1

Geophysical Research Letters

Supporting Information for

The 2 March 2016 Wharton Basin MW 7.8 earthquake: High stress drop north-south strike-slip rupture in the

diffuse oceanic deformation zone between the Indian and Australian Plates

Thorne Lay1, Lingling Ye2, Charles J. Ammon3, Audrey Dunham3, Keith D. Koper4

1Department of Earth and Planetary Sciences, University of California Santa Cruz, Santa Cruz, California, 2Seismological Laboratory, California Institute of Technology, Pasadena, California, USA, 3Department of Geosciences, Pennsylvania State University, University Park, Pennsylvania, USA, 4Department of Geology and Geophysics, University of Utah, Salt Lake City, Utah, USA

Correspondence to: Thorne Lay, tlay@ucsc.edu Contents of this file Figures S1 to S8

Table S1

Animations M1 to M2

Introduction

Supporting information includes 8 figures, 1 table, and 2 animations.

1

Figure S1. Maps of seismicity from 1976 to 2016 from the NEIC catalog (left) and GCMT solutions (right) for the deformation zone between the Indian and Australian plates and the Sumatran subduction zone. Symbols are color-coded for source depth, with radius scaled proportional to MW. Note the widespread intraplate seismicity, comprised primarily of strike slip and thrusting events.

1

Figure S2. The array response functions [Xu et al., 2009] of the European and Australia station configurations for a P wave with a period of 1 s. Also known as point spread functions, these images show the distorting effects of the finite and discrete wavefield sampling achieved by the arrays. The response of an ideal array would be a 2D delta function at the epicenter. The smearing and sidelobes are mitigated by Nth root stacking, but are not completely eliminated. In addition to the array response, the distance range of the network influences the streaking effects in the back-projections, as apparent in Movie S1.

2

Figure S3. Broadband (50 to 200 s period) surface wave peak amplitudes for waveforms equalized to a propagation distance of 90° for the 2 March 2016 mainshock. Love wave (G1) and vertical component Rayleigh wave (R1) amplitudes are shown in rose diagrams (left) and as functions of azimuth (right). The symbols are color-coded by epicentral distance of the recording prior to distance equalization. The solid lines indicate the azimuthal radiation patterns expected for the double-couple focal mechanism of the mainshock with arbitrary amplitude scales.

3

Figure S4. Love wave (G1) relative source time functions (RSTFs) obtained by iterative time-domain deconvolution with positivity constraint of the 2 March 2016 mainshock recordings by the corresponding station recordings for the 3 March 2016 EGF event (Table 1). A Gaussian filter with width parameter of 0.2 is applied to each deconvolution. The data are aligned with respect to directivity parameter, Γ = cos(θ–θr)/c, where q is the station azimuth, θr is the rupture direction, and c is a reference phase velocity (assumed to be 4.0 km/s), defined for the two possible fault plane orientations with rupture azimuths of (a) 5° and (b) 95°. The data are aligned on the hypocentral reference time for the mainshock. Reference lines at 30 s duration are provided to help assess the duration variability.

4

Figure S5. SH body wave relative source time functions (RSTFs) obtained by iterative time-domain deconvolution with positivity constraint of the 2 March 2016 mainshock recordings by the corresponding station recordings for the 3 March 2016 EGF event (Table 1). A Gaussian filter with width parameter of 0.5 is applied to each deconvolution. The data are aligned with respect to directivity parameter, Γ = cos(θ–θr)/c, where q is the station azimuth, θr is the rupture direction, and c is a reference phase velocity (assumed to be 10.0 km/s), defined for the two possible fault plane orientations with rupture azimuths of (a) 5° and (b) 95°. The data are aligned on the hypocentral reference time for the mainshock.

5

Figure S6. Observed SH body wave relative source time functions (RSTFs) at different azimuths obtained by deconvolving the corresponding EGF signals are compared with predicted source time functions from the finite-fault source models that use varying rupture expansion speeds from 1.5 to 3.0 km/s and a strike of 5°. The overall RSTF duration is less than 30 s at all azimuths, but the high apparent velocity (~10 km/s) of the SH waves results in little sensitivity to rupture speed and spatial extent.

6

Figure S7. Comparison of observed (black lines) and computed (red lines) P-wave and SH-wave broadband ground displacement waveforms for the finite-fault model in Figure 5 with strike of 5° and rupture speed of 2.0 km/s. The azimuth and distance of each station are shown below the station name. The blue numbers indicate the peak-to-peak amplitude of the observed signals in microns.

7

Figure S8. The stress change distribution on the fault model shown in Figure 5. The stress change at the mid-point of each subfault is computed for the entire distribution of slip over the model surface. The shear stress amplitude and direction at each subfault are sown by the color scale and length and angle the vectors in each subfault. The stress drops calculated by trimming off those subfaults that have a seismic moment less than 15% of the peak subfault moment and using the remaining average slip and residual fault area, Δσ0.15 is 15 MPa. The slip-weighted stress drop measured from the variable stress change distribution, ΔσE is 20 MPa.

8

Table S1 Source velocity model used in finite-fault inversions

P-Velocity km/s

S-Velocity km/s

Density kg/m3

Thickness km

1.5 0 1000 5.0 5.0 2.5 2600 1.7 6.6 3.65 2900 2.3 7.1 3.9 3050 2.5 7.7 4.5 3300 halfspace

9

Animations Movie M1. Back-projections of 0.5-2.0 Hz P wave signals from large regional networks of broadband stations in Europe and Australia. Movie M2. Rupture animation for slip history (top) cumulative slip (bottom) for the preferred finite-fault rupture model with a strike of 5° and rupture expansion speed of 2.0 km/s.