these regionalizations are well-defined by the tomographic data should contribute valuable information about how to optimize the parameterization of the GFM.

SCEC Publications

Eymold, W. K., and T. H. Jordan (2019), Tectonic regionalization of the Southern California crust from tomographic cluster analysis, J. Geophys. Res., 124, 11,840-11,865, doi:10.1029/2019JB018423.

References & Publications Eymold, W. K., and T. H. Jordan (2019), Tectonic regionalization of the Southern California crust from tomographic cluster analysis, J. Geophys. Res., 124, 11,840-11,865, doi:10.1029/2019JB018423.

Graves, R., T. H. Jordan, S. Callaghan, E. Deelman, E. Field, G. Juve, C. Kesselman, P. Maechling, G. Mehta, K. Milner, D. Okaya, P. Small, and K. Vahi (2011). CyberShake: A physics-based probabilistic hazard model for Southern California, Pure Appl. Geophys., 167, 367-381.

Hearn, E.H., M.E. Oskin, W.R. Thatcher, G. Hirth, W.M. Behr, and M.R. Legg, Progress toward a Community Rheology Model of Southern California, SCEC Annual Meeting, 2018, P152.

Jordan, T. H., S. Callaghan, R. W. Graves, F.Wang, K. R. Milner, C. A. Goulet, P. J. Maechling, K. B. Olsen, Y. Cui, G. Juve, K. Vahi, J. Yu, E. Deelman, D. Gill (2018), CyberShake Models of Seismic Hazards in Southern and Central California. Proceedings of the 11th National Conference in Earthquake Engineering, Earthquake Engineering Research Institute, Los Angeles, CA.

Lee E.-J., P. Chen, T. H. Jordan, P. B. Maechling, M. A. M. Denolle and G. C. Beroza (2014). Full-3D tomography for crustal structure in Southern California based on the scattering-integral and the adjoint-

wavefield methods, J. Geophys. Res., 119, 6421-6451.

Lekic, V., S. W. French, and K.M. Fischer, (2011) Lithospheric Thinning Beneath Rifted Regions of Southern California, Science, 334, 783-787.

Magistrale, Harold & Day, Steven & Clayton, Robert & Graves, Robert. (2000). The SCEC Southern California Reference Three-Dimensional Seismic Velocity Model Version 2. Bulletin of The Seismological Society of America - BULL SEISMOL SOC AMER. 90. 10.1785/0120000510.

Olsen, K.B., 2000, Site Amplification in the Los Angeles Basin from Three-Dimensional Modeling of Ground Motion. Bulletin of the Seismological Society of America ; 90 (6B): S77–S94. doi: https://doi.org/10.1785/0120000506

Plesch, A., J. H. Shaw, C. Benson, W. A. Bryant, S. Carena, M. Cooke, J. Dolan, G. Fuis, E. Gath, L. Grant, E. Hauksson, T. Jordan, M. Kamerling, M. Legg, S. Lindvall, H. Magistrale, C. Nicholson, N. Niemi, M. Oskin, S. Perry, G. Planansky, T. Rockwell, P. Shearer, C. Sorlien, M. P. Süss, J. Suppe, J. Treiman, and R. Yeats, (2007), Community Fault Model (CFM) for Southern California, Bulletin of the Seismological Society of America, Vol. 97, No. 6, doi: 10.1785/012005021

Shaw, J. H., A. Plesch, C. Tape, M. P. Suess. T. H. Jordan, G. Ely, E. Hauksson. J. Tromp, T. Tanimoto, R. Graves, K. Olsen, C. Nicholson, P. J. Maechling, C. Rivero, P. Lovely, C. M. Brankman, and J. Munster (2015). Unified Structural Representation of the southern California crust and upper mantle, Earth Planet. Sci. Lett., 415, 1-15.

Taborda, Ricardo & Bielak, Jacobo. (2014). Ground-Motion Simulation and Validation of the 2008 Chino Hills, California, Earthquake Using Different Velocity Models. Bulletin of the Seismological Society of America. 104. 1876-1898. 10.1785/0120130266.

Tape, C. , Liu, Q. , Maggi, A. and Tromp, J. (2010), Seismic tomography of the southern California crust based on spectral‐element and adjoint methods. Geophysical Journal International, 180: 433-462. doi:10.1111/j.1365-246X.2009.04429.x

Thatcher, W. R., Hearn, E. H., Oskin, M. E., Montesi, L. G., Hirth, G., Behr, W. M., Plesch, A., & Shaw, J. H. (2019, 08). Preliminary SCEC Community Rheology Model. Poster Presentation at 2019 SCEC Annual Meeting.