Figure S1. Complete data records for (A) event 20101103, (B) event 20081217, and (C)
event 20110519. The selected representative records for the whole array are displayed in
Fig. 1.
Figure S2. Comparison between the P data and synthetics for the model proposed by
Savage (2012).
Figure S3. Sensitive test for models with different slab heterogeneities characterized by
different ax, az, and σp in von Kármán distribution function. The numbers on top of
synthetics frames are width of the slab, ax, az, and σp.
Figure S4. Test for the attenuation effects. Comparing to the model without considering
attenuation (A), an attenuated model (B) shows significant amplitude drop, especially at
the smaller distance while rays sample the low-Q mantle wedge. (C) shows the same
seismograms in (B) but with amplitude amplified by 2. Note the waveforms in (C) are
almost identical as those in (A) at distance larger than 0.6°. The red dash line in (A)
displays the LVL trapped high frequency arrivals, which is not shown for the attenuated
model.
Figure S5. Comparison between the synthetics without and with attenuation. The
attenuation is added by applying t* correction assuming Q = 700.
Figure S6 Spectrograms for the data and synthetics for models with different slab
heterogeneities characterized by different ax, az, and σp.
Figure S7. Sensitive test for models with different extended depth of the LVL.
Figure S8. Sensitive test for models with different width of the slab.
Figure S9. Sensitive test for models with different slab dipping angle.
20101103
15 sec
Figure S1
(A)
Figure S1
(B)
20081217
15 sec
Figure S1
(C)
20110519
15 sec
0.8
1.2
Dis
tanc
e (d
eg)
0 5 10
Time (s)
0 5 10
Time (s)
0.2289
0.8
1.2
Dis
tanc
e (d
eg)
(A) Velocity envelope
(B) Velocity
Data
Figure S2
0.8
1.2
Dis
tanc
e (d
eg)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
0.8
1.2
Dis
tanc
e (d
eg)
(A) Velocity envelope
(B) Velocity
Data 80km/10/0.5/2.5 80km/10/1/2.5 80km/5/1/2.5
Figure S3
0.8
1.2
Dis
tanc
e (d
eg)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
0.8
1.2
Dis
tanc
e (d
eg)
(A) Velocity envelope
(B) Velocity
Data 80km/5/0.5/2.5 80km/20/1/2.5 80km/2.5/2.5/2.5
Figure S3continued
0.0
0.4
0.8
1.2
Dis
tanc
e(d
eg)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
Without attenuation With attenuation With attenuationAmplitude × 2
(A) (B) (C)
Figure S4
0.8
1.2
Dis
tanc
e (d
eg)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
0.8
1.2
Dis
tanc
e (d
eg)
(A) Velocity envelope
(B) Velocity
Data 80km/10/0.5/2.5no Q
80km/10/0.5/2.5Q = 700
Figure S5
0
5
10
Fre
quen
cy (
Hz)
0
5
10
Fre
quen
cy (
Hz)
36 38 40 42 44 46 48 50 52 54
Time (sec)
36 38 40 42 44 46 48 50 52 54
Time (sec)
80km/10/0.5/2.5
80km/10/0.5/2 80km/10/0.5/3
80km/20/1/2.5 80km/2.5/2.5/2.5
0.0 0.2 0.4 0.6 0.8
Normalized Amplitude
Figure S6
0
5
10
Fre
quen
cy (
Hz)
Data
0.8
1.2
Dis
tanc
e (d
eg)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
0.8
1.2
Dis
tanc
e (d
eg)
(A) Velocity envelope
(B) Velocity
Data 80km/10/0.5/2.5LVL to 100km
80km/10/0.5/2.5LVL to 300 km
Figure S7
0.8
1.2
Dis
tanc
e (d
eg)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
0.8
1.2
Dis
tanc
e (d
eg)
(A) Velocity envelope
(B) Velocity
Data 80km/10/0.5/2.5 50km/10/0.5/2 110km/10/0.5/3
Figure S8
0.8
1.2
Dis
tanc
e (d
eg)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
0 5 10
Time (s)
0.8
1.2
Dis
tanc
e (d
eg)
(A) Velocity envelope
(B) Velocity
Data 80km/10/0.5/2.5Dip angle = 75°
50km/10/0.5/2.5Dip angle = 70°
110km/10/0.5/2.5Dip angle = 80°
Figure S9