BRIEF ARTICLE THE AUTHOR

Transcription

1 BRIEF ARTICLE THE AUTHOR 1

2 2 THE AUTHOR S Pd K Pd S mantle OC IC CMB Figure 1 Figure 1. Illustration of the SPdKS / SKPdS ray-paths with sub-segments labeled. SPdKS is an SKS that intersects the source-side CMB at the ScP critical angle and propagates along the mantle-side of the CMB as a diffracted P-wave (Pdiff) for some distance. SKPdS is the same phenomena on the receiver-side CMB. Star shows earthquake source and triangle shows receiver location. IC = inner core, OC = outer core, and CMB = coremantle boundary.

3 BRIEF ARTICLE 3 Grid Configuration with 1024 vertical nodes Vertical (dr) Horizontal (dθ) Surface Model Radius (km) CMB km CMB CMB km 2000 dr min = 2.5 km dr max = 8 km ICB Grid Spacing (km) Figure 2. Radial r and lateral θ grid node spacing as a function of model radius. Minimum radial grid spacing of 2.5 km is maintained for 100 km above and below the CMB, and then gradually increased with a cosine Figure 4-4: Radial ( r) and lateral ( θ) grid node spacing as a function of model shape to 8 km for the rest of the grid. Lateral grid spacing is constant radius. Minimum angular radial measurement grid spacing and therefore of 2.5varies km is linearly maintained kilometers for 100 as akm above and below the CMB, functionand of radius. then gradually increased with a cosine shape to 8 km for the rest of the grid. Lateral grid spacing is constant in angular measurement and therefore varies linearly in kilometers as a function of radius. θ at the CMB is 5.2 km. ICB = inner core boundary. 122

4 4 THE AUTHOR 108 (a) (b) (c) DISTANCE (deg) SKS SpdKS SKIKS + SKiKS sppp 128 sskp T (sec) -3.4* D Figure 3. Radial component of displacement synthesized for a radial vector point source at 500 km depth synthesized by the (a) pseudospectral method with no earth attenuation, (b) by the pseudospectral method with no S attenuation but the P attenuation model of PREM doubled, and (c) by an asymptotically approximate full wave method with no attenuation. PREM predicted ray theoretical travel times of sskp, SKIKS, and SKiKS are shown by solid lines; predicted earliest arrival times of diffracted SPdKS and the caustic diffraction of sskp by dashed lines; and sppp and pppp by a dash-dot lines.

5 BRIEF ARTICLE D incep = 0.65 D V P D incep ( ) D V P (%) Figure 2 Figure 4. SPdKS Pdiff inception epicentral distance, incep, plotted against P-velocity perturbation at the base of the mantle, V P for PREM (star) and 1-D ULVZ models with heights between 5 and 40 km, V P between 2.5 and 10%, and V P : V S velocity ratios of 1:1 (triangles) and 1:3 (circles).

6 6 THE AUTHOR a b c d e f Figure 3 Figure 5. Illustrations of various ultra-low velocity zone model configurations. Dark gray line shows SPdKS ray path on the source side of the CMB. Light shaded patch shows approximate ULVZ geometry, with exaggerated vertical dimensions. (a) 1-D ULVZ models. (b) Source-sided ULVZ models. (c) Receiver-side ULVZ models. Finite width ULVZ models centered on (d) the Pdiff inception point with varying widths; (e) the Pdiff exit point into the core; and (f) the middle portion of the P diffracted path.

7 BRIEF ARTICLE 7 a) PREM b) 40 km 1-D ULVZ c) 40 km One-sided ULVZs D ( ) time (s) s / * D ( ) time (s) s / * D ( ) time (s) s / * D ( ) Figure 4 Figure 6. Synthetic SV displacement seismograms produced by running the pseudospectral seismic wave propagation code on a) PREM, b) the 1-D, 40 km-thick, 10% P and 30% S-velocity reduction ULVZ model (Figure 5 a), and c) the 40 km-thick, source-side ULVZ velocity model (Figure 5 b). The solid blue line shows the PREM SKS arrival and the solid red line shows the PREM SPdKS arrival. On (b) the solid green line shows the SPdKS picks for that model. On (c), dashed red lines show the SPdKS picks for the one-sided model and the solid green line repeats the SPdKS for the equivalent 1-D model. Cyan, yellow, and magenta dashed lines show the predicted travel times of sppp, PPPP, and pppp, respectively.

8 8 THE AUTHOR mean SPdKS delay (s) 8 dt = 0.02 h*dvp ULVZ height (h/km) * dvp (%) Figure 5 Figure 7. Relationship between ULVZ strength and the mean SPdKS delay relative to the SPdKS travel time in PREM, dt. The ULVZ strength is parameterized as the thickness of the ULVZ, h, times the P-velocity perturbation at the base of the ULVZ, V P. For the 40 km-thick full gradient model the effective thickness is taken to be 20 km. For the 40 km-thick half gradient model, the effective thickness is taken to be 30 km. Circles show mean SPdKS delays for all 1-D ULVZ models and line shows the linear fit to those points given by the equation for dt.

9 BRIEF ARTICLE 9 a) 20 km Mid 5%P 15%S b) 20 km Mid 10%P 30%S D ( ) 115 D ( ) time (s) s/ * D time (s) s/ * D Figure 6 Figure 8. Synthetic seismograms for finite width middle models (Figure 5e). Blue line shows SKS picks from PREM pseudospectral synthetic seismograms. Solid red line shows PREM SPdKS picks. Dashed red line shows SPdKS picks for this model. Green lines show SPdKS picks for equivalent one-sided ULVZ model. Dashed cyan, yellow, and magenta lines show travel time predictions for sppp, PPPP, and pppp, respectively.

10 10 THE AUTHOR SPdKS ULVZ - SPdKS PREM travel time (s) ULVZ Width 1-D 1-sided 1920 km 960 km 840 km 720 km 600 km 480 km 240 km km 60 km 30 km 20 km 10 km ULVZ Strength (height km x dvp %) Figure 9. SPdKS delay versus ULVZ strength (parameterized as ULVZ height multiplied by the ULVZ P-velocity perturbation) for various ULVZ widths.

11 BRIEF ARTICLE 11 Slope of SPdKS delay vs ULVZ strength Width of ULVZ (km) Figure 10. Slopes of 1-D fits to inception model results from Figure 9 versus ULVZ width. This slope is a reasonable proxy for SPdKS sensitivity to ULVZs of a given finite width centered on the Pdiff inception point.

12 12 THE AUTHOR Event _ D Modeling Event _ MM17 PREM-like 116 MM17 D ( ) MM12 MM11 MM10 MM09 MM08 MM06 MM05 MM02 MM travel time relative to SKS (s) PREM CRZ / ULVZ CRZ / CMTZ / ULVZ PREM-like thin ULVZ PREM-like PREM PREM thin ULVZ MM12 MM11 MM10 MM09 MM08 MM06 MM05 MM02 MM travel time relative to SKS (s) Figure 9 Figure 11. Data section for August 14, 1995 earthquake with a moment magnitude of 6.3 and a depth of 126 km, located near Papua New Guinea and recorded at ten stations in the MoMa array (Thorne and Garnero, 2004). Left section shows data aligned on SKS picks, labeled with bestfitting 1-D CMB region model(s) for each individual trace (Thorne and Garnero, 2004). ULVZ = ultra-low velocity zone, CRZ = core-rigidity zone (small finite rigidity at top of outer core), CMTZ = core-mantle transition zone (linear gradient between lower mantle properties and upper outer core properties). PREM-like results indicates that the best fitting models are PREM and thin (10 km) ULVZs with density perturbations but no velocity perturbations. Right section shows same data but with SKS picks (blue), SPdKS picks (green and magenta), and PREM predicted SPdKS relative to observed SKS arrivals (red line).

13 BRIEF ARTICLE N a) Grand [2002] 30 N 0 30 S E 150 E W W 90 W 60 W Shear Velocity Tomography at 2800 km 20 N b) S20RTS [2000] c) SAW24B16 [2000 d) S362D1 [2001] e) SBW4L18 [2000] E 160 E E 160 E E 160 E E 160 E 180 Figure 10 Figure 12. a) Location of the source (yellow star) and receivers (white triangles) for the seismograms shown in Figure 11, as well as the expected SPdKS source-side (cyan lines) and receiver-side (magenta lines) sampling regions on the CMB, plotted on the 2800-km deep slice of the 3-D mantle shear wave model of (Grand, 2002). The same source position and source-side Pdiff paths are also plotted on the 2800-km deep slice of the 3-D mantle shear wave models of b) Megnin a nd R omanowicz 2 000, c)masters e tal 2 000, d)ritsema a nd v anheijst 2 000, ande)gu e tal 2 001, allcompi //mahi.ucsd.edu/gabi/rem.html).