ABSTRACT 
 
 
 
 
Title of Dissertation: THE SEDIMENT AND CRUSTAL 
STRUCTURE OF THE SOUTHEASTERN 
UNITED STATES 
  
 Erin Cunningham, Doctor of Philosophy, 2019 
  
Dissertation Directed By: Professor Vedran Leki?, Department of Geology  
 
 
 
The Southeastern United States (SEUS) preserves a detailed geologic record of the 
continental collision and rifting that has shaped it over billions of years. Currently the 
SEUS lies on a passive margin far away from ongoing tectonics, yet retains the ability 
to produce damaging earthquakes. It has long been suspected that the current 
seismicity in the SEUS is related to zones of weakness inherited from  past structural 
boundaries; however, the mechanisms that dictate the size and location remain poorly 
understood and the ground shaking hazard posed by widespread sedimentary basins 
remains poorly quantified. P-to-S converted waves, analyzed through the receiver 
function technique, enable detailed imaging of lithospheric structure, but suffer from 
contamination in sedimented regions. Therefore, I implement a method for 
constraining average crustal thickness and seismic velocities while correcting for the 
effects of sediments. I then use high-frequency constraints on sediment structure to 
construct a reference velocity model for the Atlantic Coastal Plain, discussing its 
implications for seismic hazard. Finally, using Sp and sediment-corrected Ps receiver 
functions, I image the structure of the crust and mantle lithosphere across the SEUS, 
applying identifying dipping crustal structures that appear to be associate with active 
seismic zones. I discuss these structures in the context of the region?s tectonic past 
and future seismic hazards 
 
 
 
 
 
   
 
 
 
 
 
 
 
 
 
THE SEDIMENT AND CRUSTAL STRUCTURE OF THE SOUTHEASTERN 
UNITED STATES 
 
 
 
by 
 
 
Erin Cunningham Your Full Name as it appears in University Records 
 
 
 
 
 
Dissertation submitted to the Faculty of the Graduate School of the  
University of Maryland, College Park, in partial fulfillment 
of the requirements for the degree of 
Doctor of Philosophy  
2019 
 
 
 
 
 
 
 
 
 
Advisory Committee: 
Professor Vedran Leki?, Chair 
Dr. Nicholas Schmerr  
Dr. Mong Han Huang 
Dr. Lara Wagner 
Dr. Brian Phillips 
 
 
 
   
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
? Copyright by 
Erin Cunningham 
2019 
 
 
 
 
 
 
 
 
 
 
  
 
   
 
 
Table of Contents 
 
 
Table of Contents ii	
 ?
List of Tables iv	
 ?
List of Figures v	
 ?
Chapter 1: Introduction to the Lithosphere and the Southeastern United States 1	
 ?
1.1	
 ?Overview	
 ?of	
 ?Chapter	
 ?2	
 ? 3	
 ?
1.2	
 ?Overview	
 ?of	
 ?Chapter	
 ?3	
 ? 4	
 ?
1.3	
 ?Overview	
 ?of	
 ?Chapter	
 ?4	
 ? 5	
 ?
1.4	
 ?Technical	
 ?Background:	
 ?Receiver	
 ?Function	
 ?Methodology	
 ? 6	
 ?
1.4.1	
 ?Phase	
 ?picking	
 ?and	
 ?P-??SV-??SH	
 ?rotation	
 ? 12	
 ?
1.4.2	
 ?Deconvolution	
 ? 14	
 ?
1.4.3	
 ?Frequency	
 ?Domain	
 ?Deconvolution	
 ? 16	
 ?
1.4.4	
 ?Time	
 ?Domain	
 ?Deconvolution	
 ? 19	
 ?
Chapter 2: Constraining Crustal Structure in the presence of Sediment: A 
Multiple Converted Wave Approach 21	
 ?
2.1	
 ?Introduction	
 ?and	
 ?Rationale	
 ?for	
 ?New	
 ?Methodology	
 ? 21	
 ?
2.2	
 ?SRTC	
 ?H-??	
 ??	
 ?-??VP	
 ?Stacking	
 ? 24	
 ?
2.2.1	
 ?Ps	
 ?receiver	
 ?functions	
 ?in	
 ?traditional	
 ?H-???-??VP	
 ?stacks	
 ? 25	
 ?
2.2.2	
 ?SRTC	
 ?Ps	
 ?receiver	
 ?function	
 ?H-???-??VP	
 ?stacks	
 ? 28	
 ?
2.2.3	
 ?Determining	
 ?when	
 ?to	
 ?correct	
 ?for	
 ?sedimentary	
 ?layers	
 ? 35	
 ?
2.2.4	
 ?Sp	
 ?receiver	
 ?functions	
 ?in	
 ?H-???-??VP	
 ?stacks	
 ? 36	
 ?
2.2.5	
 ?SsPmp	
 ?receiver	
 ?functions	
 ?in	
 ?H-??	
 ??	
 ?-??VP	
 ?stacks	
 ? 39	
 ?
2.2.6	
 ?Calculating	
 ?combined	
 ?SRTC	
 ?H-??	
 ??	
 ??VP	
 ?stacks	
 ?with	
 ?error	
 ? 41	
 ?
2.3	
 ?Receiver	
 ?Function	
 ?Data	
 ?for	
 ?H-??	
 ??	
 ??VP	
 ?Stacks	
 ? 43	
 ?
2.4	
 ?Results	
 ?and	
 ?Observations	
 ?from	
 ?the	
 ?SRTC	
 ?H-??	
 ??	
 ??VP	
 ?Method	
 ? 46	
 ?
2.4.1	
 ?SRTC	
 ?H-???-??VP	
 ?stacking	
 ?with	
 ?synthetic	
 ?waveforms	
 ? 48	
 ?
2.4.2	
 ?TA	
 ?station	
 ?B30A:	
 ?eastern	
 ?Williston	
 ?Basin	
 ? 51	
 ?
2.4.3	
 ?TA	
 ?station	
 ?U60A:	
 ?Atlantic	
 ?Coastal	
 ?Plain	
 ? 55	
 ?
2.4.4	
 ?TA	
 ?statin	
 ?448A:	
 ?Mississippi	
 ?Embayment	
 ? 59	
 ?
2.5	
 ?Discussion	
 ?of	
 ?H-???-??VP	
 ?Methodology	
 ? 62	
 ?
2.6	
 ?H-???-??VP	
 ?Conclusions	
 ? 67	
 ?
Chapter 3: Constraining Basin Fundamental Frequency and Velocity Profiles of 
the Atlantic Coastal Plain using Receiver Functions 70	
 ?
3.1	
 ?Introduction	
 ?to	
 ?Hazard	
 ?in	
 ?the	
 ?Atlantic	
 ?Coastal	
 ?Plain	
 ? 70	
 ?
3.2	
 ?Receiver	
 ?Function	
 ?Data	
 ?for	
 ?the	
 ?ACP	
 ? 75	
 ?
3.3	
 ?Basin	
 ?Fundamental	
 ?Frequency	
 ? 76	
 ?
3.4	
 ?Basin	
 ?Velocity	
 ?Profiles	
 ? 81	
 ?
3.4.1	
 ?ACP	
 ?Basin	
 ?VS	
 ?Profile	
 ? 81	
 ?
3.4.2	
 ?ACP	
 ?Basin	
 ?VP	
 ?Profile	
 ? 86	
 ?
3.5	
 ?Validation	
 ? 92	
 ?
3.6	
 ?Discussion	
 ?of	
 ?ACP	
 ?results	
 ? 98	
 ?
3.7	
 ?Conclusions	
 ? 101	
 ?
 ii 
 
 
Chapter 4: Crustal and Lithospheric Structure in the Southeastern US using 
Receiver Function Common Conversion Point (CCP) Stacking 102	
 ?
4.1	
 ?Introduction	
 ? 102	
 ?
4.2	
 ?A	
 ?Brief	
 ?History	
 ?of	
 ?SEUS	
 ?Tectonics	
 ? 6	
 ?
4.3	
 ?Receiver	
 ?Function	
 ?Data	
 ?for	
 ?Common	
 ?Conversion	
 ?Point	
 ?(CCP)	
 ?Stacking	
 ? 105	
 ?
4.4	
 ?Common	
 ?Conversion	
 ?Point	
 ?(CCP)	
 ?stacking	
 ? 107	
 ?
4.4.1	
 ?CCP	
 ?stacking	
 ?overview	
 ? 107	
 ?
4.4.2	
 ?Regional	
 ?Velocity	
 ?Model	
 ?with	
 ?SRTC	
 ?H-??	
 ??-??VP	
 ?Stacking	
 ? 111	
 ?
4.4.3	
 ?Sediment	
 ?Removed	
 ?PRF	
 ?CCP	
 ?stacks	
 ?and	
 ?SRF	
 ?CCP	
 ?stacks	
 ? 115	
 ?
4.5	
 ?Results	
 ?and	
 ?Observations	
 ? 118	
 ?
4.5.1	
 ?Crustal	
 ?Thickness	
 ? 119	
 ?
4.5.2	
 ?Relationship	
 ?between	
 ?Structure	
 ?and	
 ?Seismicity	
 ? 128	
 ?
4.6	
 ?Discussion	
 ? 134	
 ?
4.6.1	
 ?The	
 ?Extent	
 ?of	
 ?the	
 ?Grenville	
 ?Front	
 ?in	
 ?the	
 ?SEUS	
 ? 134	
 ?
4.6.2	
 ?Evidence	
 ?of	
 ?the	
 ?Suwanee	
 ?Suture	
 ? 136	
 ?
4.6.3	
 ?Westward	
 ?dipping	
 ??double	
 ?Moho?	
 ? 137	
 ?
4.6.4	
 ?Correlation	
 ?of	
 ?Structure	
 ?to	
 ?SEUS	
 ?Seismicity	
 ? 139	
 ?
4.6.5	
 ?On	
 ?the	
 ?lack	
 ?of	
 ?a	
 ?lithosphere	
 ?asthenosphere	
 ?boundary	
 ?or	
 ?intra-??lithospheric	
 ?
layering	
 ? 140	
 ?
4.7	
 ?Conclusions	
 ? 141	
 ?
Summary 143	
 ?
Appendices 148	
 ?
Appendix A: Table A.1 148	
 ?
Appendix B: CCP Figures for Chapter 4 150	
 ?
Bibliography 163	
 ?
 
 
 
 
 
 
 
 
 
 
 iii 
 
 
List of Tables 
 
Table 2.1-Number of events and slowness range for each seismic station used in this 
study.           40 
 iv 
 
 
List of Figures 
Figure 1.1 - Ideally the displacement recorded at a seismogram (u(t)) is the 
convolution of the source function (s(t)) and the earth response (g(t)). (Figure  
modified from Shearer, 2009) 10	
 ?
Figure 1.2 - Cartoon of converted P-wave energy across a Moho over mantle half 
space if no tilted axes of anisotropy are present.  (Left) converted paths from a  
direct P wave at the Moho (P phases in red and S phases in dotted blue) and  
(right) their relative arrival times in a Ps receiver function.                                 
12	
 ?
Figure 1.3- Cartoon of converted S-wave energy across a MLD over mantle half 
space if no tilted axes of anisotropy are present Left - converted paths from a  
direct S wave at the MLD (P phases in red and S phases in dotted blue) and  
Right - relative arrival times in a Sp receiver function.
 1
2	
 ?
Figure 1.4? Ps RF for station 127A located in Hobbs, New Mexico. The y-axis 
shows  the fraction of the amplitude of direct P arriving phase, the x-axis is in  
time.                                                                                                                     17	
 ?
 Figure 1.5- Fourier transform from time to frequency domain for common functions. 
Arrows indicate the Fourier transform was taken. Top - delta function in the  
time domain yields a flat spanning all frequencies in the frequency domain.  
Bottom - boxcar function in the time domain yields a sync function with  
multiple frequencies (spectral leakage) in frequency domain.                           19	
 ?
Figure 1.6? Time (ITDD in red) and Frequency (IFDD in blue) domain RF 
comparison for station 127A located in Hobbs, New Mexico. IFDD RF was  
filtered at 0.03-1 Hz and appears overall smoother than the ITDD RF.  However, 
it is difficult to determine if the Ps arrival associated with the Moho  occurs at 
3.5 or 6 s. The ITDD RF was created with 300 iterations and appears  to have 
more arriving phases, however, the spectral leakage has been reduced  and the 
peaks narrow. The RF amplitudes for both frequency and time domain  RFs have 
been normalized to 1. 20	
 ?
Figure 2.1- A. Ray-path geometry and B. synthetic receiver functions of P-to-s 
phases converted across the Moho at 30 km depth, from an incident plane  wave 
of horizontal slowness 0.06 s/km. Crustal VP = 6.3 km/s, VS = 3.7 km/s,  while 
mantle VP = 8 km/s and VS = 4.6 km/s. Arrival times of the direct Moho  P-to-s 
conversion and first-order multiples are relative to the arrival time of  the direct 
Pwave.                                                                                                               27	
 ?
Figure 2.2-  A. Ray-path geometry of P-to-s phases converted across the base of a 0.5 
km thick sediment layer, along with the largest-amplitude first- and  second-
order multiples for an incident plane wave of horizontal slowness  0.0789 s/km. 
B. Synthetic receiver functions computed for a model with a 0.5  km thick 
sediment layer and horizontal slowness 0.06 s/km (VP = 2.5 km/s, VS  = 1 km/s), 
crustal VP = 6.2 km/s, VS = 3.5 km/s, a Moho at 35 km depth, and  mantle VP = 8 
 v 
 
 
km/s, VS = 4.6 km/s. Direct conversions are in black, while the  oscillatory 
pattern is attributed to sediment multiples, labeled in red, blue and  purple. 
Orange dots indicate where the expected arrival of phases with more  than 2 P-
wave legs would arrive.                                                                                       29	
 ?
Figure 2.3- Absolute value of the reflection coefficients of PP*PP (left) and SS*SS 
(right) for a range of different sediment velocity variables using equations  from 
Aki and Richards (1980) Because each reverberation will need to be  multiplied 
by an PPPP or SSSS reflection coefficient, the reverberations will  get smaller, 
but the PPPP reverberation will get smaller faster, and SS*SS  reverberations 
will remain visible in the Ps receiver function (See Figure 2.2)                        29	
 ?
Figure 2.4-  The SRTC H-k-VP stacking method may not work when the PPbs and 
Pms phase arrival intersect. (Left) Calculated PPbs arrival time from sediment  
with varying thickness and VS. The VP/VS ratio changes according to the mud  
line equation in Brocher (2005). (Right) the Pms ?PPbs arrival time for a crust  
with VP = 6 km/sand VP/VS = 1.76. Area where the PPbs arrival may interfere  
with the Pms Moho arrival is shown in light blue to red. SRTC stacking may  not 
work for an extremely thin crust of 20 km, if sediment thickness is greater  than 
1.5 km with excessively low VS of 0.5 km/s.                                                      30	
 ?
Figure 2.5-  A. The autocorrelation of the mean synthetic receiver function (gray) and 
the best fit decaying sinusoid (red) used for determining the parameters of  the 
resonance removal filter. The first large amplitude negative peak  corresponds to 
the two-way s travel time in sediment (?t) on the x-axis and  the amplitude of the 
sediment reverberations (r0) on the y-axis. B. Mean  synthetic Ps receiver 
functions before (gray) and after (red) the resonance  removal filter is applied; 
the filter successfully removes the reverberations and  the direct P to s 
conversion across the Moho becomes clear.                                                       32	
 ?
Figure 2.6- Dependence of resonance removal filter on ray parameter for an idealized 
synthetic case. The best-fit reverberation removal filter (red line) calculated  
from the autocorrelation of the mean synthetic receiver function provides a  good 
fit to the autocorrelations of Ps RFs computed across the range of  possible ray 
parameters (black lines). The synthetic model used has a  sediment layer 0.5 km 
thick with a VP = 2.5 km/s and a VS = 1.7 km/s  overlying a 39.5 km crust with 
VP = 6.2 km/s and VS = 3.5 km/s and mantle  with VP = 8 km/s and VS = 4.5 
km/s.                                                                                                                    32	
 ?
Figure 2.7? A. Ray-path geometry of S to p converted phases across the Moho at 30 
km depth calculated for an incident plane wave of 0.14 s/km and the post  
critical P wave reflection across the 30 km deep Moho (SPmp) with incident  
plane wave of 0.1486 s/km. The Ps conversions from the Moho are plotted for  
comparison (gray line). B. Synthetic receiver function of the S-to-p phases  
converted across the Moho. Time is relative to the direct S arrival (positive  
values indicate earlier arrivals). C. Synthetic receiver function of the Hilbert- 
transformed post-critical P phase at the Moho. Time is relative to the direct S  
arrival (positive values indicate later arrivals). Crustal VP = 6.3 km/s, VS = 3.7  
km/s , while mantle VP = 8 km/s and VS = 4.6 km/s.                                          37	
 ?
 vi 
 
 
Figure 2.8? SRTC H-?-VP stacking of synthetic receiver functions, where negative 
values in the stack are clipped. Input model has a 0.5 km thick sedimentary  
layer, VP of 2 km/s, and ? = 2.3, and a Moho at 37 km depth with crustal VP =  
6.3, and ? = 1.76. A. The mean Ps receiver function before (black) and after  
(red) the sediment removal filter is applied. B. Ps H-? stack at assumed VP =  6.3 
shows no clear maxima. C. The Ps H-k stack after the sediment removal  filter 
and time correction is applied shows a clear maximum. D, E, & F. Three  cross-
sections through the maximum of the H-?-VP ?triple stack; black cross  hairs 
denote 1? error bars.                                                                                            42	
 ?
Figure 2.9? H-?-VP stacks generated from synthetic waveforms with increasing noise 
levels. The input model has a sedimentary layer 0.5 km thick, VP of 2.3  km/s 
and ? = 2.3, a Moho at 42.5 km depth with crustal VP = 6.3 km/s, and a ?  of 
1.76. White noise is filtered to have similar frequency content as typically  
encountered noise, with amplitude up to three-quarters of the maximum of the  
daughter waveform (s for Ps; p for Sp and SPmp). At increased noise levels,  we 
are still able to constrain crustal H, ?, and VP within error.                                51	
 ?
Figure 2.10?  (Top) Mean 4 Hz (blue) and 1Hz (black) Ps RF for station B30A 	
 ?
shown with picked arrival time for ?tP and ?t shown in the red and magenta  
circles respectively. The arrival time of the two phases (Pbs and PPbs) likely  
overlap at this station. (Bottom) Autocorrelation from the mean 4Hz (blue)  and 
1Hz (black) RF with picked arrival time for ?t. Autocorrelation of a  random 
subset of individual RFs are shown as gray lines in the background.                 52	
 ?
Figure 2.11? SRTC H-?-VP stacking of B30A receiver functions, where negative 	
 ?
values in the stack are clipped. Station B30A is located on the eastern edge of  
the Williston Basin in Montana. A. The mean Ps receiver function before  
(black) and after (red) the sediment removal filter is applied. B. Ps H-? stack  at 
assumed VP = 6.3 km/s shows a maximum of the stack at H = 28.5 km and  ? = 
1.62. C. The Ps H-k stack after the sediment removal filter and time  correction 
is applied; the stack maximum now corresponds to larger H and ?  values D, E, 
& F Three cross-sections through the maximum of the H-?-VP  triple stack; 
black cross-hairs denote 1? error bars.                                                                54	
 ?
Figure 2.12 - 1Hz Ps receiver functions before (black) and after (red) sediment 
removal filter is applied plotted from 0 to 20 seconds after the direct P arrival.  
(Top) station B30A (Middle) station U60A (Bottom) station 448A                   54	
 ?
Figure 2.13?  (Top) Mean 4 Hz (blue) and 1Hz (black) Ps RF for station U60A 
shown with picked arrival time for ?tP and ?t shown in the red and magenta  
circles respectively. The arrival time of the two phases (Pbs and PPbs) likely  
overlap at this station. (Bottom) Autocorrelation from the mean 4Hz (blue)  and 
1Hz (black) RF with picked arrival time for ?t. Autocorrelation of a  random 
subset of individual RFs are shown as gray lines in the background.  At station 
B30A, the arrival time of ?t does not depend on event location and  only 
changes slightly with frequency of autocorrelated RFs.                                      57	
 ?
Figure 2.14? SRTC H-?-VP stacking of U60A receiver functions, where negative 
values in the stack are clipped. Station U60A is located on the Atlantic  Coastal 
 vii 
 
 
Plain in North Carolina. A. The mean Ps receiver function before  (black) and 
after (red) the sediment removal filter is applied. B. Ps H-? stack  at assumed VP 
= 6.3 km/s shows no clear maxima. C. The Ps H-k stack after  the sediment 
removal filter and time correction is applied has a clear  maximum. D, E, & F. 
Three cross-sections through the maximum of the H-?- VP triple stack; black 
cross-hairs denote 1? error bars.                                                                          58	
 ?
Figure 2.15? (Top) Mean 4 Hz (blue) and 1Hz (black) Ps RF for station 448A shown  
	
 ? with picked arrival time for ?tP and ?t shown in the red and magenta circles  
respectively. The arrival time of the two phases (Pbs and PPbs) likely overlap  at 
this station. (Bottom) Autocorrelation from the mean 4Hz (blue) and 1Hz  
(black) RF with picked arrival time for ?t. Autocorrelation of a random subset  
of individual RFs are shown as gray lines in the background. At station 448A,  
the arrival time of ?t is challenging to pick, likely due to the multiple  sediment 
layers (multiple negative bumps seen in the 4Hz autocorrelation).                    60	
 ?
Figure 2.16? SRTC H-?-VP stacking of 448A receiver functions. Station 448A is 
located on the Mississippi Embayment in Mississippi. A. The mean Ps  receiver 
function before (gray) and after (red) the sediment removal filter is  applied, 
which do not appear significantly different at this station. B. Ps H-?  stack at 
assumed VP = 6.3, and the maximum of the stack is H = 24 km and ?  = 1.5. C. 
The Ps H-k stack after the sediment removal filter and time  correction is applied 
has a maximum at H = 36.25 km and ? = 1.83. D, E, &  F. Three cross-sections 
through the H-?-VP ?triple stack through the maximum  of the stack; black cross-
hairs denote 1? error bars.                                                                                   61	
 ?
Figure 2.17? SPmp RFs for events between 30 and 50 epicentral distances. Each line 
is a bin of backazimuthal (BAZ) directions from station ANMO, chosen due  to 
its large range in BAZ of SPmp events. The amplitude and location of the  SPmp 
at ~6 s does not change significantly based on BAZ direction which  indicates 
that the SPmp waveforms show evidence of neither crustal  anisotropy nor crustal 
thickness variations with BAZ. We expect this  observation to hold true for most 
other stations, due to the low frequency of  SPmp waveforms.                           65	
 ?
Figure 3.1- Basin thickness for stations on the Atlantic Coastal Plain (Fenneman and 
Johnson 1946 ) inferred from basement depth interpolated from AAPG  
Basement Map of North America, 1978                                                              75	
 ?
Figure 3.2- A. Ray-path geometry of P-to-s phases converted across the base of a 0.5 
km thick sediment layer, along with the largest-amplitude first- and second- 
order multiples for an incident plane wave of horizontal slowness 0.0789  s/km. 
B. Synthetic receiver functions computed for a model with a 0.5 km t hick 
sediment layer and horizontal slowness 0.06 s/km (VP = 2.5 km/s, VS = 1  
km/s), crustal VP = 6.2 km/s, VS = 3.5 km/s, a Moho at 35 km depth, and  
mantle VP = 8 km/s, VS = 4.6 km/s. Direct conversions are in black, while the  
oscillatory pattern is attributed to sediment multiples, labeled in red, blue and  
purple. Orange dots indicate where the expected arrival of phase with more  than 
2 P-wave legs would arrive.                                                                                 78	
 ?
 viii 
 
 
 Figure 3.3- TSs and r0 measurements from 4 Hz Ps receiver function 
autocorrelations. With increasing sediment thickness, TSs increases. A.  
Autocorrelation of Synthetic RF calculated for velocity model described in  top 
right corner. TSs and r0 are picked as the first negative large amplitude,  shown 
by the red dot. B. Autocorrelation of 4 Hz Ps RFs for station U60A  located on 
0.137 km of basin material. Mean autocorrelation shown in  blue and a range of 
Ps RF autocorrelation events shown in gray. TSs and  r0  pick shown as red dot. 
C. Autocorrelation of 4 Hz Ps RFs for station Q60A  located on 1.07 km of basin 
material. Mean autocorrelation shown in blue and  a range of Ps RF 
autocorrelation events shown in gray. TSs and r0 pick shown  as red dot.         78	
 ?
Figure 3.4- - Measurements from the 4Hz Auto-correlated RFs across the ACP. A. 
Measurements of TSs at each station across the ACP. As basin thickness  
increases TSs increases. B. Measurements of r0 across the ACP. As basin  
thickness increases, the r0 (related to the impedance contrast) decreases. C.  
Fundamental Frequency which is related to TSs as F0 = 12TSs ?across the ACP  
.                                                                                                                            79	
 ?
Figure 3.5- Station measurements plotted vs. basin thickness. A.TSs increases with 
increasing basin thickness B. r0 decreases with increasing basin thickness C.F0  
decreases with increasing basin thickness D. TPPbs increases with increasing  
basin thickness E. PPbs decreases with increasing basin thickness.                    81	
 ?
Figure 3.6- VS velocity profile from inversion.  A. TSs measurements from and 4Hz 
Ps RFs (cyan circles) compared to TSs predicted from the mean Vs profile  from 
our best-fit inversion (black crosses) B. VS velocity inversions using TSs  
measurements and assuming VS increases in a power-law fashion with depth.  A 
range of velocity inversions using all stations is plotted in gray, and the  mean of 
these models is plotted in black                                                                           84	
 ?
Figure 3.7- Station grouping between stations near the Mississippi Embayment (ME), 
red circles, and Coastal Plain (CP), blue circles                                                 
Error! Bookmark not defined.	
 ?
Figure 3.8- VS velocity profile from inversion.  A. TSs measurements from and 4Hz 
Ps RFs (cyan circles) compared to TSs predicted from the mean Vs profile  from 
our best-fit inversion (black crosses) B. VS velocity inversions using TSs  
measurements and assuming VS increases in a power-law fashion with depth.  A 
range of velocity inversions using stations with basin thickness less than  2.5 km 
are plotted in gray, and the mean of these models is plotted in black.  The mean 
velocity inversion for the stations in the Mississippi Embayment  (ME) is plotted 
in red, and the mean velocity inversion for the stations in the  Costal Plain (CP) 
is plotted in blue.                                                                                                 85	
 ?
Figure 3.9- Parameters from VS profile inversion (p, a1, and VS0) for all stations 
with basement thickness less than 2.5; ACP stations; ME stations                     86	
 ?
Figure 3.10- TPPbs and PPbs measurements from 4 Hz Ps receiver functions. With 
increasing sediment thickness, TPPbs increases. A. Autocorrelation of  Synthetic 
RF calculated for velocity model described in top right corner.  TPPbs and PPbs 
 ix 
 
 
are picked as the largest positive amplitude, shown by the red  dot. B. 4 Hz Ps 
RFs for station U60A located on 0.137 km of basin material.  Mean Ps RF 
shown in blue and a range of Ps RFs shown in gray. TPPbs and  PPbs pick shown 
as red dot. C. 4 Hz Ps RFs for station Q60A located on 1.07  km of  basin 
material. Mean Ps RF shown in blue and a range of Ps RFs  shown in gray. 
TPPbs and PPbs pick shown as red dot                                                               87	
 ?
Figure 3.11- Measurements from the 4Hz RFs across the ACP. A. Measurements of 
TPPbs at each station across the ACP. As basin thickness increases TPPbs  
increases. B. Measurements of r0 across the ACP. As basin thickness  increases, 
the amplitude of PPbs (related to the impedance contrast) decreases.                88	
 ?
Figure 3.12? VP velocity profile from inversion.  A. TPPbs measurements from and 
4Hz Ps RFs (magenta circles) compared to TPPbs predicted from the mean VP  
profile from our best-fit inversion (black crosses) and TPPbs predicted from the  
Brocher (2005) relationship of VP and VS (green crosses) B. VP velocity  
inversions using TPPbs measurements and assuming VP increases in a power- law 
fashion with depth. A range of velocity inversions using stations with  basin 
thickness less than 2.5 km are plotted in gray, and the mean of these  models is 
plotted in black. The mean velocity inversion for the stations in the  Mississippi 
Embayment (ME) is plotted in red, and the mean velocity  inversion for the 
stations in the Costal Plain (CP) is plotted in blue, and the  VP profile from the 
Brocher (2005) relationship is plotted in green. The  velocity profile is not well 
constrained by any of the data.                                                                            90	
 ?
Figure 3.13? Parameters from VP profile inversion (p, a1, and VS0) for all stations 
with basement thickness less than 2.5; ACP stations; ME stations                     91	
 ?
Figure 3.14? VP velocity profile from inversion.  A. TPPbs measurements from and 
4Hz Ps RFs (magenta circles) compared to TPPbs predicted from the mean VP  
profile from our best-fit inversion (black crosses) and TPPbs predicted from the  
Brocher (2005) relationship of VP and VS (green crosses) B. VP velocity  
inversions using TPPbs measurements and assuming VP increases in a linearly  
with depth. A range of velocity inversions using stations with basin thickness  
less than 2.5 km are plotted in gray, and the mean of these models is plotted in  
black. The mean velocity inversion for the stations in the Mississippi  
Embayment (ME) is plotted in red, and the mean velocity inversion for the  
stations in the Costal Plain (CP) is plotted in blue, and the VP profile from the  
Brocher (2005) relationship is plotted in green. The linear relationships for VP  
agree with each other, and are consistent with the Brocher (2005) regression  
relationship.                                                                                                        91	
 ?
Figure 3.15? Parameters from linear VP profile inversion b1 and VS0 for all stations 	
 ?
with basement thickness less than 2.5; ACP stations; ME stations                     92	
 ?
 Figure 3.16- Comparison of site fundamental frequency (F04) measurements 	
 ?
obtained in this study compared to more traditional spectral ratio techniques  of 
estimating site fundamental frequency from peak amplitude. HVSR  techniques 
applied to Radial (R, dark blue) and Transverse (T, cyan)  component data 
recorded at the seismic station. SSR techniques applied to  PRF data for create 
 x 
 
 
"PRF spectral ratios? using 1Hz (pink) and 4Hz (red)  PRFs. Method applied to 
three stations on differing basin thicknesses (U60A,  Q59A, Q60A). 95	
 ?
Figure 4.1- Major regions of interest in the southeastern United States modified from 
Wagner (2018) Topography is shown in the background.                                9	
 ?
Figure 4.2- The locations of USArray TA stations used in this study colored by the 
number of P-to-s (A) and S-to-p (B) receiver functions calculated at each  
station. Dark red circles are stations that have been deployed for a longer time.     
.                                                                                                                          107 
Figure 4.3- Locations of CCP stacking points (blue) and stations (red triangles). 	
 ?  
Cross sections shown are calculated along vertical and horizontal lines in the  
gird. Each cross section number indicates a W-E cross section, while letters  
indicate a N-S cross-section.                                                                             110	
 ?
Figure 4.4- The velocity assumption in H- stacking is extremely important and can 
significantly change inferred crustal properties. Results from traditional H-k  
stacks calculated for station 060Z on Atlantic Coastal Plain sediments. Top.  As 
the assumed average crustal VP velocity used to calculate the stacks  increases, 
the crustal thickness estimate increases until a new maximum is  inferred (at 6.5 
km/s). Bottom. As the assumed average crustal VP increases,  the VP/VS ratio 
(or ??) decreases, until again a new maximum is inferred                                  111	
 ?
Figure 4.5- Measurements of sediment phase arrival times at each station.The two 
way S-wave travel time in sediment (left) and the arrival time of the PPbs  phase 
(right) are shown. In regions with thicker sediment, both TSs (?t) and  TPPbs (?tP) 
are large, while in regions with thinner sediment TSs and TPPbs are  small. 
Stations in black do not demonstrate a sedimentary signal in the  receiver 
functions.                                                                                                          113	
 ?
Figure 4.6- Values of crustal H, ??, and VP for each station in our study area obtained 
from H- ?? -VP stacking method (described in text). These parameters are used  to 
calculate a velocity model for the migration of RFs from time to depth in  the 
CCP stack                                                                                                          115	
 ?
Figure 4.7- Results of CCP stacking for cross-section H (location shown in top panel 
of figure). Red indicates positive RF amplitude associated with a velocity  
increase with depth while blue indicates negative RF amplitude associated  with 
a velocity decrease with depth. A. In traditional PRF CCP stacking we  see large 
amplitude sediment reverberations that create a Moho step at about  327 km 
along the profile B. Sediment removed (SR) PRF CCP stacks remove  much of 
the sediment reverberation contamination in the regions highlighted  and show a 
flat Moho at 327 km along the profile. C. SRF CCP stacks show a  similar Moho 
profile to SR PRF CCP stacks, but have lower resolution.                               116	
 ?
Figure 4.8- Examples of the resonance filter applied to PRFs. Gray is the mean of the 
PRFs before the resonance removal filter is applied, red line shows the mean  of 
the PRFs after the sediment is removed. A. Station W61A in New Bern,  NC. 
Before the sediment is removed, no clear Moho phase arrival (Pms) exists  in the 
RF. After the sediment is removed, a small amplitude peak appears at  ~5s which 
 xi 
 
 
is likely the Moho arrival. B.  Station S60A in Water view, VA.  Again, before 
the sediment is removed, no clear Moho phase arrival (Pms)  exists in the RF. 
After the sediment is removed, a evidence of a Moho arrival  appears between 4-
5 s.                                                                                                                     117	
 ?
Figure 4.9 - Figure 9- W-E Cross-section 22 showing a strong positive (red) arrival 
between 35 and 45 km depth. Right is the sediment removed (SR) PRF CCP  
stack and left is the SRF CCP stack. Stacks below show the variance of the RF  
amplitude at each point in the CCP stack. We interpret this positive Labels in  
black are features from Figure 1 that the cross sections pass through. RT=  Rome 
trough, GF= Grenville front, SGR = South Georgia rift, ETSZ = eastern  
Tennessee seismic zone, BFZ = Brevard fault zone, CPS= central Piedmont  
suture, NCSQ = North Carolina seismic quiescence, AB = Appalachian basin,  
SCSZ = South Carolina seismic zone, SSZ = Suwanee suture zone, CVSZ =  
central Virginia seismic zone. Colored labels above each cross section denote  
the physiographic province that the cross sections pass through. CL = Central  
Lowland; AP = Appalachian Plateau, V&R = Valley and Ridge, BR = Blue  
Ridge, ACP = Atlantic Coastal Plain, LP = Interior Lower Plateau.                 120	
 ?
Figure 4.10- Crustal thickness/Moho depth from SR PRF CCP stacks. Contour figure 
of Moho depths shows a deep (~65km) Moho beneath the Blue Ridge and  
Appalachian Plateaus (~60 km) that shallows towards the coast (~25km).      121	
 ?
Figure 4.11- W-E Cross-section 10, which passes through the maximum crustal 
thickness estimate in the Blue Ridge. Evidence of multiple crustal layers  
dipping sharply westward can be seen beneath the ETSZ and BFZ. The intra- 
crustal layer is at ~10 km beneath the CPS and deepens to ~40 km beneath the  
ETSZ, while the Moho phase in the same region begins at ~40km depth  beneath 
the CPS and dips to ~60 km beneath the ETSZ and BFZ. The SRF  CCP stacks 
also show a deep ~60 km Moho beneath the ETSZ and BFZ. A  flat intra-crustal 
layer can also be seen beneath the NCSQ at about 15 km  depth. . Stacks below 
show the variance of the RF amplitude at each point in  the CCP stack. See 
caption in Figure 9 for description of labels and  abbreviations.                       125	
 ?
Figure 4.12- W-E Cross section 19 that shows an eastward dipping intra-crustal layer 
beneath the V&R and BR provinces in the SR PRF CCP stack. This intra- crustal 
layer appears at ~15 km beneath the Appalachian basin and dips  eastward to 
~20 km beneath the Piedmont. Stacks below show the variance of  the RF 
amplitude at each point in the CCP stack. See caption in Figure 9 for  description 
of labels and abbreviations.                                                                               126	
 ?
Figure 4.13- Dip direction of intra-crustal layering at each CCP stacking latitude and 
longitude.  Points indicate locations in the CCP stacking grid that show intra- 
crustal layering. Red points indicate locations where a westward dipping intra- 
crustal layer is seen, blue points indicate the latitude and longitudes where a  flat 
intra crustal layer is seen, and blue indicates the location where an  eastward 
dipping intra-crustal layer is identified. Purple shaded circles note  the location 
of seismic zones in the SEUS, the brown line notes the predicted  location of the 
Grenville Front  (GF), and the green dotted lines indicate  locations of suture 
 xii 
 
 
zones. In the south, the green line is interpreted as the  Suwannee suture zone 
(SSZ) where in the north the green dotted line indicates  a suture of unknown 
origin, but may be the suture between the Laurentian and  Amazonian crust.  128	
 ?
Figure 4.14- N-S cross section D which crosses through the main seismicity in the 
ETSZ. Here a strong mid-crustal discontinuity is seen dipping northward from  
~10 km beneath the CPS/SSZ to ~20 km beneath the ETSZ and BFZ. In the  
same region, there is a sharp change in the crustal thickness from ~40 km to  ~60 
km beneath the BFZ/ETSZ. Stacks below show the variance of the RF  
amplitude at each point in the CCP stack. See caption in Figure 9 for  description 
of labels and abbreviations.                                                                               129	
 ?
 Figure 4.15- W-E cross-section 16 through the CVSZ. Beneath the CVSZ we see an 
intra-crustal layer dipping slightly westward from around 15 km beneath the  
ACP and only deepens to 22-23 km beneath the BR. Directly below the CVSZ  
there is a change in crustal thickness from ~25 km to ~45 km over a later  
distance of about 100 km. Stacks below show the variance of the RF  amplitude 
at each point in the CCP stack. See caption in Figure 9 for  description of labels 
and abbreviations                                                                                              131	
 ?
Figure 4.16- N-S cross section H through the SCSZ. In this cross section the Moho 
appears fairly flat beneath the CL, AB, BR, and V&R. The only large change  in 
the Moho depth occurs almost directly below the SCSZ in the Piedmont.  Here 
the Moho changes from ~42 km to 30 km.  Unlike Figures 15 & 16, no  
westward dipping intra-crustal layering is seen. Stacks below show the  variance 
of the RF amplitude at each point in the CCP stack. See caption in  Figure 9 for 
description of labels and abbreviations                                                             132	
 ?
Figure 4.17- W-E Cross-section 10 that passes through the NCSQ. Beneath the 
NCSQ in the ACP we see both a flat Moho at 40km and no strong evidence  for 
intra-crustal layering. Stacks below show the variance of the RF amplitude  at 
each point in the CCP stack. See caption in Figure 9 for description of  labels 
and abbreviations                                                                                              133	
 ?
 xiii 
 
 
List of Abbreviations 
 
 
AB: Appalachian basin 
ACP: Atlantic Coastal Plain  
AP: Appalachian plateau 
BFZ: Brevard Fault Zone 
BR: Blue Ridge 
CCP: common conversion point  
CPS: central piedmont suture 
CVSZ: Central Virginia seismic zone 
E-HVSR: earthquake horizontal to vertical spectral ratio  
ETSZ: Eastern Tennessee seismic zone 
GF: Grenville Front 
F0: fundamental frequency 
HVSR: horizontal to vertical spectral ratio 
LAB: lithosphere-asthenosphere boundary 
LP: lower plateau  
Moho: Mohorovi?i? discontinuity 
MLD: mid-lithospheric discontinuity  
NCSQ: North Carolina seismic quiescence  
PRFs/Ps RFs: P-to-s receiver functions  
RFs: receiver functions 
RT: Rome trough  
SCSZ: South Carolina seismic zone 
 xiv 
 
 
SEUS: Southeastern United States 
SGR: South Georgia Rift 
SR: sediment removed 
SRFs:/Sp RFs S-to-p receiver functions 
SRTC: sediment-removed time corrected 
SSR: standard spectral ratio 
SSZ: Suwanee Suture Zone 
TA: transportable array 
VP: P-wave velocity  
Vs: S-wave velocity 
V&R: valley and ridge 
Z1.0/Z2.5: depth to VS = 1.0 or 2.5 km/s respectively 
   
 
 
 
 
 
 
 
 xv 
 
 
Chapter 1: Introduction to the Lithosphere and the Southeastern United 
States 
 
Tectonic plates glide across the underlying mantle, shaping the history of our 
continents through collision and subduction. These plates make up the lithosphere, 
the Earth?s topmost layer that includes the crust and rigid upper portion of the mantle. 
As continents collide and break apart making new continents and creating new 
tectonic plates, the lithosphere is deformed. Lithospheric structure then contains a 
history of the continental formation and evolution of a region as well as provides 
evidence for ongoing mantle dynamics. Observed variations in the lithospheric 
structure are often used to understand tectonic processes; thinned lithosphere implies 
extensional or rifting forces are at play, while thickened lithosphere implies 
compressional or collisional forces. 
 
The southeastern United States (SEUS) has undergone multiple episodes of 
continental convergence and rifting, so mapping the boundaries of accreted terrains 
from previous collisional events ? or inherited structures ? can clarify the tectonic 
history of this region. Boundaries between inherited structures may also correlate 
with the ongoing seismicity of the region. Being distant from the nearest tectonic 
plate boundary, the damaging earthquakes observed in the SEUS cannot be explained 
by release of energy accumulated between plates moving with respect to one another. 
The long recurrence time of large intra-plate earthquakes makes them particularly 
dangerous and the location and magnitude of these events are difficult to predict. 
Mechanisms controlling the location of intra-plate seismicity are poorly understood, 
 1 
 
 
but are likely related to preexisting zones of weaknesses in the crust or lithosphere 
inherited from past tectonic history of the SEUS.  
 
A common seismological tool for imaging boundaries beneath the surface is receiver 
functions. Receiver functions are sensitive to discontinuities from differences in 
material properties such as the basin-bedrock interface, the crust-mantle boundary 
(often referred to as the Moho) and the lithosphere-asthenosphere boundary (LAB). 
The basin-bedrock boundary is important for understanding the long-period (low 
frequency) effects of shaking during an earthquake, which affects large buildings and 
structures. The largest ground motions are related to the depth and contrast between 
the basin-bedrock interface, and so constraining this across the SEUS will have 
implications for hazard analysis.  
 
Constraining crust and Moho properties across the SEUS using receiver functions 
may help to understand the history of inherited structures and seismicity. Sharp 
changes in the Moho boundary may correlate with regions of seismicity or indicate 
locations of inherited crust; while the thickness and structure of the lithosphere in the 
SEUS is still debated, and further constraints might clarify the depth extent of the 
plate boundary. Variations in the LAB and other intra-lithospheric discontinuities 
may also be weak-zones related to inherited structures that correlate with seismicity. 
Currently, hazard estimates of intra-plate earthquakes use the ongoing seismicity in 
the area. These hazard assessments likely overestimate the hazard in regions with 
small seismicity and simultaneously underestimate the hazard in regions that have 
 2 
 
 
may have the potential to produce large earthquakes but haven?t in the historic past 
(Stein and Liu, 2009). By better understanding how lithospheric variations relate to 
currently active regions of seismicity, we may be able to predict locations of large 
and damaging intra-plate earthquakes in the future.  
 
Although it is one of the more straightforward methods to studying lithospheric 
discontinuities, calculating receiver functions in the SEUS has proven difficult. 
Atlantic Coastal Plain sediments overlie the majority of the SEUS, and signal from 
sedimentary basins can overprint signals from deeper earth structure. Therefore, in 
my dissertation, I first develop a method to constrain the crust/mantle boundary in 
regions with sedimentary basins (Chapter 2), and then I constrain important 
basin/bedrock properties for hazard analysis in the SEUS (Chapter 3). Finally, I apply 
the new methodology to the SEUS to investigate the underlying lithosphere without 
sediment contamination (Chapter 4). This work has implications for improving 
characterization of lithospheric structure beneath regions with sediment, which may 
be up to 70% of the continental United States.  
1.1 Overview of Chapter 2 
Receiver functions are sensitive to sharp seismic velocity variations with depth and 
are commonly used to constrain crustal thickness. The H-? stacking method of Zhu 
and Kanamori (2000) is often employed to constrain both the crustal thickness (H) 
and ?!/?! ratio (?) beneath a seismic station using P-to-s converted waves (Ps). 
However, traditional H-? stacks require an assumption of average crustal velocity 
(usually ?!). Additionally, large amplitude reverberations from low velocity shallow 
 3 
 
 
layers, such as sedimentary basins, can overprint sought-after crustal signals, 
rendering traditional H- ?? stacking uninterpretable. We overcome these difficulties in 
two ways. When S-wave reverberations from sediment are present, they are removed 
by applying a resonance removal filter allowing crustal signals to be clarified and 
interpreted. We also combine complementary Ps receiver functions, Sp receiver 
functions, and the post-critical P wave reflection from the Moho (SPmp) to remove 
the dependence on an assumed average crustal ?!. By correcting for sediment and 
combining multiple data sets, the crustal thickness, average crustal P-wave velocity, 
and crustal ?!/?! ratio is constrained in geologic regions where traditional H- ?? 
stacking fails, without making an initial P-wave velocity assumption or suffering 
from contamination by sedimentary reverberations.  
1.2 Overview of Chapter 3 
Thickness and seismic velocities of sedimentary basins strongly affect their response 
during earthquakes, prolonging and amplifying ground motions. We characterize 
shallow basin structure of the Atlantic Coastal Plain (ACP) using a passive-seismic 
approach based on high-frequency P-to-S receiver functions. We map the site-specific 
basin fundamental frequency for 64 USArray TA stations and confirm that the 
method yields results similar to those from traditional spectral ratio techniques, with 
basin fundamental frequencies between 0.1 and 1 Hz. In addition, using sediment S-
wave reverberations and P-to-s phase arrival times measured directly from the 
receiver function, we invert for average basin S and P-wave velocity profiles of the 
ACP. We find that that VS increases with depth following a power-law relationship 
(?! ? ?) while the increase of VP with depth appears to be more linear. Finally, we 
 4 
 
 
use variation of measured S-reverberation and P-to-s phase amplitudes with depth to 
validate the P and S wave velocity profiles for the ACP. These results have 
implications for seismic shaking across the ACP, which is home to millions of 
Americans. 
1.3 Overview of Chapter 4 
The Southeastern United States lies on a passive margin shaped by its complicated 
tectonic history of collisional events. Despite its distance from active plate margins, 
damaging earthquakes still occur whose mechanisms are poorly understood. 
Weaknesses in the lithosphere inherited from past collisional events are the most 
likely explanation, but the extent and magnitude of lithospheric variations in the 
SEUS has not been fully explored. Here, we use sediment removed P-to-s receiver 
functions (PRFs), combined with complimentary S-to-p receiver functions (SRFs), to 
constrain crustal and lithospheric variations across the entire SEUS. Through our 
investigation, we find evidence for deep mid-crustal layering (or a ?double-Moho?) 
deepening westward beneath the blue ridge in eastern Tennessee as well as a shallow 
mid-crustal layer deepening eastward which has been interpreted to mark the location 
of the suture between Laurentia and Gondwana during the formation of the 
supercontinent Pangea (Hopper et al. 2017). We find no evidence for a strong 
lithosphere-asthenosphere boundary or mid-lithospheric discontinuities associated 
with the location of intra-crustal earthquakes; however, we find a correlation with 
sharp changes in the crustal thickness.    
 
 5 
 
 
1.4 A Brief History of SEUS Tectonics   
Here we provide a brief overview of the tectonic history of the southeastern United 
States (SEUS) beginning in the Proterozoic and continuing to modern day; generally 
following Wagner et al. (2018) who provide a more detailed tectonic history of the 
region. 
 
 The continental cratonic core of North America contains some of the oldest 
lithosphere on the planet. In the present day, this same cratonic lithosphere extends 
from the Rocky Mountains in the west to near the edge of the Appalachian Mountains 
in the east (Yuan et al., 2011). This core made up the continent Laurentia, and has 
remained stable since its formation in the Proterozoic. Between about 1.4 and 1.1 Gya 
the Grenville orogeny occurred which brought together multiple continental cores to 
create the supercontinent Rodinia (Whitmeyer and Karlstrom, 2007).  The continent-
continent collisions of the Grenville orogeny formed an extensive mountain range on 
the east coast of what is now the United States and Canada. Although the large 
mountains have eroded over time, evidence of their roots remain (Wagner et al., 
2018). Much of what is now the SEUS did not exist at this time, and the exact 
location of westward extent of deformation that occurred during the Grenville 
orogeny, called the Grenville front (GF- Figure 1.1), remains elusive. Most studies 
agree that the extent of the GF is preserved past the Appalachian Mountains in what 
is now Tennessee and Ohio (Whitmeyer and Karlstrom 2007; Rivers et al., 2014). 
However, another suggests that the GF lies beneath the mid-continent rift in 
Wisconsin and Minnesota (Stein et al., 2018) 
 6 
 
 
  
After a period of stability, Laurentia broke apart from the supercontinent Rodinia 
about 700 Mya and the vast Iapetus Ocean was formed.  This major rifting event left 
other failed rifts within the SEUS including the Reelfoot Rift and the Rome Trough 
(RT- Figure 1.1) (Whitmeyer and Karlstrom, 2007). After this rifting event, further 
accretion occurred including the accretion of the Taconic island arc to New England 
and Canada during the Taconic Orogeny around 490-460 Mya. This event 
downwarped the crust and later erosion of these mountains led to the creation of the 
inland Appalachian Basin that overlaps some of the RT (AB- Figure 1.1) (Clark, 
2001) With the subsequent closure of the Iapetus Ocean, additional accretion added 
an unknown number of terranes (Hibbard, 2002; Hatcher, 2007; Hatcher, 2010; 
Hibbard 2010) including some arc accretions. The accretion of the Taconic island arc 
and creation of the Taconic Mountains in New England downwarped the crust and 
created the Appalachian basin (AB-Figure 1.1).  Constraining the number and extent 
of subsequent terrane accretions at depth is particularly challenging due to the 
massive re-working of the continent including crustal shortening and emplacement 
and possible changes in subduction vergence directions meaning that the underlying 
basement crust is often unknown (Merschat and Hatcher, 2007; Wagner et al., 2012). 
There is some evidence for the extent of the Piedmont terrane and Carolina terrane 
accretions (Hibbard, 2000) between 360-345 Mya and the boundary between the two, 
called the central piedmont suture is fairly well constrained (CPS- Figure 1.1) 
 
The supercontinent Pangea formed with the collision of Gondwana and Laurentia 
during the Alleghanian orogeny beginning around 350 Mya (Clark, 2001). The 
 7 
 
 
Alleghanian orogeny compressed and shortened the eastern margin of Laurentia at 
least 200 km (Hopper et al., 2017), creating the Appalachian Mountains and leaving 
the Brevard fault zone (BFZ- Figure 1.1), a series of major thrust faults between the 
Blue Ridge and the Piedmont that extend into the crust (Christensen and Szymanski, 
1988). The suture zone of Gondwana and Laurentia is located off the coast of the 
SEUS for the most part, except for between Laurentia and the accreted Tallahassee-
Suwanee terrane (of modern day Florida and Georgia) (SSZ- Figure 1.1-  Mueller et 
al., 2014; Hopper et al. 2017). 
  
The breakup of Pangea around 175 Mya marks the end of accreted terranes to the 
SEUS (Whitmeyer and Karlstrom, 2007). With the initiation of rifting came the 
formation of the South Georgia Rift basin (SGR Figure 4.1) and subsequent Central 
Atlantic Magmatic Province magmatism (Whalen et al., 2015; Wagner et al. 2018). 
As the SEUS transitioned into a passive margin, erosion and sedimentation along the 
eastern margin created the Atlantic Coastal Plain (ACP- Figure 1.1). 
 
 8 
 
 
 
Figure 1.1- Major regions of interest in the southeastern United States modified from 
 Wagner (2018) Topography is shown in the background. Red circles indicate 
 the location of earthquake Mw>2.5 since 1900.  The brown line notes the 
 location of the Grenville Front (GF) (Witmeyer and Karlstrom, 2007); 
 Magenta indicates the region associated with the Rome Trough (RT), The 
 Brevard Fault zone (BFZ) is shown as a series of lines with triangles (Powell 
 et al., 2016). The blue line indicates the location of the Central Piedmont 
 Shear Zone (CPS). Dotted black line is the location of the Suwannee suture 
 zone  (SSZ). Basins and rifts labeled in white with shaded regions showing 
 their approximate extent. The seismogenic zones in purple are:  ETSZ = 
 eastern Tennessee seismic zone, SCSZ = South Carolina seismic zone, and 
 CVSZ = central Virginia seismic zone while cyan is the NCSQ = North 
 Carolina seismic quiescence. The SGR is the South Georgia Rift basin in 
 black, AB = the 4000 ft contour of the sediment thickness of the Appalachian 
 Basin which defines the deepest portions of the basin in orange, and light blue 
 ACP region notes the location of the Atlantic coastal plain  sediments. 
 9 
 
 
 
1.5 Technical Background: Receiver Function Methodology 
Layers in the earth (e.g. crust, mantle, core, etc.) contain different material properties 
that distinguish them from one another. The density, bulk, or shear modulus of a 
material can change the velocity of the seismic wave, reflecting changes in 
temperature and composition. Velocity and density differences (impedance contrasts) 
between different layers can be isolated using the receiver function method 
(Langston, 1979) and used to constrain their material properties.  
 
Earthquakes produce a seismic wavefront that travels in all directions away from the 
source. Some energy from the seismic wave is recorded at distant seismic stations 
after traveling through the body of the earth. The resulting seismogram recorded at a 
station u(t) can be considered the convolution (*) of functions describing the source, 
s(t), and the structure, g(t), of the earth that the seismic waves have passed through 
(Figure 1.2).  
 
 
Figure 1.2 - Ideally the displacement recorded at a seismogram (u(t)) is the 
 convolution of the source function (s(t)) and the earth response (g(t)). (Figure 
 modified from Shearer, 2009) 
 10 
 
 
In receiver function analysis, we aim isolate the earth response, g(t), which, in the 
case of receiver functions, is a result of seismic wavefield interactions with 
impedance discontinuities. To do this, the three-component seismograms are rotated 
in the direction of the earthquake source, upgoing P and S waves are estimated using 
near-surface velocities, and then the source-related signals s(t) are removed through 
the process of deconvolution.  
 
For an incoming P wave crossing a sharp velocity discontinuity, some of the energy 
from the direct P body waves is converted to a vertically polarized S wave (referred 
to as SV) across the discontinuity (Figure 1.3). A sharp discontinuity is one that 
occurs over a distance less than ? wavelength of the parent wave. In the case of the 
direct P wave, P is the parent and SV is the daughter waveform. Conversely, for an 
incoming S wave some of the SV energy is converted to a P wave (Figure 1.4); In 
this case SV is the parent wave and P is the daughter waveform. The parent 
waveforms contain the source information, s(t), while the daughter waveforms 
contain the earth response information g(t) convolved with source information s(t). 
Receiver functions result from deconvolving source-related signals from the daughter 
using the parent waveforms. The relative amplitudes and arrival times of features in 
the receiver functions help constrain the earth?s structure beneath the station (e.g. 
Langston 1979, Farra and Vinnik, 2000) 
 11 
 
 
 
Figure 1.3 - Cartoon of converted P-wave energy across a Moho over mantle half 
 space if no tilted axes of anisotropy are present.  (Left) converted paths from a 
 direct P wave at the Moho (P phases in red and S phases in dotted blue) and 
 (right) their relative arrival times in a Ps receiver function. 
S-Converted Ray Paths at MLD Sp Receiver Function 
Direct S arrival  
Rela%ve'Velocity' Time (S to p) 
Surface Sp Structure''
Crust 
Lithosphere 
Lithosphere 
P wave 
S wave   
Figure 1.4- Cartoon of converted S-wave energy across a MLD over mantle half 
 space if no tilted axes of anisotropy are present Left - converted paths from a 
 direct S wave at the MLD (P phases in red and S phases in dotted blue) and 
 Right - relative arrival times in a Sp receiver function. 
1.5.1 Phase picking and P-SV-SH rotation 
Often in receiver function studies, direct parent wave arrival times to each station are 
handpicked. However, due to the amount of data produced by USArray, handpicking 
waveforms for the entire array becomes impractical. For this reason, the automated 
 12 
 
 
phase picking algorithm of Abt. et al. (2010) is used. In this algorithm, the predicted 
arrival of a parent phase is calculated through a l-D velocity model (AK135) for the 
earth. To determine the actual arrival time, 25 seconds before and after the predicted 
arrival time is searched to find the arrival time that maximizes the ratio of the short 
term average (STA) to long term average (LTA), following the algorithm of Earle and 
Shearer (1994). If the maximum STA/LTA ratio is smaller than 2 then the predicted 
arrival from the 1-D model is used. Seismic stations record waveforms in 3 
orthogonal components, often oriented to geographic North (N), East (E), and normal 
to the surface (Z). The N and E components are rotated relative to the earthquake 
source location to obtain motions in the radial (R) and transverse (T) directions. The 
R component describes the horizontal direction along which the P wave is 
propagating from the source, and the T component is orthogonal to R in the 
horizontal plane. The rotation is expressed mathematically given the back azimuth, ? 
(the angle of the earthquake location at the station from north): 
 
1.1  ? ? ? ? ? ? ? ? ? ?? =  ?
???? ???? ?
????? ???? ?  
 
In order to isolate the converted phases in receiver functions, the components are 
transformed onto a coordinate system that isolates the parent and the daughter 
waveforms (P-SV-SH system). The waveform interaction at the free surface boundary 
can be described mathematically through the free surface transform matrix of Kennett 
(1991) and Bostock (1998) in terms of the P wave velocity (?), S wave velocity (?), 
and the horizontal slowness (?): 
 13 
 
 
 
! ?!?! ? 1??
0 2? ??
? !1 ?
 ? ? ? ? ? 1.2  ? ? ? ? ? ? ? ? ? ?? =  ? ? !2? ? ?
! ?  
?? 0 ????! ?
1
0 2 0
 
where ?! and ?! are vertical slownesses of P and S waves, respectively, defined by 
the equations:  
 
1.3  ? ? ? ? ? ? ? ? ? ? ? ? ? ?? !! !! =  ? ? ? ?          and  ? ? ? ? ? ??! ? =  ? ?!! ? ?! 
 
This transformation matrix describes how the seismic wavefield interacts at the free 
surface. The matrix transformation should isolate the parent and daughter phases 
allowing for deconvolution to take place.  
1.5.2 Deconvolution 
Deconvolution is the method by which I can isolate converted phases arrivals (s in Ps 
and p in Sp) to infer underlying velocity and geologic structure. In terms of the parent 
and daughter waveforms, the daughter waveform (d) can be written as the 
convolution (*) of the earth response (g) with the upgoing parent (p) waveform. 
Convolution in the time domain can be described as the integral of one signal with 
another that is reversed and shifted in time. Mathematically this is written as:  
 14 
 
 
!
1.4  ? ? ? ? ? ? ? ? ?? ? =  ?? ? ? ? ? = ? ? ? ?  ?? ?  ??? ? 
!!
 
Convolution in the time domain is equivalent to multiplication in the frequency 
domain and so equation can be expressed as: 
 
1.5  ? ? ? ? ? ? ? ? ?? ? =  ?? ?  ??(?),  
 
where ? ? ,? ?  ?, ?and ? ?  ?are the Fourier transform of the time domain signals 
d(t), g(t), and p(t)m respectively. It seems simple then to isolate earth response as the 
spectral division (though this is usually a bad idea in practice due to zeros and near-
zeros of P(w)):  
 
 ?? ?
1.6  ? ? ? ? ? ? ? ? ?? ? = ? ?  
 
The earth structure ? ?  ?can be transformed back into the time domain by taking the 
inverse Fourier transform. Because the time domain is related to the frequency 
domain through as Fourier transform, it should be possible to calculate equivalent and 
interchangeable receiver functions in either the time or frequency domain. There are 
two common methods of deconvolution that have been used most often to calculate 
stable RFs. The first is a variation of frequency domain deconvolution method of 
Bostock (1998); the second is an iterative deconvolution in the time domain following 
Ligorria and Ammon (1999).  Because each deconvolution method has tradeoffs, both 
 15 
 
 
methods are used and compared in our study.  In the future, we could include the 
trans-dimensional hierarchical Bayesian deconvolution (THBD) developed by Kolb 
and Lekic (2014). The THBD method allows for an estimation of inversion likelihood 
through Bayesian statistics, but comes at the cost of much higher computation time. 
1.5.3 Frequency Domain Deconvolution 
Although the spectral division shown in equation 6 looks straightforward, 
complications can arise that make the deconvolution unstable.  For example, if the 
denominator becomes sufficiently small, the deconvolution can in effect ?blow up?. I 
can stabilize (i.e. ?regularize?) the deconvolution by multiplying both the numerator 
and denominator by the complex conjugate ??(?) of the parent waveform and adding 
a small factor to the denominator. This is the frequency domain damped least squares 
deconvolution (Bostock 1998): 
 
?
1.7  ? ? ? ? ? ? ? ? ?? ? =  ?! !  ?! (!) , 
! !  ?!?(!)!!
 
the small factor, ?, in this case is the best regularization parameter. The best 
regularization parameter is the one that minimizes the generalized cross validation 
function (GCV). The GCV computes G(w), convolves it with the parent to calculate a 
predicted daughter waveform, and then computes its misfit to the observed daughter. 
This is repeated over a range of regularization parameters (?). The smallest GCV 
indicates which regularization parameter should be used. Adding the regularization 
parameter means that the least squares frequency domain deconvolution is not a pure 
 16 
 
 
deconvolution, but it prevents the spectral division from ?blowing up? by ensuring a 
non-zero denominator for division. Figure 1.5 demonstrates the effect of changing 
the regularization parameter on the resultant receiver function for TA station J27A 
located in Hobbs, New Mexico.  
1e6  *?  
 1e3 *?  
 1     *?  
 
Figure 1.5? Ps RF for station 127A located in Hobbs, New Mexico. The y-axis 
 shows  the fraction of the amplitude of direct P arriving phase, the x-axis is in 
 time.  
 
The resulting RFs are shown when the regularization parameter ? is multiplied by 1 
(blue line), 1000 (green line), or 106 (pink line). The conclusion drawn from this 
figure is that ? alone (without any further modification) is enough to stabilize the 
deconvolution.  
As expected, a larger multiple of the regularization parameter in the denominator 
increases the stability but reduces the amplitudes of arriving phases to the solution. 
However, the integrity of the receiver function and phases arrivals remains consistent. 
A benefit of using the frequency domain deconvolution is that it is computationally 
 17 
 
 
efficient. This is especially important when considering the large amount of data 
collected in the USArray. 
 
Frequency domain RFs are computationally fast and fairly reliable, however, other 
methods may sometimes be preferred. Narrow spikes (relating to the impedance 
contrast of earth structure) exist in the time domain, and are associated with Fourier 
transforms whose power spans multiple frequencies. For example, the Fourier 
transform of a delta spike signal in the time domain has infinite width in the 
frequency domain. A longer duration signal, like a boxcar function will produce a 
Fourier transform that looks like the sinc function (Figure1.6).  Simply put, the more 
defined or sharp a pulse is in the time domain, the larger its extent in the frequency 
domain. This spectral extent of signals can interfere with spectral contribution of later 
arriving phases on a receiver function, or be distorted by the band-limited nature of 
the data itself, which only has usable signal-to-noise ratios across a relatively narrow 
band of frequencies. This means that when computing RFs, there is a tradeoff 
between localization of signals in time versus frequency. This is also the case with 
time-domain deconvolution methods.  
 
 
 18 
 
 
	
 ?	
 ?Domain:          Time                                Frequency  
Spectral	
 ?
leakage 
 
Figure 1.6- Fourier transform from time to frequency domain for common functions. 
 Arrows indicate the Fourier transform was taken. Top - delta function in the 
 time domain yields a flat spanning all frequencies in the frequency domain. 
 Bottom - boxcar function in the time domain yields a sync function with 
 multiple frequencies (spectral leakage) in frequency domain.  
 
1.5.4 Time Domain Deconvolution 
The time domain simultaneous least squares deconvolution of Kikuchi and Kanamori 
(1982) and Ligorria and Ammon (1999) presents an alternate method of 
deconvolution. Instead of employing spectral division, a cross correlation of the 
parent and daughter components is performed. A Gaussian peak is introduced at a 
time lag that corresponds to the maximum cross correlation; this is then convolved 
with the parent component and subtracted from the daughter component. The width 
of the Gaussian peak used in the deconvolution is related to the frequency-time 
localization tradeoff discussed above in Chapter 1.1.3.  The process is repeated until 
the set number of iterations is met.  This method limits the addition of spectral 
artifacts in station TA 127A (Figure 1.6). The signal from the time domain receiver 
functions appears to be much more clear that the previous frequency domain 
 19 
 
 
deconvolution, and the presence of side lobes is greatly reduced. However, the 
number of iterations determines the resolvability of each receiver function ? a 
parameter chosen a priori, and the computation time is significantly larger. 
 
Figure 1.2? Time (ITDD in red) and Frequency (IFDD in blue) domain RF 
 comparison for station 127A located in Hobbs, New Mexico. IFDD RF was 
 filtered at 0.03-1 Hz and appears overall smoother than the ITDD RF. 
 However, it is difficult to determine if the Ps arrival associated with the Moho 
 occurs at 3.5 or 6 s. The ITDD RF was created with 300 iterations and appears 
 to have more arriving phases, however, the spectral leakage has been reduced 
 and the peaks narrow. The RF amplitudes for both frequency and time domain 
 RFs have been normalized to 1. 
 
 
 
 
 
 
 
 20 
 
 
Chapter 2: Constraining Crustal Structure in the presence of Sediment: A 
Multiple Converted Wave Approach 
 
2.1 Introduction and Rationale for New Methodology  
Across a sharp seismic impedance contrast, some of the seismic energy from direct 
teleseismic P waves converts to SV wave energy, and vise-versa. Receiver functions 
(RFs) isolate the time series of converted wave energy directly beneath a seismic 
station by deconvolving the parent waveform (P for P-to-s conversions and S for S-
to-p) from the daughter waveform (S for Ps and P for Sp), removing complexity in 
the signal due to earthquake rupture and distant structure. RFs are sensitive to sharp 
seismic velocity variations with depth and are therefore used to constrain shallow 
crustal and lithospheric velocity discontinuities such as the boundary between the 
crust and mantle (Mohorovi?i? discontinuity or Moho) (e.g Vinnik, 1977; Langston, 
1979; Owens et al., 1984). Stacking the amplitudes of RFs at the expected arrival 
times of the converted phase across the Moho and its primary reverberations 
(multiples) for a range of crustal thickness (H) and relative velocity (?!/?! ratio or ?) 
values will produce a maximum of the stack corresponding to the best estimate for H 
and ?? (Zhu and Kanamori, 2000). Due to its simplicity and exploitation of both direct 
converted and reverberating phases, H-? stacking has become a widely used method 
to obtain an estimate of the average crustal thickness and velocity beneath a seismic 
station.  Traditional H-? stacking has been used extensively to estimate the thickness 
and ?!/?! of the crust  (Kumar et al., 2001; Ramesh et al., 2002; Eaton et al., 2006; 
Rychert et al., 2007; Audet et al., 2009; French et al., 2009; Yuan et al., 2010; 
Levander and Miller, 2012; Parker et al., 2013; Reeves et al., 2015; Biryol et al., 
 21 
 
 
2018; Soto-Cordeo et al., 2018).  With the deployment of large-scale seismic arrays, 
such as the EarthScope USArray, automated RF analysis in the form of H-? stacking 
has been deployed to constrain crustal structure on a continental scale (Crotwell and 
Owens, 2005). While straightforward, the traditional method of Ps H-? stacking is 
limited by two major factors: 1) The prior assumption of an average crustal velocity 
beneath a seismic station and 2) Contamination of crustal signal by sediment or other 
shallow-layer reverberations. These limitations have proved difficult to overcome, 
and can significantly bias estimates of crustal thickness and ?!/?! ratios based on 
them. 
 
Several studies have quantified how assuming an incorrect average ?! can bias crustal 
thickness and ?!/?! estimates from H-? stacks by up to 10s of kilometers for H 
(Sheehan, 1995; Zelt and Ellis, 1999; Yeck et al., 2013). At some locations, an 
accurate crustal velocity may be obtained from active source experiments or surface 
wave inversion studies; however, a continental scale study requires that these 
measurements are available at all locations and reflect the same region of lateral 
sensitivity as the receiver functions (Chai et al., 2015; Rychert and Harmon, 2016). 
Though Ps RFs do have some sensitivity to average crustal ?!  (Kumar and Bostock, 
2008; Bostock and Kumar, 2010), it is typically insufficient to provide robust 
estimates of ?! beneath the seismic station. Therefore, to remove the dependence on 
?!, Sp RFs can be combined with Ps RFs to create stacks that constrain average 
crustal H, ?!, and ?! (Rychert and Harmon, 2016). The S-to-p wave conversion from 
the Moho arrives before the direct S, meaning that Sp RFs are not contaminated by 
 22 
 
 
sediment reverberations in the same way as Ps RFs, although, Sp RFs have a smaller 
amplitude conversions and typically contain energy at lower dominant frequencies 
than Ps RFs. Because vertical resolution depends on frequency content, lower 
frequency Sp RFs yield a lesser vertical resolution. Due to the relatively low signal to 
noise ratio of Sp converted phases, the large amplitude post critical P reflection from 
the Moho, called the SPmp phase, can be used to constrain average crustal P wave 
velocity (Langston 1996; Owens and Zandt, 1997; Yu et al., 2012; Kang et al., 2016; 
Parker et al., 2016). While the SPmp phase has a broad region of lateral sensitivity, it 
is less impacted by sediment reverberations and depends on average crustal ?! and 
crustal thickness, rather than ?!.  
 
However, even once crustal ?! is well constrained, a slow sedimentary layer beneath 
a seismic station can significantly bias crustal thickness estimates (Yeck et al., 2013) 
and large amplitude sediment reverberations can directly overprint Ps conversions 
from the Moho, rendering interpretation of Ps H-? stacks challenging (Zelt and Ellis, 
1999). Previous studies have removed sediment reverberations with some success, but 
these methods fail if more than one sedimentary layer exists or if sedimentary phase 
arrivals directly overlap Moho arrivals. (Yeck et al., 2013, W?lbern and R?mpker 
2017), and require an assumption about average crustal ?!  (Yu et al., 2015). 
Currently no single method overcomes limitations of traditional H-? stacking, the 
assumption of crustal ?!, and contamination by one or many sedimentary layers.  
 
 23 
 
 
To constrain average crustal thickness, ?!, and ?!/?! in sediment dominated regions, 
contamination from sediment reverberations should be removed and data with 
complementary sensitivity should be incorporated. We propose a method that applies 
a resonance removal filter to Ps RFs when contaminated by sediment, and combines 
these sediment-removed Ps RFs with Sp RFs as well as the envelopes of SPmp RFs. 
Stacking these three complementary datasets across a range of H, ?, and ?! and 
accounting for slowness in sedimentary layers yields a maximum at H, ?, and ?! 
values corresponding to the best estimate of all three crustal parameters. We call this 
method is called the sediment-removed, time-corrected (SRTC) H-?-?! triple stack. 
The ability of this method to constrain the crustal structure is demonstrated using 
synthetic data and results from three different geologic/tectonic regions using 
recordings from USArray Temporary Array (TA) stations. The two main benefits of 
this method are that it constrains all three crustal parameters in regions where 
sediment reverberations contaminate Ps RFs and that it can be automated to calculate 
these values for continental scale seismic arrays such as the EarthScope USArray. 
 
2.2 SRTC H- ? -VP Stacking 
By removing source and path effects through deconvolution, receiver functions 
isolate the near receiver structure. Deconvolution can be performed in the frequency 
domain using modified spectral division (Clayton and Wiggins, 1976; Bostock, 1996; 
Dueker and Sheehan, 1997; Lawrence and Shearer, 2006); however, we choose to 
employ the time domain simultaneous least squares deconvolution of Kikuchi and 
Kanamori (1982) and Ligorria and Ammon (1999) to reduce the appearance of side 
 24 
 
 
lobes. (For a full discussion of receiver function deconvolution techniques see Pesce, 
2010) Traditionally, these Ps RFs are calculated and then stacked along predicted 
phase arrival times over a range of crustal thickness (H) and crustal velocity (?! /?! 
ratio or ?) to create H-? stacks (Zhu and Kanamori 2000).  
 
In this study, we review traditional H-? stacking (section 3.1) and then introduce the 
approach for removing and correcting for shallow-layer reverberations (section 3.2), 
which is based on the resonance removal filter of Yu et al (2015). We then introduce 
and discuss metrics that allow us to determine when the shallow-layer corrections are 
needed (section 3.3). Once contamination from sediment reverberations has been 
removed, the complementary Sp (section 3.4) and SPmp (section 3.5) data are added 
to form a SRTC H-?-?! triple stack (section 3.6). This SRTC H-?-?! stack explicitly 
removes the need to assume a crustal velocity, and thereby improves confidence in 
our estimates of crustal thickness and velocity. 
2.2.1 Ps receiver functions in traditional H-?-VP stacks 
The direct P-to-s conversion across the Moho (Pms) is often used to constrain the 
Moho depth. The Moho depth estimate can be improved by including the multiple 
later arriving P-to-s primary reverberations, PPms, and PSms+ PmsPms whose ray paths 
are shown in Figure 2.1A and relative arrival times are shown in Figure 2.1B. The 
expected arrival time of the direct conversion and primary reverberations relative to 
the direct P wave arrival depend differently on crustal thickness (?) and crustal 
velocity (?! and ?!) given a ray parameter (?) of the teleseismic P wave:  
 25 
 
 
 
2.1  ? ? ? ? ? ? ??!!! = ? ?
!!
! ? ?! ?  ? ? !! ? ?!!  ? ?  
2.2  ? ? ? ? ? ? ? ?? !! ! !! !!!!! = ? ?! ? ? +  ? ?! ? ?  ? ?  
2.3  ? ? ? ? ? ??!!!! + ?
!! !
!!!!!! = 2? ?! ? ?  ? ? ? 
Each Ps receiver function, ?!(?), corresponds to incident P waves of ray parameter, 
?!, where ? is the index of the individual RF. By assuming H, ?, and an average 
crustal velocity, ?!, eqns. 1-3 can be used to compute the expected arrival times of 
the direct conversions and reverberations. The weighted sum of all N Ps receiver 
functions obtained at a seismometer are evaluated at the arrival times expected across 
a range of trial H and ? values to yield an H-? stack:   
 
!
2.4  ? ? ? ? ? ? ? ??!" ?, ? =  ? ? ? ?!?! ?!!! + ?!?! ?!!!! ? ?!?! ?!!!! + ?!!!!!!  
!!!
 
Where ?!, ?!, and ?!are the relative weights of the three phases. The most likely H 
and ? beneath the seismic station will be associated with largest amplitudes of the 
weighted sum, ?!" using the same weight values as Zhu and Kanamori (2000) where 
?! = 0.7, ?! = 0.2 and ?! = 0.1.  
 26 
 
 
PmsPms Pms PSms PPms
S wave
P wave
Distance from station (km)
PSms+PmsPms
Pms PPms
2 4 6 8
Relative Arrival Time (s)  
Figure 2.1- A. Ray-path geometry and B. synthetic receiver functions of P-to-s 
 phases converted across the Moho at 30 km depth, from an incident plane 
 wave of horizontal slowness 0.06 s/km. Crustal VP = 6.3 km/s, VS = 3.7 km/s, 
 while mantle VP = 8 km/s and VS = 4.6 km/s. Arrival times of the direct Moho 
 P-to-s conversion and first-order multiples are relative to the arrival time of 
 the direct P wave.  
    
In theory, Ps receiver functions spanning a range of ray parameters should contain 
information needed to constrain ?! on their own when stacked over a range of H, ?, 
and ?! values (Kumar and Bostock, 2008; Bostock and Kumar, 2010). Therefore Ps 
RF stacks are calculated over a range of average crustal ?! values creating a Ps RF H-
?-?! triple stack, which is calculated in eq. (2.7) and is called ?!" ?, ?,?! , where the 
absolute maximum should correspond to the most likely average crustal values 
beneath the seismic station. In practice, when shallow-layer reverberations are 
present, it is necessary to first remove their signature before determining ?!; this 
procedure is discussed procedure in Section 1.3.2. Furthermore, noise in the receiver 
 27 
 
 Amplitude
 
functions and the range of incident ray parameters can prevent robust estimation of 
?! , ?requiring the incorporation of independent data constraints; these independent 
constraints are discussed in Sections 1.3.3 and 1.3.4. 
2.2.2 SRTC Ps receiver function H-?-VP stacks  
When sedimentary layers are present, energy from conversions produced across the 
impedance contrast at the base of the sediment and reverberations within the sediment 
can get trapped within the low-velocity layer. These conversions and reverberations 
can produce large amplitude, long duration, and oscillatory (ringy) signals, which can 
directly overprint smaller amplitude signals from the Moho. The ray paths for the 
direct sediment conversion Pbs, and reverberations PPbs, PSbs, Pbs-2S, PPbs-2S, PSbs-
2S, Pbs-4S, etc. are shown in Figure 2.2A, and relative arrival times are shown in 
Figure 2.2B. In Ps receiver functions, the largest amplitude, longest duration 
oscillatory reverberations are those that contain two S-legs in the sediment (the S-
wave reverberations: PSbs, Pbs-2S, PPbs-2S, Pbs-4S, etc. Figure 2.2B), this is due to 
fact that Ps receiver functions are rotated, and the larger ?? ??? reflection coefficient 
at the base of the layer, meaning the reverberations at longer times contain primarily 
S-wave energy (Figure 2.3). Even though the PPbs phase arrival has large amplitude, 
it would only interfere with the Pms Moho conversion when the crust is extremely 
thin (<20 km) and the sediment is both thick and unreasonably slow (Figure 2.4). 
Therefore, to interpret later arriving Moho signals in the Ps receiver function, the 
periodic S-wave sediment reverberations must be removed.   
 28 
 
 
 
Figure 2.2-  A. Ray-path geometry of P-to-s phases converted across the base of a 
 0.5 km thick sediment layer, along with the largest-amplitude first- and 
 second-order multiples for an incident plane wave of horizontal slowness 
 0.0789 s/km. B. Synthetic receiver functions computed for a model with a 0.5 
 km thick sediment layer and horizontal slowness 0.06 s/km (VP = 2.5 km/s, VS 
 = 1 km/s), crustal VP = 6.2 km/s, VS = 3.5 km/s, a Moho at 35 km depth, and 
 mantle VP = 8 km/s, VS = 4.6 km/s. Direct conversions are in black, while the 
 oscillatory pattern is attributed to sediment multiples, labeled in red, blue and 
 purple. Orange dots indicate where the expected arrival of phases with more 
 than 2 P-wave legs would arrive.  
 
PP  PP SS  SS
0.8
4 4
0.6
3 3 0.4
0.2
2 2
0
2 2.5 3 2 2.5 3
VP/VS of sediment (km/s) VP/VS of sediment (km/s)
 
Figure 2.3- Absolute value of the reflection coefficients of PP*PP (left) and SS*SS 
 (right) for a range of different sediment velocity variables using equations 
 from Aki and Richards (1980) Because each reverberation will need to be 
 29 
 
VP of sediment (km/s)
VP of sediment (km/s)
 
 multiplied by an PPPP or SSSS reflection coefficient, the reverberations will 
 get smaller, but the PPPP reverberation will get smaller faster, and SS*SS 
 reverberations will remain visible in the Ps receiver function (See Figure 2.2).  
 
 
Travel time of Pms - PPbs 
Travel time of PPbs in sediment Crustal Thickness 20km Crustal Thickness 30km
2 8 2 2 2
6 1
1.5 1.5 1.5
4 0
1 1 1
2
0.5 0 0.5 0.5
1 2 3 1 2 3 1 2 3
sediment thickness (km) sediment thickness (km) sediment thickness (km)
 
Figure 2.4-  The SRTC H-k-VP stacking method may not work when the PPbs and 
 Pms phase arrival intersect. (Left) Calculated PPbs arrival time from sediment 
 with varying thickness and VS. The VP/VS ratio changes according to the mud 
 line equation in Brocher (2005). (Right) the Pms ?PPbs arrival time for a crust 
 with VP = 6 km/sand VP/VS = 1.76. Area where the PPbs arrival may interfere 
 with the Pms Moho arrival is shown in light blue to red. SRTC stacking may 
 not work for an extremely thin crust of 20 km, if sediment thickness is greater 
 than 1.5 km with excessively low VS of 0.5 km/s.  
 
The S-wave sediment reverberations can be suppressed from the Ps RF by applying a 
resonance removal filter proposed by Yu et al. (2015). The S reverberations have a 
resonant frequency associated with the two-way travel time of the S-wave in the 
sediment layer. A resonance removal filter can be constructed using the travel time of 
the S reverberation in sediment, ??, and the relative strength of the S-wave 
reverberation, r0 While these values can be inferred directly from the Ps RF, both ?? 
and r0 can be more reliably measured on the autocorrelation of the receiver function. 
In sediment, the autocorrelation of the Ps RF will have a decaying sinusoid pattern, 
showing a large, negative peak with amplitude r0, at time lag ?? (Figure 2.5A). A 
consistent method for determining these parameters, even when the autocorrelation is 
 30 
 
Vs of sediment (km/s)
 Travel time of PPbs (s)
Vs of sediment (km/s)
Travel time of Pms - PPbs 
 
complex with multiple minima (which suggests multiple shallow layers) involves 
finding the best-fit decaying sinusoid to the autocorrelation function of the form:  
 
2.5  ? ? ? ? ? ? ? ?? ? = ??!!"cos ? !" , 
?!
 
where t is the lag time of the autocorrelated RF, and the three parameters sought 
through the fitting procedure are the half-period of the oscillation, ??, ?the amplitude 
of the autocorrelation at zero lag time, ??, and the decay constant ?. Because the half-
period of the oscillation is precisely the travel time of the S reverberation, ? ?? =
 ??!, both parameters of the resonance removal filter can be estimated (Figure 2.5A). 
While these parameters should depend somewhat on ray parameter, the lateral 
proximity of the S reverberation bounce points to the station for the range of 
teleseismic ray parameters means that this dependence is in practice small enough 
that computing a single best-fitting set of parameters across all ray parameters is 
justified (at least in the idealized synthetic case of Figure 2.6). For example, when 
sediment is 2 km thick with a ??!= 2.5 km/s and ?! = 2.1 km/s, the PSbs reflection for 
an event with epicentral distance of 90 degrees may occur just 0.19 km from the 
station, while that for an event with epicentral distance of 30 degrees will occur 0.37 
km from the station. Therefore the best-fit decaying sinusoid is found to the 
autocorrelation of the mean Ps RF at the station. The benefit to using eq. (2.5) is that 
it can be easily automated and may simplify the interpretations of ??? in complicated 
or noisy auto correlated signals. Furthermore, calculating the resonance removal filter 
on the mean Ps receiver function at each station has the benefit of leveraging more 
 31 
 
 
waveforms to suppress noise and thereby stabilizes the inference of the optimal 
resonance removal filter parameters.  
 
Figure 2.5-  A. The autocorrelation of the mean synthetic receiver function (gray) 
 and the best fit decaying sinusoid (red) used for determining the parameters of 
 the resonance removal filter. The first large amplitude negative peak 
 corresponds to the two-way s travel time in sediment (?t) on the x-axis and 
 the amplitude of the sediment reverberations (r0) on the y-axis. B. Mean 
 synthetic Ps receiver functions before (gray) and after (red) the resonance 
 removal filter is applied; the filter successfully removes the reverberations and 
 the direct P to s conversion across the Moho becomes clear.  
 
Autocorrelated Receiver Function (RF) 
1 Autocorrelated RFs
Best-fit Decaying Sinusoid
0
0 2 4 6 8 10
Time (s) 
 
Figure 2.6- Dependence of resonance removal filter on ray parameter for an idealized 
 synthetic case. The best-fit reverberation removal filter (red line) calculated 
 from the autocorrelation of the mean synthetic receiver function provides a 
 good fit to the autocorrelations of Ps RFs computed across the range of 
 32 
 
RF amplitude
 
 possible ray parameters (black lines). The synthetic model used has a 
 sediment layer 0.5 km thick with a VP = 2.5 km/s and a VS = 1.7 km/s 
 overlying a 39.5 km crust with VP = 6.2 km/s and VS = 3.5 km/s and mantle 
 with VP = 8 km/s and VS = 4.5 km/s.    
 
Once the parameters ?? and r0 are found, the resonance removal filter can then be 
constructed in the frequency domain:  
 
2.6  ? ? ? ? ? ? ? ?? ? = 1+ r ?!!"?!!  
 
The resonance removal filter is applied in the frequency domain by multiplying it 
with the Fourier Transform of the Ps RFs. We verify that the reverberations are 
successfully suppressed from the resulting receiver function, clarifying the later 
arriving P-to-s Moho phase (Figure 2.5B).  
 
Although applying a resonance removal filter makes conversions from the Moho 
interpretable, it does not correct for the time delay that the Moho-related phases 
accumulate through the slow sedimentary layer. If the sediment time delay is not 
corrected when stacking the Ps receiver functions in the H-?-?! stacks, an incorrect 
crustal thickness estimate will be obtained (Yeck et al., 2013; Yu et al., 2015). 
Because slow sedimentary layers cause the assumed average crustal ?! to be lower 
than the actual ?!, the crustal thickness will be underestimated and the ?!/?! will be 
overestimated. 
 
 33 
 
 
To address this problem, the Moho phase arrivals used in the H-?-?! stack can be 
adjusted for the sediment time delay by using either the Pbs phase arrival, which will 
be the first large amplitude arrival on the receiver function (Yu et al., 2015), or the 
PPbs phase which is the largest amplitude arrival on the receiver function.  The arrival 
time of the Pbs phase (??) and the PPbs (???) phase can overlap, especially in lower 
frequency (1Hz) Ps RFs. It is easiest to identify these phases on high frequency (4 
Hz) Ps RFs. Using the high frequency RFs allows for tighter constraints on ?? or ??? 
and helps separate the Pbs phase arrival from the PPbs phase arrival when the 
sediment layer is thin (less than 0.5 km). In our automated procedure, we found it was 
easier to measure the arrival time of the larger phase ??? instead of the first arriving 
phase ??. ?Once ??? is determined, the predicted Moho phase arrival times are 
accounted for in computing the time adjusted H-?-?! stack:  
 
2.7  ? ? ? ? ? ? ??!" ?, ?,?!
!
= ?!?! ?!!! + ?? ? ??? + ?!?! ?!!!! + ???
!!!
? ?!?! ?!!!! + ?!!!!!! + ??  
 
When sedimentary layers are present, eq. (2.7) corrects for the sediment delay time 
and yields an estimate of the sub-sediment crustal thickness. Using the mud-line 
equation from Brocher et al. (2005), which is appropriate for unconsolidated 
sediments, we assume a relationship between ?! and ?! in sedimentary layers. This 
relationship can be used to approximate the ?!, ?!, and thickness of sedimentary or 
shallow layer from the measurements of ?? and ???. ? To further constrain the crustal 
 34 
 
 
thickness and crustal velocity, we add the Sp and SPmp phase arrivals (sections 2.2.4 
& 2.2.5) 
2.2.3 Determining when to correct for sedimentary layers 
Applying a sediment removal filter and correcting phase arrival times on the Ps 
receiver function may introduce error if the data do not exhibit pronounced sediment 
reverberations. To avoid this, a set of criteria based on quantitative metrics is defined 
to decide whether the sediment removal filter and arrival time corrections should be 
applied.  
 
The first criterion that needs to be satisfied for the sediment removal filter to be 
applied is that the variance of the differences between the sediment removed Ps RF 
and the original Ps RF (v1) be larger than the variance of the differences between the 
autocorrelation of the original Ps RF and the resonance filter (v2). If v1 is larger than 
v2, the resonance filter fits the autocorrelation of the receiver function well (small v2) 
and applying the resonance removal filter changes the resulting RF (large v1). The 
second criterion is that the amplitude of the PPbs arrival, ?(???), is at least 30 percent 
of the largest amplitude signal in the receiver function. Alternatively, the sediment 
removed receiver functions are used if the direct sediment conversion is 90 percent of 
the largest amplitude signal in the receiver function, regardless of v1 or v2. If the 
direct sediment conversion is small, the sediment reverberations should not mask 
deeper arrivals, but if the direct sediment conversion is very large, then reverberations 
may be overprint deeper signal. If these criteria are met, then the sediment removal 
 35 
 
 
filter is applied and delay time corrected to create the SRTC H-?-?! stack.  By 
applying a single set of criteria, SRTC H-?-?! stacking can be automated for stations 
across various tectonic/geologic settings without having to decide a priori whether 
the station is significantly affected by sediment. 
 
2.2.4 Sp receiver functions in H-?-VP stacks 
To further constrain crustal properties beneath a seismic station, it is helpful to add 
the complimentary Sp receiver function. Like Ps RFs, S-to-p converted waves 
reverberate between the Moho and the free surface, producing multiples. The S-to-p 
conversion across the Moho occurs further from the station (Figure 2.7) and the S 
wave contains lower frequencies than the P wave and so Sp RFs are not used as often 
as the Ps RFs to constrain crustal properties. However, the Sp phase arrives before the 
direct S wave arrival and is not directly affected by sedimentary layers as the Ps 
phase.  While the direct Smp conversion across the Moho and its reverberations have 
been used to constrain crustal thickness and velocity (Rychert and Harmon, 2016), we 
find that signals from sediment multiples arriving immediately after the main S phase 
are difficult to isolate from source-time function complexity and distinguish from the 
noise; therefore, we choose to only use the direct Sp phase to constrain the crustal 
thickness.  
 36 
 
 
Smp Pms S wave
P wave SPmp
150 100 50 0 50 100 150 200
Distance from Station (km)
SPmp
x 10 Smp
6
4 0
2
0
1 2 3 4 5 6 7 0 5 10
Relative Arrival Time (s) Relative Arrival Time (s)
 
Figure 2.7? A. Ray-path geometry of S to p converted phases across the Moho at 30 
 km depth calculated for an incident plane wave of 0.14 s/km and the post 
 critical P wave reflection across the 30 km deep Moho (SPmp) with incident 
 plane wave of 0.1486 s/km. The Ps conversions from the Moho are plotted for 
 comparison (gray line). B. Synthetic receiver function of the S-to-p phases 
 converted across the Moho. Time is relative to the direct S arrival (positive 
 values indicate earlier arrivals). C. Synthetic receiver function of the Hilbert-
 transformed post-critical P phase at the Moho. Time is relative to the direct S 
 arrival (positive values indicate later arrivals). Crustal VP = 6.3 km/s, VS = 3.7 
 km/s , while mantle VP = 8 km/s and VS = 4.6 km/s. 
 
To compare, and eventually combine, the Sp and Ps RFs the Sp RFs polarity is 
flipped so that a conversion across an impedance increase with depth corresponds to a 
positive phase, and then the Sp RF is time-reversed so Smp phase arrives after the 
direct S. In this convention, the relative arrival time of the direct Sp conversion 
depends on the crustal thickness (?), crustal velocity (?! and ?!), and S-wave ray 
parameter (?!") in exactly the same way as Ps arrival time (eq. 1). The sum of all Sp 
receiver functions (?!") computed at a station, ?!! (?), and evaluated at the expected 
 37 
 
Amplitude
Amplitude
 
arrival times for a range of crustal thickness (H), ?!/?! ratios (?), and average 
crustal ??! yields the expression analogous to eq. (2.4):  
!!"
 ? ? 2.9  ? ? ? ? ??!!! ?, ?,?! = ?
!
! (?!!!) 
!!!
 
If the Ps RFs indicate the presence of a sufficiently thick shallow layer (see section 
2.2.3), the delay time of the converted P-wave in sediment compared to the direct S-
wave is estimated to be ?? (same as the Pbs time above) by assuming vertical 
incidence in the sediment and correct for it when constructing the Sp H-?-?! stack: 
 
!!"
2.10  ? ? ? ? ? ?? !!!! ?, ?,?! = ?! ?!!! + ?? ? ???  
!!!
 
Stacking the direct Smp conversions on their own does not result in a maximum at a 
single, optimal H-?-?! combination; rather, ?!!! ?yields a range of possible H, ?, and 
?! values. This is nevertheless beneficial because traditional Ps RF H- ?? stacks with 
noisy data can be characterized by spurious maxima arising from constructive 
interference of noise or multiples that do not correspond to correct crustal H and  ?? 
values. Therefore, in addition to helping to constrain ?!, the strength of including the 
complementary Sp stack is that it significantly restricts the range of possible maxima, 
especially regarding crustal thickness (H). To further improve the ability to determine 
crustal thickness, ?!/?!, and especially average crustal ?!, constraints from the SPmp 
conversion are incorporated.  
 
 38 
 
 
2.2.5 SsPmp receiver functions in H- ? -VP stacks 
The SPmp phase provides constraints complementary to those from Sp and Ps RFs 
since it is sensitive to crustal thickness and average crustal ?! . The SPmp phase 
involves a post-critical reflection of the P-wave at the Moho which is produced when 
ray parameters of incoming teleseismic S-waves are sufficiently large, and are greater 
than the maximum ray parameters used in constructing typical Sp stacks (e.g. Wilson 
et al., 2006). As a result, the horizontal location of the conversion point is far away 
from the station (> 50 km) (Figure 2.7A). The raw SPmp waveform contains 
complexity due to the earthquake source-time function and source-side propagation 
effects. Therefore, just as for the Sp RF, these source effects are suppressed by 
deconvolving the parent (S) component from the daughter component (P) and 
computing an SPmp receiver function. Unlike the direct Sp phase, SPmp arrives after 
the direct S-wave, and so it can be combined with the Ps RF without being flipped in 
time. Despite arriving after the direct S-wave, the SPmp RF is less sensitive to 
sediment reverberations (Parker et al. 2016). In regions where the assumed average 
crustal ?! is inaccurate and sedimentary layers may mask Moho signals on the Ps RF, 
the large amplitude SPmp phase has proven useful in obtaining crustal thickness 
estimates (e.g. Langston, 1996; Owens and Zandt, 1997; Yu et al., 2012; Parker et 
al., 2013; Yu et al., 2013; Tian et al., 2015; Parker et al., 2016; Kang et al., 2016).  
 
The SPmp RF should not be directly incorporated into a H-?-?! stack because the 
post-critical reflection produces phase shifts that result in waveform distortion. We 
remove these distortions by computing the envelope of each SPmp receiver function 
 39 
 
 
(Bracewell, 1968). The expected phase arrival time of the SPmp conversion relative to 
the direct S wave arrival depends only on ?, ?!, and the ray parameter, ?!"#$%:  
 
2.11  ? ?? !! !!"!!! = 2? ?! ? ?!"!!!  
 
Because the post critical P wave reflection occurs for events within a small range of 
epicentral distances, fewer data are available for SPmp RFs than for Ps and Sp RFs. 
So that SPmp arrives at positive lag times, we phase weight stack the SPmp RFs in ray 
parameter bins, and require at least 3 events in each bin (Schimmel and Paulssen, 
1997), yielding a set of ?!!! (?). As with Sp and Ps RFs, a H-?-?! stack can be 
constructed by summing all available ?!!! ?  ?(?!) at expected arrival times, ??!"#$% eq. 
(11), for a range of H, ?, and ?!values:  
 
!!
 ? 2.12  ? ? ? ? ??!"!!! ?, ?,?! = ?
!!
! (?!"!!!) 
!!!
 
Note that this stack is identical for all values of ?. If the analysis of Ps receiver 
functions indicates that sediment corrections are necessary (see section 2.2.3), we 
correct the SPmp travel time by again assuming vertical incidence within the 
sedimentary layer. The time-corrected H-?-?! stack is given by:  
!!
 ? 2.13  ? ??!"!!! ?, ?,?! = ?
!!
! (?!"!!! + 2??? ? ??) 
!!!
 
 40 
 
 
The additional constraints provided by the SPmp RFs, when combined with Ps and Sp 
stacks, will allow us to constrain ?!, as well as H and ? beneath a seismic station even 
when thick sediment layers are present. 
 
2.2.6 Calculating combined SRTC H- ? ?VP stacks with error 
The final step in creating the SRTC H-?-?! triple stack is to determine the best 
method to combine the Ps, Sp and SPmp SRTC H-?-?! ?stacks and quantify error. 
Each stack is normalized by its absolute maximum, and the triple stacks (Ps, Sp, and 
SPmp) are combined by adding them to create a joint SRTC H-?-?! ?triple-stack:  
 
2.14  ? ? ? ? ? ?? ?, ?,?! =  ? ???(?, ?,?!)+  ???? ?, ?,?! + ???!?(?, ?,?!), 
 
where the absolute maximum of the stack will be the best fit H, ?, and ?! ?crustal 
parameters to the Ps, Sp, and SPmp RFs. This volumetric stack can be visually 
represented on three planes cutting through the maximum of the triple stack (see 
Figure 2.8).  
 41 
 
 
A Mean Ps RF B Traditional Ps H-  stack C Ps H-  stack with
sediment removed
2 1 2 1
Ps RF before sediment
 removal filter
0
Ps RF after sediment 
removal filter
0 0
0 2 4 6 8 10 20 30 40 50 60 20 30 40 50 60
Time after P arrival (s) Thickness (km) Thickness (km)
D H-?-V stack at V E H-?-V stack at H F H-?-V stack at ?  
P P P P 
2 1 2 1 60 1
50
40
30
0 0 20 0
20 30 40 50 60 6 6
Thickness (km) Average V (km/s) Average V (km/s)
P P  
Figure 2.8? SRTC H-?-?! stacking of synthetic receiver functions, where negative 
 values in the stack are clipped. Input model has a 0.5 km thick sedimentary 
 layer, V! of 2 km/s, and ? = 2.3, and a Moho at 37 km depth with crustal V! = 
 6.3, and ? = 1.76. A. The mean Ps receiver function before (black) and after 
 (red) the sediment removal filter is applied. B. Ps H-? stack at assumed V! = 
 6.3 shows no clear maxima. C. The Ps H-k stack after the sediment removal 
 filter and time correction is applied shows a clear maximum. D, E, & F. Three 
 cross-sections through the maximum of the H-?-V! ?triple stack; black cross 
 hairs denote 1? error bars. 
 
To quantify the uncertainty of the estimates of optimal H, ?,?! values, we look at the 
spread of the stack amplitude along its three axes. We interpret stack amplitude as 
proportional to log-likelihood. Therefore, ignoring higher order terms of the Taylor 
series expansion of the stack around the maximum, the curvature of the stack can be 
related to the posterior covariance matrix representing uncertainty of our parameter 
estimates. Specifically, the posterior covariance matrix is given by the negative 
inverse of the second-derivatives matrix evaluated at the triple stack maximum (Sivia, 
2006): 
 42 
 
Average V /V RF amplitude
P S
Average V /V Average V /V
P S P S
Thickness (km) Average V /VP S
 
 
?!? ?!? ?!? !!
?! ! ! ??
! ????! ????
! ?!!! ?!" !
! ! ! ? ? ?
!? ?!?
2.15  ? ? ? ? ? ? ? ? ? ?!!! ?!! ?!!! =  ?? !  
?! ! !
????! ??! ??!??
!" ?!!! ?! ?!? ?!? ?!?
???? ??!?? ??!
 
Where ? is the triple stack (from eq. 2.14) amplitude at the optimal H, ?,?! values. 
The original paper introducing the H-? stacking method (Zhu and Kanamori, 2000) 
proposed calculating the uncertainty of H and ? estimates made from the stack by 
dividing stack variance by the second derivative for H and ?, which ignores the co-
variance terms. This corresponds to computing the uncertainty on each parameter at 
the preferred value of the other parameter(s). Due to tradeoffs among parameters 
captured in the off-diagonal terms of the covariance matrix, however, this estimate is 
a lower bound on the true uncertainty. The uncertainty estimates represented by the 
variances in eq. (2.15), on the other hand, are a more appropriate measure of 
uncertainty since they capture additional uncertainty arising from our ignorance about 
the true values of the other parameters. It should be noted that even for traditional H-? 
stacks, the error should be calculated including the co-variance terms. 
 
2.3 Receiver Function Data for H- ? ?VP Stacks 
We evaluate the effectiveness of the H-?-?! stacking method for constraining crustal 
?!, ?!/?! and thickness using synthetic receiver functions as well as data recorded at 
three EarthScope Transportable Array (TA) stations, which span various geologic and 
 43 
 
 
tectonic settings, but all suffer from shallow layer contamination. The stations are: (1) 
B30A located in Edmore, ND on the eastern edge of the Williston basin; (2) 448A in 
Bay Minette, AL located within the Mississippi embayment; and (3) U60A in 
Pendleton, NC within the Atlantic coastal plain (ACP) Figure 2.9.  
  50 oN
  45 oN
  B30A40 oN
  35 oN
U60A
  30 oN
  25 oN 448A
 120 oW o 108o
W
W 2  96oW   84
oW   7
Figure 2.9 -Locations of USArray TA stations B30A, U60A, and 448A used in this 
 study shown as teal circles.  
 
To calculate Ps RFs for each station, we obtain 300 second three-component 
waveforms around the P arrival time for events with Mw>5.6 at epicentral distances 
between 30 and 90 degrees from the station. (Abt et al., 2010). Sp receiver functions 
are further restricted to events with epicentral distances between 55 and 90 degrees 
from the station that occur at depths less than 300 km (Wilson et al., 2006). SPmp 
waveforms are restricted to events with epicentral distances between 30 and 50 
degrees (Yu et al., 2013). A summary of the number of Ps, Sp and SPmp events, along 
with the slowness range for each station is shown in Table 2.1.  
 44 
 
 
 
Table 2.1-Number of events and slowness range for each seismic station used in this 
study    
# of 
Station Geologic 
# of Ps Ps # of Sp Sp SPmp 
Setting RF slowness RF slowness 
SPmp slowness 
events range (?) events range (?) RF events range (?) 
B30A Williston 167 0.0475- 49 0.1002- 37 0.1276-Basin, MT 0.774 0.1196 0.1411 
Atlantic 
U60A Coastal 120 0.0487- 0.1001- 0.1262-
Plain, NC 0.0772 
113 0.1201 8 0.1409 
Mississippi 
448A Embay., 36 0.0504- 0.1006- 0.1293-
MS 0.773 
81 0.1201 8 0.1339 
 
The automatic quality-control procedure is detailed in Abt et al. (2010). In this study, 
we culled the dataset to have a minimum Z-to-R cross correlation of 0.3, and a 
maximum difference of 25s between the automatically determined arrival time and 
prediction for ak135. The minimum signal to noise ratio is set to 0 for Ps and Sp 
receiver functions, while for SPmp RFs we require that the Hilbert transform of the 
SPmp arrival is greater than 1.2 of the S-wave arrival amplitude. We follow the Abt et 
al. (2010) procedure for calculating RFs, using a free-surface transform matrix 
(Kennett, 1991) and finding the surface velocity that minimizes the parent amplitude 
(P for Ps and S for Sp) on the daughter component to project the waveforms onto the 
P-SV-SH system, pick the arrival times, and apply a fourth order Butterworth band-
pass filter to waveforms of 0.03-1 Hz and 0.03-4 Hz for Ps, and 0.03-0.5 Hz for both 
Sp and SPmp. We then use the iterative time domain deconvolution with Gaussian 
half-amplitude half-width of about 0.12 s 0.5 s and 1 s for 4 Hz, 1 Hz, and 0.5 Hz 
respectively. to calculate the receiver functions (Ligorria and Ammon, 1999).  
 45 
 
 
 
To calculate synthetic receiver functions, the synthetic seismograms are generated 
using the flat-layer reflectivity algorithm ANIREC (Levin and Park, 1997) for a range 
of slowness and backazimuths. Then a fourth order Butterworth band-pass filter is 
applied to waveforms (0.03-1 Hz and 0.03-4 Hz for Ps, and 0.03-0.5 Hz for both Sp 
and SPmp) and the parent waveform is deconvolved from the daughter waveform 
using the Wiener frequency deconvolution (Wiener, 1964; Langston 1979).  
 
We calculate the triple stacks over a range of possible crustal H, ? and ?!, values 
where H ranges from 20 to 60 km, ? ranges from 1.5 to 2, and ?! ranges from 5.6 to 
6.8 km/s.  
2.4 Results and Observations from the SRTC H- ? ?VP Method 
We show the results of SRTC H-?-?! stacking applied to synthetic waveforms and 
three EarthScope TA stations with a low velocity sediment layer. We compare the 
estimates of crustal thickness before and after the SRTC H-?-?! stacking method is 
used and also to crustal thickness estimates in these regions from previous studies. 
 
Because time corrections are made to each phase to account for the slow sedimentary 
layer, the average crustal thickness, ?!/?!, and ?! at each station will correspond to 
the crust below the sediment. To determine the Moho depth from our study and 
compare with other studies, the thickness of the sediment should be added to the 
crustal thickness estimates from the SRTC H-?-?! stacks. We estimate the sediment 
thickness at each station using the measurements of ?? and ??? ?and assuming a 
 46 
 
 
relationship of ?! and ?! relationship from the mud-line equation (Brocher et al., 
2005). We find that the sediment thickness agrees within 0.1 km of the shallowest 
sediment layer from the CRUST 1.0 model (Laske et al., 2013). We interpret this 
shallowest layer of sediment as the unconsolidated sediment thickness at each station. 
At station TA B30A, we find a sediment thickness of 0.3 km, which agrees with 
estimates of 0.3-0.5 from multiple datasets  (AAPG, 1967; Laske et al., 2013). At 
station U60A we find a sediment thickness of 0.78, similar to the estimate from 
CRUST 1.0 of 0.9 km (Laske et al., 2013), but much thicker than the estimate of 0.3 
km interpolated from drill hole data using Bouger gravity maps (Lawerence and 
Hoffman, 1993). This discrepancy in sediment thickness values at U60A could arise 
from differences in interpreting what constitutes the shallowest sedimentary layer; 
alternatively, they could be due to differences in the velocity used to determine the 
thickness of the sedimentary layer.  Finally, at station TA 448A we calculate a 
sediment thickness to be 0.46 km, which lies in between the estimates of 0.6 km from 
CRUST 1.0 (Laske et al., 2013) and 0.339 for nearby station Y46A from horizontal to 
vertical ambient noise measurements (Langston and Horton, 2014). Tighter 
constraints and better uncertainty estimates can be achieved by either using higher 
frequency receiver functions (e.g. Leahy, 2012) or by stacking receiver functions for 
predicted sedimentary multiples (e.g. Yeck et al., 2013).  However, since the goal of 
this study is accurate estimates of total crustal thickness, the rough estimation of 
sediment thickness from the travel time is sufficient, as the uncertainty of the 
resulting crustal thickness estimates are smaller than those achievable with the H-?-
?! stacking method.  
 47 
 
 
2.4.1 SRTC H-?-VP stacking with synthetic waveforms 
The model used for calculating the synthetic waveforms has a sedimentary layer of 
thickness (H) of 0.5 km, VP of 2.3 km/s, and ? of 2.1 underlain by 36.5 km thick layer 
more representative of crystalline rock, with VP = 6.4 km/s and ? = 1.76. Before 
carrying out H-?-?! stacking, how the application of the sediment removal filter 
affects the appearance of mean Ps RFs relative to the results of traditional H-? 
stacking is compared. (Figure 2.8 A-C). 
 
The mean Ps RF computed from the synthetic waveforms is dominated by large 
amplitude sediment reverberations (black line, Figure 2.8A). After the sediment 
reverberation removal filter is estimated and applied (eq. 2.6), the mean Ps RF shows 
one clear arrival at ~5 s (red line, Figure 2.8A) corresponding to the Ps conversion 
across the Moho, but which was masked by sediment reverberations before the 
application of the resonance removal filter. 
 
Ps RFs computed with and without the resonance removal filter yield dramatically 
different results when used for traditional and SRTC H-? stacking, respectively. To 
illustrate this, traditional H-? stacks computed using the original Ps RFs are plotted 
(black line - Figure 2.8A) and compared to SRTC H-? stacks computed using the 
resonance removed Ps RFs (red line - Figure 2.8A) at an assumed crustal VP of 6.3 
km/s. While the traditional Ps H-? stack shows no clear Moho maximum (Figure 
2.8B), one might be interpreted at H = 42 km and ? = 1.63 (Figure 2.8B). SRTC H-? 
stacks, on the other hand, show a much stronger maximum at H = 37 km and ? =1.79 
 48 
 
 
(Figure 2.8C), which is nearly identical to the model values used to calculate the 
synthetic waveforms. 
 
Having validated that the SRTC H-? stacks are much better at resolving the input 
crustal structure in the presence of a shallow low-velocity layer, complementary Sp 
RFs are added, as sediment reverberations will not contaminate them. If the sediment 
removal filter is not successful in removing all of the amplitude or there are multiple 
basin layers with phase arrivals overprinting Moho arrivals, the Sp H- ? -VP stacks 
can help clarify crustal thickness. Finally, the SPmp RFs are introduced to remove the 
dependence on an assumed average crustal VP and better constrain the crustal 
thickness (H). A combined SRTC H-?-?! stack is computed (eq. 2.14) for the same 
input model by adding the weighted time-corrected Sp (eq. 2.10) and SPmp (eq. 2.13) 
RFs to the SRTC Ps stacks, Figure 2.8. Unlike H-? stacking, which can be fully 
visualized on a 2D contour plot, the H-?-?! stack varies in amplitude along three 
axes; therefore, slices through the stack volume are shown. In Figure 2.8D, when the 
Sp and SPmp RFs are added, again there is one clear maximum in the SRTC H-?-VP 
stack (Figure 2.8D-E). The crustal parameters are H = 41.25 ? 2 km, ?! = 6.3 ? 0.3 
km/s, and ? = 1.75 ? 0.09.  
 
To investigate the dependence of the SRTC H-?-?! stacking results on noise, this 
analysis is repeated for increasing amounts of white noise (filtered to have similar 
frequency content as typically encountered noise) with the maximum amplitude of the 
noise being 75% of the maximum amplitude of the daughter waveform (P for Sp and 
 49 
 
 
S for Ps) added to all of the waveform components. The H, ?, and ?! values of the 
input model within error are still able to be resolved (Figure 2.10). This illustrates the 
ability of the SRTC H-?-?! stack to resolve the three crustal parameters from 
synthetic waveforms persists even under realistic noise levels. It also demonstrates 
that when a sedimentary layer is present, the sediment-removed, time-corrected Ps 
RFs, time-corrected Sp RFs, and time-corrected SPmp are all needed to constrain 
parameters in the Ps H-?-?! stacks. Additionally, the sediment thickness is roughly 
estimated to be H = 0.508 km, which is consistent with the input sediment thickness 
value. Therefore, both the sediment and the crustal structure of the synthetic model 
are successfully retrieved using the SRTC H-?-?! stacking method.  
 50 
 
 
N = 0.25    H = 42?3.58  ? = 1.84?0.15  Vp = 6.4?0.37 
Traditional Ps H-?  stack H-?-Vp Stack at Vp = 6.4 H- ?- Vp Stack at H = 42 H-?-Vp Stack at ? = 1.842 60 1
1.9 50 0.8
1.8 0.6
40
1.7 0.4
1.6 30 0.2
1.5 0
20 30 40 50 60 20 30 40 50 60 5.6 5.8 6 6.2 6.4 6.6 6.8 205.6 5.8 6 6.2 6.4 6.6 6.8
Thickness (km) Thickness (km) Average Vp (km/s) Average Vp (km/s)
N = 0.5                     H = 42.75?3.0  ? = 1.86?0.1  Vp = 6.35?0.27 
Traditional Ps H-?  stack H-?-Vp Stack at Vp = 6.35 H- ?- Vp Stack at H = 42.75 H-?-Vp Stack at ? = 1.86
2 60 1
1.9 50 0.8
1.8 0.640
1.7 0.4
1.6 30 0.2
1.5 20 0
20 30 40 50 60 20 30 40 50 60 5.6 5.8 6 6.2 6.4 6.6 6.8 5.6 5.8 6 6.2 6.4 6.6 6.8
Thickness (km) Thickness (km) Average Vp (km/s) Average Vp (km/s)
N = 0.75                   H = 44.75?3.0  ? = 1.8?0.09  Vp = 6.25?0.42 
Traditional Ps H-?  stack H-?-Vp Stack at Vp = 6.25 H- ?- Vp Stack at H = 42.75 H-?-Vp Stack at ? = 1.8
2 60 1
1.9 50 0.8
1.8 0.6
40
1.7 0.4
1.6 30 0.2
1.5 20 0
20 30 40 50 20 30 40 50 605.6 5.8 6 6.2 6.4 6.6 6.8 5.6 5.8 6 6.2 6.4 6.6 6.8
Thickness (km) Thickness (km) Average Vp (km/s) Average Vp (km/s)
 
Figure 2.10? H-?-?! stacks generated from synthetic waveforms with increasing 
 noise levels. The input model has a sedimentary layer 0.5 km thick, V! of 2.3 
 km/s and ? = 2.3, a Moho at 42.5 km depth with crustal V! = 6.3 km/s, and a ? 
 of 1.76. White noise is filtered to have similar frequency content as typically 
 encountered noise, with amplitude up to three-quarters of the maximum of the 
 daughter waveform (s for Ps; p for Sp and SPmp). At increased noise levels, 
 we are still able to constrain crustal H, ?, and ?! within error. 
2.4.2 TA station B30A: eastern Williston Basin 
Having validated the SRTC H-?-?! stacking method on synthetics, we apply it to data 
recorded at a station where reverberations from shallow sediments can be expected. 
EasrthScope Temporary Array (TA) station B30A lies on the eastern edge of the 
Williston basin in northeastern Montana. The Ps RFs were calculated using the best-
 51 
 
Average Vp/Vs Average Vp/Vs Average Vp/Vs
Average Vp/Vs
Thickness (km) Thickness (km) Thickness (km)
 
fit rotation velocities of VP = 5.3 km/s and VS=2.9 km/s.  The measurement of ??? is 
made on the 4Hz Ps RFs (Figure 2.11) and is. 0.6 s.  The measurement of ?? at 
station B30A is 1.17 s and is measured from the autocorrelation of the 1Hz Ps RF. 
However, there no difference between the 1Hz and 4Hz RF for this measurement 
(Figure 2.11). At station B30A, the arrival time of ?t does not depend on event 
location or frequency of autocorrelated RFs.  
Station B30A
2
4 Hz Ps RF 
1 Hz Ps RF 
0
0 2 4 6 8 10
Time after P arrival (s)
1
4 Hz Ps RF autocorrelation 
0 1 Hz Ps RF autocorrelation  
1 Hz Ps RF autocorrelation  
?t
0 2 4 6 8 10
Lag Time (s)
 
Figure 2.10?  (Top) Mean 4 Hz (blue) and 1Hz (black) Ps RF for station B30A 
 shown with picked arrival time for ?tP and ?t shown in the red and magenta 
 circles respectively. The arrival time of the two phases (Pbs and PPbs) likely 
 overlap at this station. (Bottom) Autocorrelation from the mean 4Hz (blue) 
 and 1Hz (black) RF with picked arrival time for ?t. Autocorrelation of a 
 random subset of individual RFs are shown as gray lines in the background. 
 
The Ps RF before the sediment removal filter is dominated by large amplitude, 
oscillatory signals due to S-wave reverberations within the low velocity shallow layer 
(black line, Figure 2.12A, Figure 2.13). From the measurements of ?? and ???, 
sediment thickness is estimated to be 0.37 km. The traditional Ps H-? stack has a 
maximum at H = 28.5 km and ? = 1.62 (Figure 2.12B), which would suggest 
anomalously thin crust and a crustal ?!/?! representative of quartz sandstone (>57% 
 52 
 
Ampltidue RF Amplitude
 
SiO2) to the base of the crust (Christensen, 1996). After the sediment removal filter is 
applied to the Ps RFs, the amplitude of the receiver function decreases with 
increasing lag-time (red line ? Figure 2.12A), and the oscillatory behavior is 
suppressed. The SRTC Ps H-? stack shows a stronger maximum, indicative of a 
better fit to the direct conversions and multiples, which occurs at a greater depth, H = 
40 km and a higher ?!/?!, ? = 1.77 (Figure 2.12C). When complimentary Sp and 
SPmp RFs are added to create a SRTC H-?-?! stack, the maximum occurs at H = 
44.25 ? 2.63 km, ?! = 6.75? 0.43 km/s, and ? = 1.76? 0.09. (Figure 2.12D-E). By 
applying the SRTC H-?-?!triple stacking method to this station, a thicker continental 
crust is found with a more realistic average ?!/?! ?? consistent with a generally more 
mafic crustal composition (e.g. diorite or amphibolite per Christensen, 1996) 
compared to traditional H-? stacking, and estimate average crustal ?! ?instead of 
having to assume it a priori.  
A Mean Ps RF B Traditional Ps H-  stack C Ps H-  stack with
sediment removed
0.8 2 1 2 1
Ps RF before sediment
 removal filter 1.9 0.8 1.9 0.8
0.4
1.8 0.6 1.8 0.6
0
1.7 0.4 1.7 0.4
1.6 0.2 1.6 0.2
Ps RF after sediment 
removal filter
1.5 0 1.5 0
0 5 10 15 20 20 30 40 50 60 20 30 40 50 60
Time after P arrival (s) Thickness (km) Thickness (km)
D H- V stack at V = 6.75 E H- V stack at H = 44.25 F H- V stack at   = 1.76
P P P P 
2 1 2 1 60 1
1.9 0.8 1.9 0.8 0.8
50
1.8 0.6 1.8 0.6 0.6
40
1.7 0.4 1.7 0.4 0.4
30
1.6 0.2 1.6 0.2 0.2
1.5 0 1.5 0 20 0
20 30 40 50 60 5.6 6 6.4 6.8 5.6 6 6.4 6.8
Thickness (km) Average V (km/s) Average V (km/s)
P P  
 53 
 
Average V /V RF amplitude
P S
Average V /V Average V /V
P S P S
Thickness (km) Average V /VP S
 
Figure 2.12 ? SRTC H-?-?! stacking of B30A receiver functions, where negative 
 values in the stack are clipped. Station B30A is located on the eastern edge of 
 the Williston Basin in Montana. A. The mean Ps receiver function before 
 (black) and after (red) the sediment removal filter is applied. B. Ps H-? stack 
 at assumed V! = 6.3 km/s shows a maximum of the stack at H = 28.5 km and 
 ? = 1.62. C. The Ps H-k stack after the sediment removal filter and time 
 correction is applied; the stack maximum now corresponds to larger H and ? 
 values D, E, & F Three cross-sections through the maximum of the H-?-VP 
 triple stack; black cross-hairs denote 1? error bars. ? 
Station B30A
Ps RF before sediment
 removal filter
Ps RF after sediment 
removal filter
0
0 5 10 15 20
Time (s) after P arrival 
Station U60A
0
0 5 10 15 20
Time (s) after P arrival 
0
0 5 10 15 20
Time (s) after P arrival 
 
Figure 2.13 - 1Hz Ps receiver functions before (black) and after (red) sediment 
 removal filter is applied plotted from 0 to 20 seconds after the direct P arrival. 
 (Top) station B30A (Middle) station U60A (Bottom) station 448A.   
 
Previous studies near the location of TA station B30A estimate the Moho depth to be 
between 38 and 45 km (Cook et al., 1981; Langston, 1994; Chulick and Mooney, 
2002). Those values agree within error with our estimate for crustal thickness of 
44.25? 2.63 km (or a Moho depth of 44.72 km). However, the EarthScope Automated 
Receiver Survey (EARS), which uses a modified version of the traditional H-? stack, 
 54 
 
Ps RF amplitude Ps RF amplitude Ps RF amplitude
 
finds a much lower estimate for the crustal thickness of 30 ? 5.8 km (Crotwell and 
Owens, 2005). The ?!/?! (?) from the SRTC H-?-?! triple stack is 1.76? 0.09, while 
the EARS estimate of ? is much lower at 1.65? 0.07. The likely explanation for the 
inconsistency between these two results is that by implementing the sediment 
removal filter and adding Sp and SPmp stacks, the amplitude of the shallower 
maximum is reduced (Figure 2.12B) and the deeper maximum is highlighted. 
(Figure 2.12C-F). The average crustal P-wave velocity (?!) of 6.75 ? 0.43 km/s is 
consistent within error of previous results where the velocity is between 6.53 and 6.6 
km/s (Langston 1994; Chulick and Mooney, 2002; Laske et al., 2013) and agrees with 
the higher VS found under the Williston Basin (Schulte-Pelkum et al. 2017). The 
consistency of the three crustal parameters with previous studies gives us confidence 
that the H-?-?! triple stacking method is valid in this region.  
 
2.4.3 TA station U60A: Atlantic Coastal Plain 
Due to shallow sedimentary layers, one particularly challenging region for receiver 
function interpretation has been the Atlantic coastal plain. EarthScope TA station 
U60A lies on the eastern North Carolina portion of the Atlantic coastal plain. The Ps 
RFs were calculated using the best-fit rotation velocities of VP = 6 km/s and VS= 3.4 
km/s.  The measurement of ??? is made on the 4Hz Ps RFs (Figure 2.14, Figure 
2.13) and is. 0.2 s.  The measurement of ?? at station U60A is 0.8 s and is measured 
from the autocorrelation of the 1Hz Ps RF. However, there is slight difference 
between the 1Hz and 4Hz RF for this measurement (Figure 2.14). The Ps RF before 
the sediment removal filter for station U60A has a large amplitude oscillatory signal 
 55 
 
 
between at lag times <5 seconds due to shallow layer reverberations (black line, 
Figure 2.15).  From the measurements of ?? and ???, the sediment thickness is 
estimated to be 0.8 km and calculate a sediment removal filter.  When the filter is 
applied to the Ps RFs, the amplitude of the early arriving oscillatory phases is reduced 
and a peak appears in the sediment removed Ps RF (red line, Figure 2.15A) at 
approximately 4 seconds. In the traditional Ps H-? stack calculated with the Ps RFs 
before the resonance removal is applied, no clear maxima appear deeper than 20 km 
(Figure 2.15B), and no realistic constraint on crustal thickness and ?!/?! can be 
placed. However, when the SRTC Ps H-? stack calculated with the resonance 
removed Ps RFs, a deeper maximum becomes clear at H = 30.5 km and ? = 164. 
When the time-corrected Sp and SPmp RFs are incorporated in the SRTC H-k-Vp 
triple stack, a maximum is found at H = 31.75?2.28 km, ?! = 6.0?0.38 km/s, and ? = 
1.65?0.09 (Figure 2.15D-E), which suggests thinner and slower average crust than 
that present beneath the Williston basin., the sediment removal filter and the SRTC 
H-?-?! triple stack allows meaningful constraints to be placed on average crustal 
properties. Even when the sediment multiples completely mask the crustal signal in 
the Ps RF; the SRTC H-?-?! stack finds a reasonable maximum associated with the 
crust where none existed in the traditional H-? stack. 
 56 
 
 
Station U60A
2
4 Hz Ps RF 
1 Hz Ps RF 
0
0 1 2 3 4 5
Time after P arrival (s)
1
4 Hz Ps RF autocorrelation 
0 1 Hz Ps RF autocorrelation  
1 Hz Ps RF autocorrelation  
?t
0 2 4 6 8 10
Lag Time (s)
 
 
Figure 2.11?  (Top) Mean 4 Hz (blue) and 1Hz (black) Ps RF for station U60A 
 shown with picked arrival time for ?tP and ?t shown in the red and magenta 
 circles respectively. The arrival time of the two phases (Pbs and PPbs) likely 
 overlap at this station. (Bottom) Autocorrelation from the mean 4Hz (blue) 
 and 1Hz (black) RF with picked arrival time for ?t. Autocorrelation of a 
 random subset of individual RFs are shown as gray lines in the background. 
 At station B30A, the arrival time of ?t does not depend on event location and 
 only changes slightly with frequency of autocorrelated RFs.  
 
 
A Mean Ps RF B Traditional Ps H-  stack C Ps H-  stack with
sediment removed
2 1 2 1
0.4 Ps RF before sediment
 removal filter 1.9 0.8 1.9 0.8
0.2
1.8 0.6 1.8 0.6
0
1.7 0.4 1.7 0.4
1.6 0.2 1.6 0.2
Ps RF after sediment 
removal filter
1.5 0 1.5 0
0 5 10 15 20 20 30 40 50 60 20 30 40 50 60
Time after P arrival (s) Thickness (km) Thickness (km)
D H- V stack at V = 6.0 E H- V stack at H = 31.75 F H- V stack at   = 1.65
P P P P 
2 1 2 1 60 1
1.9 0.8 1.9 0.8 0.8
50
1.8 0.6 1.8 0.6 0.6
40
1.7 0.4 1.7 0.4 0.4
30
1.6 0.2 1.6 0.2 0.2
1.5 0 1.5 0 20 0
20 30 40 50 60 5.6 6 6.4 6.8 5.6 6 6.4 6.8
Thickness (km) Average V (km/s) Average V (km/s)
P P  
 57 
 
Average V /V RF amplitude
P S
Ampltidue RF Amplitude
Average V /V Average V /V
P S P S
Thickness (km) Average V /VP S
 
Figure 2.12? SRTC H-?-?! stacking of U60A receiver functions, where negative 
 values in the stack are clipped. Station U60A is located on the Atlantic 
 Coastal Plain in North Carolina. A. The mean Ps receiver function before 
 (black) and after (red) the sediment removal filter is applied. B. Ps H-? stack 
 at assumed V! = 6.3 km/s shows no clear maxima. C. The Ps H-k stack after 
 the sediment removal filter and time correction is applied has a clear 
 maximum. D, E, & F. Three cross-sections through the maximum of the H-?-
 VP triple stack; black cross-hairs denote 1? error bars. 
 
Most regional and active source studies of the Atlantic coastal plain that directly 
interpret their results in terms of the geology have taken place further south than TA 
U60A. There is a slight disagreement on the Moho depth in on the Atlantic coastal 
plain, where the Moho depth may be shallower, between 30-35 km (Cook et al., 
1981; Chulick and Mooney, 2002; Li and Fischer., 2002; Abt et al., 2010; Laske et 
al., 2013; Cook, 2016; Parker et al., 2016), or deeper, between 35-40 km (Crotwell 
and Owens, 2005; Lynner and Porrit, 2017). We find the crustal thickness at station 
U60A to be thin 31.75?3.52 km (a shallow 32.55 km Moho depth), agreeing with the 
first set of studies. We find a low with a ?!/?! (?) of 1.65?0.17 which agrees within 
error other estimates of VP/VS in the Atlantic coastal plain between 1.69-1.72 
(Musacchio et al., 1997; Parker et al. 2013). . We also find a slow average crustal ?! 
of 6.0?0.79 km/s compared to some studies which estimate that VP = 6.4-6.7 km/s 
(Laske et al., 2013) but that agrees with previous studies, which suggest the crustal P-
wave velocity is between 6.0 and 6.5 km/s. (Chulick and Mooney, 2002) The large 
error of the ?! estimate means that SRTC stacking at this station is not able to well 
resolve the average P-wave velocity at this station. 
 
 58 
 
 
2.4.4 TA statin 448A: Mississippi Embayment 
The final station on which SRTC H-?-?! triple stacks calculated is EarthScope TA 
station 448A, located within the lower Mississippi embayment. The Ps RFs were 
calculated using the best-fit rotation velocities of VP = 5.9 km/s and VS= 3.2 km/s.  
The measurement of ??? is made on the 4Hz Ps RFs (Figure 2.16A) and is. 0.85 s.  
The measurement of ?? at station 448A is 2.2 s and is measured from the 
autocorrelation of the 1Hz Ps RF. However, while there is little difference between 
the 1Hz and 4Hz RF for this measurement (Figure 2.16B), we do see two negative 
peaks in the 4Hz autocorrelation, which implies that there are multiple sedimentary 
layers beneath station 448A.  From the measurements of ?? and ??? we estimate 
sediment thickness to be 0.46 km at this station. Unlike the two previous stations, 
applying the sediment removal filter at 448A does not significantly change the 
amplitude or arrival time of phases in the mean Ps RFs (red line compared to black 
line, Figure 2.17A, Figure 2.13).  Therefore, one might expect that the traditional 
and SRTC Ps H-? stacks would be similar to SRTC stacks. Nevertheless, this is not 
the case, as implementing the time corrections (eq.7) significantly changes the 
coherence in the SRTC H-? stacks. While the traditional H-? stack has a maximum at 
H = 22 km and ? = 1.5 (Figure 2.17B), the SRTC Ps H-k ? stack (Figure 2.17C) has 
a deeper maximum at H = 35.5 km and ? = 1.67. While the sediment time corrections 
applied to the Ps RFs has a significant effect on improving the coherence, the 
multiple sediment layers that likely exist beneath station 448A may have sediment 
reverberations, which overprint Moho arrivals. Station 448A provides a good 
example on how the Sp and SPmp RFs can improve the interpretation of crustal values 
 59 
 
 
in a region with multiple sediment layers. When time-corrected Sp and SPmp RFs are 
incorporated, the crustal properties corresponding to the maximum of the SRTC H-?-
?! triple stack are H = 39.5?2.39 km, ?! = 6.5?0.29 km/s, and ? = 1.65?0.08. 
(Figure 2.17D-E). Station 448A highlights the importance of applying time 
corrections when calculating the H-? stack in the presence of a low velocity shallow 
layer, even if the Ps RFs remain similar before and after the sediment removal filter is 
applied.  
A station 448A
4 Hz Ps RF 
1 Hz Ps RF 
0
0 1 2 3 4 5
Time after P arrival (s)
B
1
4 Hz Ps RF autocorrelation 
0 1 Hz Ps RF autocorrelation  
1 Hz Ps RF autocorrelation  
?t
0 2 4 6 8 10
Lag Time (s)
 
Figure 2.13? (Top) Mean 4 Hz (blue) and 1Hz (black) Ps RF for station 448A shown 
 with picked arrival time for ?tP and ?t shown in the red and magenta circles 
 respectively. The arrival time of the two phases (Pbs and PPbs) likely overlap 
 at this station. (Bottom) Autocorrelation from the mean 4Hz (blue) and 1Hz 
 (black) RF with picked arrival time for ?t. Autocorrelation of a random subset 
 of individual RFs are shown as gray lines in the background. At station 448A, 
 the arrival time of ?t is challenging to pick, likely due to the multiple 
 sediment layers (multiple negative bumps seen in the 4Hz autocorrelation).  
 60 
 
Ampltidue RF Amplitude
 
A Mean Ps RF B Traditional Ps H-  stack C Ps H-  stack with
sediment removed
2 1 2 1
Ps RF before sediment
0.2
 removal filter 1.9 0.8 1.9 0.8
1.8 0.6 1.8 0.60.1
1.7 0.4 1.7 0.4
0
1.6 0.2 1.6 0.2
Ps RF after sediment 
removal filter
1.5 0 1.5 0
0 5 10 15 20 20 30 40 50 60 20 30 40 50 60
Time after P arrival (s) Thickness (km) Thickness (km)
D H- V stack at V = 6.5 E H- V stack at H = 39.5 F H- V stack at   = 1.65
P P P P 
2 1 2 1 60 1
1.9 0.8 1.9 0.8 0.8
50
1.8 0.6 1.8 0.6 0.6
40
1.7 0.4 1.7 0.4 0.4
30
1.6 0.2 1.6 0.2 0.2
1.5 0 1.5 0 20 0
20 30 40 50 60 5.6 6 6.4 6.8 5.6 6 6.4 6.8
Thickness (km) Average V (km/s) Average V (km/s)
P P  
 
Figure 2.14? SRTC H-?-?! stacking of 448A receiver functions. Station 448A is 
 located on the Mississippi Embayment in Mississippi. A. The mean Ps 
 receiver function before (gray) and after (red) the sediment removal filter is 
 applied, which do not appear significantly different at this station. B. Ps H-? 
 stack at assumed V! = 6.3, and the maximum of the stack is H = 24 km and ? 
 = 1.5. C. The Ps H-k stack after the sediment removal filter and time 
 correction is applied has a maximum at H = 36.25 km and ? = 1.83. D, E, & 
 F. Three cross-sections through the H-?-V! ?triple stack through the maximum 
 of the stack; black cross-hairs denote 1? error bars. 
 
Several previous studies of the lower Mississippi embayment near the location of TA 
station 448A find similar results to the crustal values obtained in the SRTC H-?-VP 
triple stacks. In southern Mississippi (slightly further west than station 448A) the 
Moho depth was found using seismic refraction to be  ~35 km with an average crustal 
?! of 6.4 km/s (Warren et al., 1966), shallower than the SRTC H-?-VP stack crustal 
thickness estimate of 39 km. A global Moho study finds that the Moho depth in this 
region is between 32 and 36 km with an average crustal ?! of 6.5 km/s (Cook et al., 
 61 
 
Average V /V RF amplitude
P S
Average V /V Average V /V
P S P S
Thickness (km) Average V /VP S
 
1981; Mooney et al., 1983). Finally, while there are no EARS results for station 
448A, the results for a nearby station, US BRAL, finds a Moho depth of 38 km, a 
?!/?! of 1.87, assuming an average crustal ?! (from Crust 2.0) of 6.53 km/s (Crotwell 
and Owens, 2005). At this station our crustal thickness estimate is closer to that of 
EARS estimates at station US BRAL, with a lower VP/VS. The low VP/VS. value 
found in this study would indicate that the crustal composition beneath this station 
would be felsic in composition, possibly quartzite or granite (Christensen, 1996).   
2.5 Discussion of H-?-VP Methodology  
By creating the SRTC H-?-?! stack, we address two of the major limitations of 
traditional H-? stacking: contamination by sediment layers and the a priori 
assumption of an average crustal ?!.  
 
In regions overlain by a sedimentary layer, such as the Williston basin, Atlantic 
coastal plain, and Mississippi embayment, sediment reverberations overprint deeper 
crustal conversions and traditional H-? stacking methods fail. Yeck et al. (2013) 
proposed the sequential H-? stacking approach, which characterizes the sediment 
?!/?! and thickness, and uses the result to correct for sediment effects and interpret 
Moho depth. By using the resonance removal filter and time corrections of Yu et al. 
(2015), we are similarly able to characterize and suppress the S basin multiples, 
which are likely to dominate the reverberation signal (Figure 2.3) before interpreting 
Moho depth. However, unlike sequential H-k stacking, the SRTC H-? stack is able to 
constrain crustal properties even when the S-basin multiple and Moho phase arrivals 
are coincident in time due to the sediment time corrections.  Unlike Yu et al. (2015), 
 62 
 
 
we find that in practice, it is easier to use the arrival time of the PPbs phase instead of 
the Pbs phase, due to its larger amplitude.   
 
Assuming an incorrect average crustal ??! may lead to biased estimates of the crustal 
thickness, even when no sedimentary layer is present (Yeck et al., 2013). The H-V 
stacking method of Rychert and Harmon (2016) uses the direct Sp arrival as well as 
Sp multiples (from the Sp RFs) to constrain average crustal ?!, ?!, and thickness, 
removing the dependence on average ?!. While this method is robust in areas without 
sedimentary reverberations, the low amplitude Sp multiples can make it difficult to 
use Sp RFs when substantial noise is present, which is often the case. For this reason, 
we choose to supplement the constraints provided by SRTC H-? by incorporating 
only the direct Sp conversion in our stacks. The Sp phase arrival is able to provide an 
additional constraint on the crustal thickness, when the sediment structure has 
multiple layers (such as at station 448A). To constrain ?! , the SPmp RF is also added, 
which has proven useful to constrain crustal thickness and velocity in a region where 
sediment is present (e.g. Parker et al., 2016).  
 
A significant concern for the SPmp phase is the effect of azimuthal anisotropy or 
dipping layers on the arrival time of the SPmp phase. SPmp is observed on a very 
restricted epicentral distance range (30-50 degrees), and so to look at the variation in 
arrival time based on back azimuth (BAZ), binned SPmp events are plotted over a 
range of BAZ bins for permeant seismic station ANMO used because it has events 
from a range of backazimuths. When plotted by BAZ, the SPmp phase does not show 
 63 
 
 
a strong directional dependence for station ANMO (Figure 2.18). This is especially 
significant as station ANMO lies near the Rio Grande Rift with the Colorado Plateau 
to the north and the basin and range to the south. The geologic regions have 
differences in crustal properties and thickness that might be expected to manifest as 
back-azimuthal variations in the SPmp phase arrival; however, that is not the case, and 
the SPmp phase shows a clear and consistent arrival from all BAZ bins with more than 
three existing events, suggesting that stacking SPmp observed at a station is justified. 
Similar concerns exist for Sp and Ps phases in H-? stacking, although the back-
azimuthal dependence of Sp due to anisotropy is expected to be very complex. In 
future studies, harmonic decomposition of Ps phases (e.g. Olugboji and Park, 2016) 
can be used to detect and isolate the signature of anisotropy and/or dipping layers, 
helping to constrain average crustal anisotropy. Additionally, extra care must be taken 
when considering SPmp RFs for stations near oceanic crust. Since the SPmp waveform 
is affected by distant crustal structure, and oceanic and continental crust differ 
dramatically in thickness, analysis should be restricted to include only events with 
back azimuths that sample the continental crust. 
 
 64 
 
 
SsPmp Phase at ANMO plotted by BAZ
0.6
BAZ bin range
60?120
0.4 120?180
180?240
0.2 300?360
0
?0.2
?0.4
?0.6
?0.8
?20 ?10 0 10 20 30
Time after S arrival (s)  
Figure 2.15? SPmp RFs for events between 30 and 50 epicentral distances. Each line 
 is a bin of backazimuthal (BAZ) directions from station ANMO, chosen due 
 to its large range in BAZ of SPmp events. The amplitude and location of the 
 SPmp at ~6 s does not change significantly based on BAZ direction which 
 indicates that the SPmp waveforms show evidence of neither crustal 
 anisotropy nor crustal thickness variations with BAZ. We expect this 
 observation to hold true for most other stations, due to the low frequency of 
 SPmp waveforms.  
 
 
The resonance removal filter will effectively remove S-wave reverberations 
associated with the low velocity layer in Ps RF. One concern is that contamination 
from additional sedimentary reverberations (such as PPbs reverberations) may still 
mask the crustal phase arrivals even after the resonance removal filter is applied. We 
find that if the sedimentary VS is very low (has an average velocity of less than 0.5 
km/s) and sediment is thick (around 3 km), while the crust is extremely shallow (less 
than 20km) it is possible for large amplitude PPbs to arrive at the same times as direct 
Pms conversion from the Moho and its reverberations to arrive directly after Moho 
phases. In these rare cases, the sediment P-wave multiples may stack coherently near 
the maxima, biasing the crustal thickness estimate 0.5-1.0 km deeper than input 
 65 
 
Amplitude
 
values and ?!/?! by 0.02-0.03 higher than input value and therefore, the SRTC H-?-
?!triple stack may be contaminated by PPbs and should not be used (Figure 2.4 
When the average sediment VS is larger or the sediment is not as thick, S-wave  
reverberations are the most dominant signal in the Ps receiver functions.(Figure 2.2, 
Figure 2.3 ).   
 
One limitation of the resonance removal filter is that it assumes a single shallow 
layer, and only removes reverberations from the largest amplitude resonances ? 
typically, the S-wave reverberations ? in the Ps RF. The way it has been implemented 
in this study, the resonance removal filter does not remove reverberations from 
multiple distinct shallow layers that may add further reverberations. While we find 
that removing the largest reverberations is sufficient to constrain the crustal thickness 
and crustal average crustal velocities (?!/?! and ?!) at these stations, applying a 
secondary resonance removal filter could further reduce oscillatory conversions 
associated with multiple layers. For example, in the case of multiple low velocity 
sediment layers, a secondary resonance removal filter, applied after the first, may be 
able to remove reverberations associated the layer bounded by second-largest 
impedance contrasts. In regions where S-wave reverberations of the whole crust are 
particularly prominent, an extension of this method may be to apply a secondary 
resonance removal filter to remove reverberations from the crust/mantle boundary 
(Moho) that can mask intra-lithospheric arrivals at approximately 7-10 seconds (about 
80-100km). 
 66 
 
 
2.6 H-?-VP Conclusions 
 
Traditional H-? stacking is commonly used to constrain crustal thickness and ?!/?! 
beneath a seismic station, but requires an assumed average crustal ?! and fails in the 
presence of a low velocity sedimentary layer. Previous studies have been able to 
remove sediment contamination (Yeck et al., 2013) or remove the dependence on 
average crustal ?! (Rychert and Harmond, 2016) with some success. We present the 
sediment-removed, time corrected (SRTC) H-?-?! stacking method that is able to 
remove sediment contamination even when multiples directly overprint crustal 
signals while also yielding constraints on average crustal ?!.  
 
The SRTC H-?-?! stacking method uses quantitative thresholds to determine whether 
large amplitude oscillatory sedimentary reverberations are present in the Ps RF. It 
then applies a resonance removal filter and time corrections from Yu et al. (2015) to 
the Ps RFs, and adds complementary Sp and SPmp RFs, with their respective time 
corrections. Synthetic data is used to verify that this method is effective in 
constraining crustal properties when sediment reverberations contaminate deeper 
crustal conversions. When applied to EarthScope TA stations in the presence of 
sediment, and find that overall, our estimates of H, ??, and ?! agree well with previous 
studies. At stations where traditional H-? stacking implies an unlikely crustal 
thickness and Vp/Vs ratio, a more reasonable estimate for the crustal properties is 
obtained after the SRTC H-?-?! stacking method is applied. Because using the H-?-
?! stacking method requires fewer a priori assumptions about the underlying crustal 
 67 
 
 
structure, including whether a sedimentary layer is present, it is straightforward to be 
applied to large seismic arrays regardless of geologic/tectonic regime.  
 
We show that both resonance removal and time corrections can reduce the complexity 
introduced into traditional H-? stacks by shallow layer reverberations. We ensure that 
error calculations capture uncertainty arising from our ignorance about the values of 
the other parameters. This allows us to better define and estimate error around one 
clear maximum in the triple stack. However, this method does not provide an efficient 
way to express error if multiple maxima are present. For a large-scale study using the 
SRTC H-?-Vp stacking, a complexity term based on the amplitude of the next closest 
maxima in the stack is suggested. In a regional study, when multiple maxima are 
present at a station, nearby stations can be used to constrain the correct crustal 
parameters in complex stacks, and a search range for H, ?, and Vp be based on 
existing crustal models near the seismic station, such as Crust 2.0 (Crotwell and 
Owens, 2005) can be implemented to constrain reasonable crustal values and reduce 
computation time.  
 
Though our method avoids making prior assumptions about average crustal ?!, which 
is necessary in traditional H-? stacking, ?! is often poorly constrained using this 
method, with error up to ? 0.6 km/s.  While the constraint on crustal ?! is sufficient 
for this study, we recognize that tighter constraints on crustal ?! ?may be necessary to 
for other applications, such as for drawing inferences about crustal composition. Even 
when including complementary Ps, Sp and SPmp data, we find that the data are not 
 68 
 
 
particularly sensitive to ?!, which is reflected in our estimated errors (between 0.3-
0.6 km/s). Where data from active source studies is available (e.g. Lithoprobe 
(Clowes et al., 1987; Cook et al., 1988; White et al., 1994; Ross et al., 1995) or 
COCORP (Cook et al., 1981; Allemdinger et al., 1983; Peterson et al., 1984) seismic 
lines), ?! ?can be constrained more tightly using them. Alternatively, joint inversion of 
receiver functions with surface wave data can better constrain crustal ?! (e.g. Julia et 
al., 2000; Lawerence and Wiens, 2004; Shen et al., 2013). Because our work suggests 
that removal of sediment multiples can clarify constraints from receiver functions, it 
is likely to improve results of such joint inversions.  
 
From the Ps receiver functions and resonance removal filter, we obtain information 
that can be used to constrain sedimentary layers. Specifically, the amplitude and 
travel time of the S-wave reverberation in the sediment that when combined with 
assumptions about the relationship of VP and VS ?may be used to estimate sedimentary 
thickness and velocity. Thick sedimentary layers lead to increased amplitude and 
duration ground shaking during a seismic event, so mapping the structure of 
sedimentary basins over large regions has implications for seismic hazard mapping. 
 69 
 
 
Chapter 3: Constraining Basin Fundamental Frequency and Velocity 
Profiles of the Atlantic Coastal Plain using Receiver Functions 
 
3.1 Introduction to Hazard in the Atlantic Coastal Plain   
Despite being situated on a passive margin, the East Coast of the United States has 
suffered from very damaging and costly earthquakes. Many densely populated East 
Coast cities that lie on Coastal Plain sediments contain old buildings and 
infrastructure (Figure 3.1). The large impedance contrast between Atlantic Coastal 
Plain (ACP) basins and underlying igneous or metamorphic bedrock cause incoming 
seismic waves to become amplified and trapped in the basin, leading to prolonged and 
damaging ground motion (e.g., Field et al., 1990; Fischer et al. 1995; Baise et 
al., 2016; Yilar et al. 2017). Amplification by sedimentary layers and matching 
fundamental frequencies of sedimentary basins and built structures is likely 
responsible for the substantial damage resulting from even moderate-sized 
earthquakes on the East Coast, such as the Mw 5.8 2011 Mineral, VA earthquake 
(Pratt et al., 2017). 
 
Mapping the fundamental frequency and amplification of basins is useful for 
geoengineering and hazard assessment as overlap between the fundamental resonant 
frequencies of basins and buildings can be particularly damaging (e.g. Flores et al., 
1987). Typically, the average velocity to 30 m depth (VS-30) is used as a proxy for 
site amplification in hazard analysis and ground motion prediction equations (e.g. 
Abrahamson and Silva; 2008). These site characterizations by VS-30 are typically 
used to inform seismic building codes (e.g. NEHRP Provisions). However, 
 70 
 
 
characterizing only the very shallow soil and sediment to 30 m may not be sufficient 
to describe damaging site effects, particularly for larger buildings and infrastructure. 
More recent studies argue that fundamental period (inverse of fundamental 
frequency), or fundamental period combined with a velocity term (e.g. VS-30, Z1.0, or 
Z2.5) is a better proxy for site amplification and response and reduces uncertainty in 
ground motion prediction equations (Luzi et al. 2011; Pitilakis et al. 2013; Zhao and 
Xu, 2013).  
 
Figure 3.1- USGS map showing the location of populations in red/black ontop of the 
intensity of potential earthquake ground shaking that has a 2% change of occurring in 
50 years (Figure from Jaiswal et al., 2015) 
 
 
 
 
 71 
 
 
High fundamental frequencies associated with soil and shallow sediment 
reverberations pose a problem to short buildings that may have matching resonant 
frequencies. However, the large impedance contrast between the bedrock and the 
basin in the ACP means that the peak amplification should occur at the basin 
fundamental frequency (Pratt et al. 2003; Narayan, 2010). Because basins within the 
ACP are relatively thick and wide when compared to West Coast (AAPG Basement 
Map of North America, 1978), the basin fundamental frequency in the ACP will be 
low and will influence shaking over larger areas. The low fundamental frequencies of 
the basin predominantly affect larger structures with matching resonant frequencies 
(e.g. bridges, tall buildings, industrial facilities), which is of particular concern for 
cities situated on the East Coast where large buildings and infrastructure were 
constructed before current seismic building codes were in effect. Thus, it is crucial for 
hazard analysis to constrain the amplification and fundamental frequencies across the 
ACP.  
 
Common methods employed to estimate fundamental frequency are the standard 
spectral ratio (SSR), and the ambient noise or earthquake horizontal-to-vertical 
spectral ratio (HVSR or E-HVSR respectively). The SSR technique requires 
comparing ground motions from local earthquakes at a target site located on sediment 
to a reference site with no sediment that overlies the same bedrock material. (e.g. 
Borcherdt 1970; Frankel et al., 2002; Pratt et al. 2017; Perron et al. 2018). 
However, good reference sites with the same underlying bedrock material may not be 
available for all ACP stations. Mismatch in the bedrock material between the target 
 72 
 
 
and reference sites will degrade the accuracy of the seismic hazard interpretation 
(Steidl et al., 1996). In contrast, the HVSR techniques do not require a reference 
station. Instead, HVSR isolates the horizontal response from the vertical through 
spectral division, as horizontally polarized S-wave will cause the most ground 
shaking during an earthquake (e.g. Nakamura, 1989; Lermo et al., 1993; Nakamura, 
2000; Mendeck et al., 2014; Pratt et al. 2017). In some cases where available data is 
sparse, such as in the central and eastern US, the resonance peak that is used to infer 
fundamental frequency may not be clear and may prove difficult to interpret (Pratt et 
al., 2017). In the Central and Eastern US dense high-frequency seismic surveys are 
rare, and so broadband seismic stations may be useful for constraining the 
fundamental frequency. However, with traditional techniques constraining the 
fundamental frequency of the Atlantic Coastal Plain basin may still prove difficult 
(Yassminh et al., 2019).  
 
Here, we propose, validate, and apply a different approach to constrain key basin 
properties by using high frequency P-to-s receiver function analysis. P-to-s receiver 
functions (PRFs) isolate the S-wave conversions across a large impedance contrast 
(product of velocity and density) through deconvolution, similar to H/V techniques. 
However, they can be analyzed in the time domain where individual arrivals are 
easily associated with S or P-wave reverberations. Although PRFs are traditionally 
used to constrain crustal and lithospheric structure, high-frequency PRFs are strongly 
sensitive to shallow impedance contrasts such as that between the basin and bedrock 
(Zelt and Ellis, 1999; Leahy et al., 2012; Yeck et al. 2013). Therefore, we propose to 
 73 
 
 
use measurements of converted S-waves and reverberations from the high-frequency 
PRFs calculated for EarthScope USArray Transportable Array (TA) broadband 
seismic stations to characterize the basin structure in the ACP.  
 
Due to the large impedance contrast between basin and bedrock material, strong S-
wave reverberations are present in the P-wave coda. Computing the autocorrelation of 
the PRFs allows us to easily measure the two-way S-wave reverberation time (TSs) in 
the basin (Yu et al., 2015), which is directly related to the fundamental frequency of 
the basin. The basin fundamental frequency is calculated from measurements of TSs  
at each broadband seismic station, allowing for a map of basin fundamental frequency 
across the ACP to be created.  
 
We find that PRFs are also useful in constraining other basin properties, including the 
P and S-wave velocity profiles. Using TSs, along with interpolated basin thickness 
(Figure 3.2), an average basin S-wave velocity profile of the ACP is inferred. By 
incorporating the arrival time of one additional phase, TPPbs, an average P-wave 
velocity profile is also obtained. We validate these velocity profiles using calculated 
reflection coefficients of the basin impedance contrast compared to amplitudes of the 
measured phase arrivals (PPbs and the S-wave reverberation), as the amplitude is 
related to the impedance discontinuity between basin and bedrock. 
 74 
 
 
BasinThickness  (km)
3.5
3
o N
 36 2.5 
2
o N
 33 1.5
1
o N
0
  3 0.5
0
 
Figure 3.2- Basin thickness for stations on the Atlantic Coastal Plain (Fenneman and 
 Johnson 1946) inferred from basement depth interpolated from AAPG 
 Basement Map of North America, 1978 
3.2 Receiver Function Data for the ACP 
High-frequency P-to-s receiver functions (PRFs) are calculated for 75 broadband 
seismometers from the EarthScope Transportable Array (TA) stations located on the 
Atlantic Coastal Plain of the United States (Fenneman and Johnson, 1946). To 
calculate PRFs, we use 300 s long windows of three-component waveform data 
(aligned so that the P arrival is at 75 s) from teleseismic events with Mw>5.6 and 
epicentral distances between 30? and 90? from each station. The events are quality 
controlled, extracting those with a minimum Z-to-R cross correlation of 0.3, and a 
maximum difference of 25s between the automatically-determined arrival time and 
prediction for ak135, with no requirement on minimum signal to noise threshold. (See 
Data and Resources, Abt et al., 2010). We use a free-surface transform matrix 
 75 
 
  39 o
N
o W o W o W o W
 75 8   7   8
1   8
4
 
(Kennett, 1991) and find the surface velocity that minimizes the parent amplitude (P 
for PRFs) on the daughter component to project the waveforms onto the P-SV-SH 
system. We pick the arrival times and apply a fourth order Butterworth band-pass 
filter to waveforms of 0.03-4 Hz.  We then use the iterative time domain 
deconvolution with Gaussian half-amplitude half-width of ~0.12 s to calculate the 
4Hz PRFs (Ligorria and Ammon, 1999). This is done following the automated 
procedure detailed in Abt et al. (2010).   
 
The 75 seismic stations within the ACP are further culled to only those that show 
clear sedimentary basin arrivals in the receiver functions. Following the guidelines 
laid out in Chapter 2, we require that the autocorrelation of the PRF can be 
sufficiently-well fit by a decaying sinusoid ? meaning that the PRF displays the 
oscillatory effect of reverberations trapped in a sediment basin ? and that amplitude 
of the PPbs conversion is at least 30% of the maximum amplitude in the signal. We 
find that 11 of the stations determined to be in the ACP are located on bedrock 
material with no significant sediment signal in the PRF. In the end, we obtain a 
dataset of 64 stations that meet these criteria, exhibiting visible sedimentary 
reverberations, and containing between 31 and 304 individual PRFs per station.  
3.3 Basin Fundamental Frequency  
In order to map the basin fundamental frequency across the ACP, we measure the 
two-way S-wave travel time in sediment from PRFs at each station. The large 
impedance contrast between the basin and basement material means that PRFs are 
particularly sensitive to s-wave reverberations. The conversions and reverberations 
 76 
 
 
produced from this impedance contrast appear as peaks on the receiver function 
corresponding to ray paths shown in Figure 3.3A and relative phase arrival times 
shown in Figure 3.3B. In PRFs, the largest amplitude, longest duration oscillatory 
phases are those that contain the two-way S-wave reverberation (PSbs, Pbs-2S,PPbs-
2S,PSbs-2S, PbS-4S, PPbs-4S, etc.), which dominate the PRF signal. The 
reverberations at longer times contain primarily S-wave energy due to fact that PRFs 
are computed from upgoing P and S waveforms estimated using the free-surface 
transform (Kennett, 1991), and because the ?? reflection coefficient at the base of the 
layer is large. While the S-wave reverberation time (TSs) and amplitude (r0) can be 
inferred directly from the PRFs, a more reliable method is to make this measurement 
on the autocorrelation of the receiver function (Yu et al., 2015; Chapter 2). When a 
low velocity basin layer is present, the autocorrelation of the receiver function will 
have a large negative peak at reverberation time TSs and amplitude ?r0. Because the S-
wave reverberations are nearly vertical, the mean PRF at each station can be used to 
calculate the autocorrelation. We validate this with a synthetic receiver function 
(Figure 3.4A) and show example station measurements (Figure 3.4B-C). 
 77 
 
 
 
Figure 3.3- A. Ray-path geometry of P-to-s phases converted across the base of a 0.5 
 km thick sediment layer, along with the largest-amplitude first- and second-
 order multiples for an incident plane wave of horizontal slowness 0.0789 
 s/km. B. Synthetic receiver functions computed for a model with a 0.5 km t
 hick sediment layer and horizontal slowness 0.06 s/km (VP = 2.5 km/s, VS = 
1  km/s), crustal VP = 6.2 km/s, VS = 3.5 km/s, a Moho at 35 km depth, and 
 mantle VP = 8 km/s, VS = 4.6 km/s. Direct conversions are in black, while the 
 oscillatory pattern is attributed to sediment multiples, labeled in red, blue and 
 purple. Orange dots indicate where the expected arrival of phase with more 
 than 2 P-wave legs would arrive. 
 
 
 
Synthetic RF Autocorrelation U60A RF Autocorrelation Q60A RF Autocorrelation
VP
VS
0 0
0
-r0 -r0 -r0
0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10
Lag Time (s) Lag Time (s) Lag Time (s)
TSs TSs TSs
 
Figure 3.4- TSs and r0 measurements from 4 Hz Ps receiver function 
 autocorrelations. With increasing sediment thickness, TSs increases. A. 
 Autocorrelation of Synthetic RF calculated for velocity model described in 
 78 
 
Amplitude
Amplitude
Amplitude
 
 top right corner. TSs and r0 are picked as the first negative large amplitude, 
 shown by the red dot. B. Autocorrelation of 4 Hz Ps RFs for station U60A 
 located on 0.137 km of basin material. Mean autocorrelation shown in 
 blue and a range of Ps RF autocorrelation events shown in gray. TSs and  r0 
 pick shown as red dot. C. Autocorrelation of 4 Hz Ps RFs for station Q60A 
 located on 1.07 km of basin material. Mean autocorrelation shown in blue and 
 a range of Ps RF autocorrelation events shown in gray. TSs and r0 pick shown 
 as red dot.  
Measurements of TSs and r0 are made for stations across the ACP. As expected, with 
increasing basin thickness from west to east, TSs increases (Figure 3.5A, Figure 
3.6A). Deeper basins will have faster velocities at the base, reducing the impedance 
contrast between basin and bedrock. Therefore, as basin thickness increases, r0 is 
expected to decrease, consistent with our observations shown in Figure 3.5B (Figure 
3.6B). 
A. B. C.
TSs  (s) r0  F0  (Hz)
3
0.8 0.8
0.7 0.7
2.5
o N N o N o N o N o
6 N6 0.6
  
3 3 6
   3 0.6 
2
0.5
0.5
o N N o N o N o N o
3 N
3 3 0.4 3
  1.5  3 3  0.4
0.3
1 0.3
o N o N o N o N o N o
0 N0 0
  3  3 0.2   3
0.2
0.5
0.1
 
Figure 3.5-  Measurements from the 4Hz Auto-correlated RFs across the ACP. A. 
 Measurements of TSs at each station across the ACP. As basin thickness 
 increases TSs increases. B. Measurements of r0 across the ACP. As basin 
 thickness increases, the r0 (related to the impedance contrast) decreases. C. 
 Fundamental Frequency which is related to TSs as F0 = ?  ?across the ACP.  
????
 79 
 
  3
9 o
N
  3
9 o
N
  3
9 o
N
o W o W o W o W
5 8 1 4
  
7  7  8    
8
o W o W o W o W
5 8 1 4
  
7   
7   
8   8
o W o W o W o W
5 8 1 4
 7 7 8       
8
 
Assuming that the S-wave reverberations are near vertical in the slow sedimentary 
basin, the relationship between TSs and basin fundamental frequency becomes quite 
simple. The basin fundamental frequency (F0) depends on the S-wave velocity (VS) 
and basin thickness (Hb) in the quarter wavelength approximation (Joyner et al., 
1981; Shearer and Orcutt, 1987; van der Baan 2009) as: 3.1  ??! = ?!/4?!. 
Assuming near-vertical incidence in the low velocity basin layer, two-way S-wave 
travel time in sediment (TSs) is: ? 3.2  ??!" = 2?!/?!, so that fundamental frequency is 
related to TSs by: 3.3  ??! = 1/2?!". Assuming vertical incidence introduces errors 
that are negligible compared to observational uncertainty; for example the 
approximation introduces an error of < 0.2% (3 ms for a 2 s travel-time) in TSs for a 
typical ray parameter of 0.05 s/km for the incoming P wave and a 1 km thick 
sediment with VS = 1 km/s. Transforming the station measurements of TSs for F0 
accordingly, we produce a map of site-specific fundamental frequency across the 
Atlantic Coastal Plain (Figure 3.5C) obtaining fundamental frequencies ranging from 
1.1 Hz in shallow basins to 0.16 Hz in deeper ones (Figure3.6C). 
 80 
 
 
A. B. C.
TSs vs Thickness   r0 vs Thickness F0 vs Thickness
3.5 0.8 1.1
1
3 0.7
0.9
2.5 0.6 0.8
2 0.5
0.7
0.6
1.5 0.4 0.5
0.3 0.41
0.3
0.5 0.2
0.2
0 0.1 0.1
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
Thickness (km) Thickness (km) Thickness (km)
D. E.
TPPbs vs Thickness PPb s vs Thickness
1.4 1.6
1.2 1.4
1 1.2
1
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
Thickness (km) Thickness (km)
 
Figure 3.6- Station measurements plotted vs. basin thickness. A.TSs increases with 
 increasing basin thickness B. r0 decreases with increasing basin thickness C.F0 
 decreases with increasing basin thickness D. TPPbs increases with increasing 
 basin thickness E. PPbs decreases with increasing basin thickness. 
 
3.4 Basin Velocity Profiles  
3.4.1 ACP Basin VS Profile  
When the basement depth and fundamental frequency is known, the S-wave velocity 
of the sediment can be found (e.g. Bodin et al., 2001; Parolai et al., 2002; 
Setphenson et al. 2019). Due to trade-offs between sediment thickness and velocity, 
inferring the average S-wave velocity structure of the ACP using measurements of 
TSs requires additional information and assumptions.  The most important additional 
information needed is that sediment thickness at each seismic station can be estimated 
 81 
 
TPPbs  (s) T  (s)Ss
  r
0
PPb s
F  (Hz)
0
 
from existing studies. Borehole data with estimates of basin thickness along the 
Atlantic Coastal Plain is available from AAPG (AAPG Basement Map of North 
America, 1978) and aggregates state-by-state surveys of basin thickness. The basin 
thickness data is interpolated from published contour maps to give an approximate 
sediment thickness at each of our station locations (Figure 3.2). Because basins 
within the ACP are relatively thick and wide containing nearly horizontal layers of 
sediment (DiPietro, 2018; Pratt, 2018), we make the assumption that sediment within 
the ACP at a given depth will have similar velocities across fairly wide geographic 
regions, allowing for the inversion of the average ACP basin velocity profiles instead 
of separate profiles beneath each station. Finally, we assume that the seismic velocity 
in the basin does not have internal interfaces and that VS increases smoothly with 
depth. These assumptions, along with measurements of TSs from receiver functions, 
allow us to estimate an average 1-D velocity profile for the ACP. 
 
To invert for the S-wave velocity of the ACP, we seek a single velocity profile that 
minimizes the misfit between predicted and observed TSs at all stations. We 
parameterize the velocity profile assuming a power-law dependence of VS with depth, 
which has been found to be a good description in the shallow subsurface (Delgado et 
al., 2000). An additional advantage of this parameterization is that multiple studies 
have computed generic surface wave dispersion and eigenfunction expressions for 
such velocity profiles (Godin and Chapman, 2001; Tsai and Atiganyanum, 2014). In 
order to explicitly account for uncertainty in our measurements, we use a Bayesian 
inversion framework. To minimize the influence of outliers on the inferred Vs 
 82 
 
 
profiles, we define an L1-norm misfit function ?!(?) for N stations: (3.4) 
! !"#!! !"#$
? ? = ! !" ?! !" ?! !"#! !!!  ?; where  ?!"  ?!  is the observed and ?
!"#$
!"  ?!  is the ?
model-predicted TSs at each station i, and ? is the standard deviation of the TSs error 
measurement. Based on variations observed using different subsets of data, we fix ? 
to be 0.1 s.  
 
The predicted TSs are calculated from eq.(2) using the S-wave velocity at each depth 
calculated as: 3.5  ??! ? = ? !! + ?!(?) , where ?! is the starting S-wave velocity at 
the surface, ? is the depth and ?!and ? are parameters which describe the shape of the 
power-law dependence of velocity with depth, and 0 < ?! < 2.7 and 0 < ? < 1 . We 
use a Markov Chain Monte Carlo approach to obtain best-fit parameter values and 
map out uncertainties and trade-offs among them. 
 
We find that the best-fit VS profile does not adequately predict the measurements of 
TSs for very thick basins, greater than 2.5 km to basement (Figure 3.7). The four 
stations that lie on top of very thick basins (454A, 553A, 554A, 555A) are within the 
South Georgia Rift Basin (SGR- Figure 1.1) (Wagner et al., 2018). We interpret the 
TSs picks at these stations to likely be coming from an intra-sedimentary interface 
with a large impedance contrast (such as between unconsolidated and consolidated 
sediment), and not from the basin/basement impedance contrast. Therefore, these 
points are excluded and the inversion is performed for stations with a basement depth 
of less than 2.5 km thick. We invert for the average profiles of the entire ACP, as well 
as for the Florida and Eastern Gulf Coastal Plain sediments (FEG) and Coastal Plain 
 83 
 
 
(CP) profiles, separately. The three resulting VS profiles are very similar (Figure 
3.8); with the parameters of the inversion also remaining similar (Figure 3.9). This 
justifies our assumption of geographic uniformity of velocity at a particular depth 
within the ACP, and produces an equation for VS velocity of the basin as: 
3.7  ?? ? = 0.46+ 0.53(?)!.!"! . ? ?Interestingly, ? remains close to 0.5 for all of the 
models, which implies a linear increase of shear rigidity with depth (Chapman and 
Godin, 2001).  
A. B.
TSs VS 
0 0
Observed TSs
Predicted TSs
0.5 0.5
1 1
1.5 1.5
2 2
2.5 2.5
3 3
3.5 3.5
Mean ACP VS
4 4
4.5 4.5
0 2 4 6 0 1 2 3 4
Time (s) VS (km/s)
 
Figure 3.7- VS velocity profile from inversion.  A. TSs measurements from and 4Hz 
 Ps RFs (cyan circles) compared to TSs predicted from the mean Vs profile 
 from our best-fit inversion (black crosses) B. VS velocity inversions using TSs 
 measurements and assuming VS increases in a power-law fashion with depth. 
 A range of velocity inversions using all stations is plotted in gray, and the 
 mean of these models is plotted in black 
 
 84 
 
Depth (km)
Depth (km)
 
A. B.
TSs VS 
0 0
0.5 0.5
1 1
1.5 1.5
2 2
Mean ACP VS
Mean FEG VS
Observed TSs Mean CP VS
Predicted TSs
2.5 2.5
0 1 2 3 4 0 1 2 3 4
Time (s)
VS (km/s)
 
Figure 3.1- VS velocity profile from inversion.  A. TSs measurements from and 4Hz 
 Ps RFs (cyan circles) compared to TSs predicted from the mean Vs profile 
 from our best-fit inversion (black crosses) B. VS velocity inversions using TSs 
 measurements and assuming VS increases in a power-law fashion with depth. 
 A range of velocity inversions using stations with basin thickness less than 
 2.5 km are plotted in gray, and the mean of these models is plotted in black. 
 The mean velocity inversion for the stations in the Florida and Eastern Gulf 
 Coastal Plain sediments (FEG) is plotted in red, and the mean velocity 
 inversion for the stations in the Costal Plain (CP) is plotted in blue. 
 
 
 85 
 
Depth (km)
Depth (km)
 
p a1 VS0
200
100 100
100 All Stations
0 0 0
0 0.5 1 0 0.1 0.2 0 0.2 0.4
100 100 100
ACP
0 0 0
0 0.5 1 0 0.1 0.2 0 0.2 0.4
200 200
200
100 100 FEG
0 0 0
0 0.5 1 0 0.1 0.2 0 0.2 0.4
 
 
Figure 3.2- Parameters from VS profile inversion (p, a1, and VS0) for all stations 
 with basement thickness less than 2.5; ACP stations; FEG stations. 
3.4.2 ACP Basin VP Profile 
Because the P reverberations within the sedimentary column are less prominent than 
S reverberations, constraining the average ACP VP profile from receiver function data 
is more difficult. Indeed, since P-to-S receiver functions isolate the S-wave 
conversions from subsurface interfaces, prominent phases always include at least one 
S-wave leg; therefore the timing of any phase arrival will depend on the S-wave 
travel time through the basin. Therefore, inverting for the VP profile requires not only 
making an arrival time measurement on a phase sensitive to the VP in the basin, but 
also involves the same data ? measurements of TSs ? and assumptions used for 
obtaining a VS profile. Here, we focus on PPbs, which is usually the largest amplitude 
phase arrival on the RF that contains a P-wave traveling in the sediment (Figure 
 86 
 
 
3.3B). Assuming near vertical incidence, the travel time of PPbs (TPPbs) depends on 
the basin thickness (Hb), P-wave velocity (VP), and basin S-wave travel time 
as ? 3.8  ?? !!!"# = ?!/?! + ?!".  The measurement of TPPbs and the PPbs phase !
amplitude is shown for a synthetic receiver function (Figure 3.10A) and for two 
example stations with varying sediment thicknesses (Figure 3.10B-C). It is important 
to note measurements of TPPbs depend on the frequency of the receiver function. For 
thin sediments, the large amplitude PPbs phase and the earlier arriving direct P-to-s 
sediment conversion, Pbs, may become indistinguishable from one another and 
overlap in time (Chapter 2). For this reason we use the high frequency 4Hz RF to 
measure ?!!"#.   
A. B. C.
Synthetic RF Autocorrelation U60A 4Hz Ps RF Q60A 4Hz Ps RF
0.2
PPbs VP [2.7 6.2 8] 0.137 km to basement 1.07 km to basement 
2
VS [1 3.5 4.5] 2
PPbs
0.1 z   [1 35 100] PPbs
0 0
0
TPPbs
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
TPPbs
Time after P arrival (s) Time after P arrival (s) Time after P arrival (s)
TPPbs
 
Figure 3.3- TPPbs and PPbs measurements from 4 Hz Ps receiver functions. With 
 increasing sediment thickness, TPPbs increases. A. Autocorrelation of 
 Synthetic RF calculated for velocity model described in top right corner. 
 TPPbs and PPbs are picked as the largest positive amplitude, shown by the red 
 dot. B. 4 Hz Ps RFs for station U60A located on 0.137 km of basin material. 
 Mean Ps RF shown in blue and a range of Ps RFs shown in gray. TPPbs and 
 PPbs pick shown as red dot. C. 4 Hz Ps RFs for station Q60A located on 1.07 
 km of  basin material. Mean Ps RF shown in blue and a range of Ps RFs 
 shown in gray. TPPbs and PPbs pick shown as red dot 
 
Measurements of TPPbs and the PPbs phase amplitude are made across the ACP. The 
 87 
 
Amplitude
Amplitude
Amplitude
 
same trends as with TSs are found. With increasing basin thickness from west to east, 
TPPbs increases (Figure 3.11A, Figure 3.6D). Deeper basins will have faster velocities 
at the base, reducing the impedance contrast between basin and bedrock. Therefore, 
as basin thickness increases, as expected, the PPbs phase amplitude decreases (Figure 
3.11B, Figure 3.6E). This systematic behavior gives us confidence in the travel time 
and amplitude measurements of PPbs made on high frequency RFs.  
A. B.
TPPbs  (s) PPbs  
1.5
1.2
o N o N
6 1 o N o N
  3   3
6 1
0.8
o N o N
3 0.6 o N o N
  3   3
3
0.5
0.4
o N o N
0 0.2 o N
o N
  3 0  3
0 0
 
Figure 3.4- Measurements from the 4Hz RFs across the ACP. A. Measurements of 
 TPPbs at each station across the ACP. As basin thickness increases TPPbs 
 increases. B. Measurements of r0 across the ACP. As basin thickness 
 increases, the amplitude of PPbs (related to the impedance contrast) decreases.  
The VP velocity profile is obtained assuming the same type of power law relationship 
with depth used for VS (Figure 3.12). We invert for the VP profile using all stations 
with basins thickness of less than 2.5 km, as well as separately for stations in the FEG 
and on the CP. We find that all three inversions prefer higher exponents (p) than for 
VS, and that the accepted models show more variation than those of the VS profile; 
 88 
 
  39 o
N
  39 o
N
o W o W o W o W
 75  78  81 4     8
o W o W o W o W
5 8 1 4
  7   7   8   8
 
both ?! and ? are poorly constrained by the data (Figure 3.13). Inverted VP profiles 
are compared to the VP profile predicted by the VS profile by the regression fit of 
Brocher (2005) (specifically, eq. (9) in that study), which is valid for sedimentary 
lithologies with S-wave velocities less than 4.5 km/s. The ?Brocher regression? VP 
profile (predicted from VS profiles inverted from our data) is significantly different 
from the VP profiles calculated with the power-law relationship. Therefore, we repeat 
the inversions for VP profiles fixing p = 1, i.e. assuming that VP increases linearly 
with depth. Assuming a linearly increasing VP with depth, i.e. (3.9) ?!(?) = ?! +
?!(?), where ?! is the VP at the surface, and ?! is the velocity gradient with depth, 
and 0 < ??! < ?2. We perform the inversion on the same station groups and find much 
better agreement between model solutions (Figure 3.14); elimination of p as a free 
parameter results in both ?! ?and ?! are well constrained (Figure 3.15) producing a VP 
profile follows the equation: 3.10 ?! ? = 1.99+ 1.95 ? . When the resulting 
linear VP profiles are compared to the VP profile obtained from applying the Brocher 
(2005) relationship to the VS profiles obtained previously, the differences are 
observed to be minimal.  
 89 
 
 
A. B.
TPPbS VP 
0 0
Observed TPPbs
Predicted TPPbs
0.5 0.5
1 1
1.5 1.5
2 2
Mean ACP VP
Mean FEG VP
Mean CP VP
2.5 2.5
0 0.5 1 1.5 2 4 5 6 7
Time (s)
VP (km/s)
 
Figure 3.5? VP velocity profile from inversion.  A. TPPbs measurements from and 
 4Hz Ps RFs (magenta circles) compared to TPPbs predicted from the mean VP 
 profile from our best-fit inversion (black crosses) and TPPbs predicted from the 
 Brocher (2005) relationship of VP and VS (green crosses) B. VP velocity 
 inversions using TPPbs measurements and assuming VP increases in a power-
 law fashion with depth. A range of velocity inversions using stations with 
 basin thickness less than 2.5 km are plotted in gray, and the mean of these 
 models is plotted in black. The mean velocity inversion for the stations in the 
 Florida and Eastern Gulf Coastal Plain sediments (FEG) is plotted in red, and 
 the mean velocity inversion for the stations in the Costal Plain (CP) is plotted 
 in blue, and the VP profile from the Brocher (2005) relationship is plotted in 
 green. The velocity profile is not well constrained by any of the data.  
 90 
 
Depth (km)
Depth (km)
 
p a1 VS0
200
200 100
100 All Stations
00 0.5 1 0 2 4 6 00 2 4
400
200 100
200 ACP
00 0.5 1 0 2 4 6 00 2 4
400 200
200
200 100 ME
00 0.5 1 0 2 4 6 00 2 4
 
Figure 3.6? Parameters from VP profile inversion (p, a1, and VS0) for all stations 
 with basement thickness less than 2.5; ACP stations; FEG stations  
 
A. B.
TPPbS VP 
0 0
Observed TPPbs
Predicted TPPbs
Predicted
 TPPbs from 
0.5 0.5
   Brocher
1 1
1.5 1.5
2 2 MMeeaann A ACCPP V VPP
MMeeaann M FE GV PVP
MMeeaann C CPP V VPP
BBrroocchheerr V VPP
2.5 2.5
0 0.5 1 1.5 2 0 2 4 6 8
Time (s) VP (km/s)
 
Figure 3.7? VP velocity profile from inversion.  A. TPPbs measurements from and 
 91 
 
Depth (km)
Depth (km)
 
 4Hz Ps RFs (magenta circles) compared to TPPbs predicted from the mean VP 
 profile from our best-fit inversion (black crosses) and TPPbs predicted from the 
 Brocher (2005) relationship of VP and VS (green crosses) B. VP velocity 
 inversions using TPPbs measurements and assuming VP increases in a linearly 
 with depth. A range of velocity inversions using stations with basin thickness 
 less than 2.5 km are plotted in gray, and the mean of these models is plotted in 
 black. The mean velocity inversion for the stations in the Mississippi 
 Florida and Eastern Gulf Coastal Plain sediments (FEG)is plotted in red, and 
 the mean velocity inversion for the stations in the Costal Plain (CP) is plotted 
 in blue, and the VP profile from the Brocher (2005) relationship is plotted in 
 green. The linear relationships for VP agree with each other, and are consistent 
 with the Brocher (2005) regression relationship.  
a1 VP0
1000
1000
500 All Stations
01.5 2 2.5 0 0 2 4
800
1000
400 ACP
01.5 2 2.5 0 0 2 4
1000 1500
1000
500 FEG
500
01.5 2 2.5 0 0 2 4
 
Figure 3.8? Parameters from linear VP profile inversion b1 and VS0 for all stations 
 with basement thickness less than 2.5; ACP stations; FEG stations  
 
3.5 Validation  
Measurements of basin fundamental frequency presented in this study agree with 
values observed in regions with similar basin thickness (e.g. Parolai et al. 2002; 
Pratt, 2018;).  For example, the basin fundamental frequency at station Y57A above a 
 92 
 
 
shallow basin (0.12 km) has a fundamental frequency of 1 compared to ~1 Hz 
predicted in Parolai et al. (2002) and ~1.1 Hz measured from HVSR in Pratt (2018), 
while station 356A above a thick basin (1.16 km) has a fundamental frequency of 
0.28 Hz compared to 0.3 Hz and from Pratt (2018). However, when shallow 
unconsolidated sediments are present within the basin, damaging amplification can 
occur at higher frequencies. Most studies of fundamental frequency in the ACP 
measure the large amplification at the fundamental frequency from these shallow 
layers, which ranges from 1.5-12 Hz (e.g. Fischer et al., 1995; Fairbanks, et al. 2008; 
Pratt et al., 2017). Amplification at these high frequencies from shallow 
unconsolidated layers may be large and damaging to shorter buildings and smaller 
infrastructure. However, in this study we focus solely on the peak amplification 
produced from basin/bedrock impedance contrasts, which may be larger and has 
implications for large buildings, and contributes to the overall basin shaking effect.  
  
We could not identify any studies in the ACP that systematically map fundamental 
frequency of the entire basin, as opposed to the shallow sedimentary layers. 
Therefore, we carry out two different analyses to validate that our measurements of 
TSs imply the same basin fundamental frequencies as would be obtained by traditional 
methods. We identify a set of teleseismic earthquakes that are recorded at three 
stations with increasing sediment thickness (Figure 3.16). We then compute radial 
and transverse component spectral ratios using a method similar to the E-HVSR 
technique, where the radial (R) and transverse (T) component multitaper power 
spectral density estimates are compared to the vertical (V) component multitaper 
 93 
 
 
power spectral density estimates event-by-event, where the time-halfbandwidth 
product is 4. Peaks in these ratios correspond to the resonant frequencies of the 
sediment and are shown in Figure 3.16. In addition, we compute a ?PRF spectral 
ratio? which is motivated by the SSR technique where multitaper power spectral 
density estimates at stations on sediment are compared event-by-event to the mean 
spectrum of stations on bedrock where the time-halfbandwidth product is 4. Here, we 
choose as our bedrock reference stations the 11 stations determined to be in the ACP, 
but that do not have significant sediment contributions, as pointed out in DATA. The 
?PRF spectral ratios? also exhibit peaks at frequencies corresponding to basin 
resonances and are plotted in Figure 3.16. For stations U60A, Q59A, and Q60A with 
basin thickness of 0.137 km, 0.534 km and 1.07 km respectively our TSs-based 
fundamental frequency estimates are 0.71 Hz, 0.39 Hz, and 0.26 Hz respectively. For 
these same stations, the E-HVSR and ?PRF spectral ratio? techniques demonstrate 
that the peak amplification occurs at congruent frequencies. As expected, the 
fundamental frequency using measurements of TSs is consistent with the more 
traditional methods, show increasing fundamental frequency with increasing sediment 
thickness.  
 94 
 
 
U60A S-Spectral Ratios Q59A S-Spectral Ratios Q60A S-Spectral Ratios 
F 4 F 4 F 4
0 00
80 0.04 60 0.06
80 0.137 km to basement 0.524 km to basement 1.07 km to basement
0.06
60 0.03
60 40 0.04
0.04
40
40 0.02
20 0.020.02
20 20 0.01
0 0 0 0 00
0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
Frequency (Hz) Frequency (Hz) Frequency (Hz)
R SR T SR 1 Hz SR 4 Hz SR
 Figure 3.9- Comparison of site fundamental frequency (F04) measurements 
 obtained in this study compared to more traditional spectral ratio techniques 
 of estimating site fundamental frequency from peak amplitude. HVSR 
 techniques applied to Radial (R, dark blue) and Transverse (T, cyan) 
 component data recorded at the seismic station. SSR techniques applied to 
 PRF data for create "PRF spectral ratios? using 1Hz (pink) and 4Hz (red) 
 PRFs. Method applied to three stations on differing basin thicknesses (U60A, 
 Q59A, Q60A).   
 
Using the S-reverberation and PPbs arrival-time measurements made on high-
frequency PRFs, we infer S and P wave velocity profiles for the ACP, for basin 
thickness thinner than 2.5 km. We find that S velocities increase with the square root 
of depth, starting at very low wavespeed at the surface (0.1 km/s) and increasing to 
about 3.5 km/s at 2.5 km depth. Our estimates of VS at 0.1-0.5 km agree with 
previous studies of the average S-wave velocity of the ME, CP and Central United 
states basins of 0.45-0.67 km/s at the near surface (Chen et al., 1996; Liu et al., 1997; 
Bodin et al., 2001). We find that the P-wave velocity profile agrees well with that 
predicted by the Brocher VP/VS regression, starting at ~2 km/s at the surface and 
increasing to ~6 km/s at 2.5 km depth. Due to the fact that VP is much greater than VS 
in the shallow sediment (VP/VS is generally greater than 2), the P-leg of the PPbs 
phase accumulates much less travel-time than the s-leg. Therefore, when using any 
 95 
 
Amplitude of R and T SR
Amplitude of RF SR
 
phase arrival visible on the PRF to infer VP profiles, even small errors in picking can 
result in large errors in estimated VP. We find that the average velocity profiles for 
FEG and CP groups of stations are consistent with each other, at least for the top 2.5 
km of the basin; below 2.5 km the FEG basin is likely to consist of 2 or more 
underlying layers.  This suggests that the CP and FEG strata above 2.5 km considered 
in this study are similar, which agrees with previous studies (Cook et al., 1981; Pratt, 
2018). The velocities obtained at 2.5 km depth are relatively fast for sedimentary 
rock, but agree well with previous S and P velocity estimates at the same depth (Cook 
et al., 1981; Dreiling et al. 2016). These studies suggest a shallow reflector of 
sedimentary material with an average VP of 2-3 km/s above a deeper and seismically 
faster Triassic basin with an average VP = 4.5-5 km/s (Herrman, 1979; Cook et al., 
1981). The velocity estimates from our VP profiles at the very near surface of 2km/s 
are fast when compared to previous studies in the ACP VP, which suggest that VP at 
the very near surface (< 2 m) is around 1.4 km/s (Yantis, 1983; Chapman, 2003; 
Wells et al. 2015); the difference is likely due to the limited resolution of our 4 Hz 
RFs at the very near surface, where our method is likely to average velocities in the 
upper ~100 m.  The ~6 km/s velocity at 2.5 km that our models obtain agree well with 
studies which predict that thick sediment layer may exist (>1 km) with VP of 6 km/s 
(Langston and Horton, 2014; Dreiling et al. 2016). 
 
We further validate our velocity profiles by comparing the amplitudes of r0 and PPbs 
measured directly on the PRF with amplitudes from the reflection/transmission 
coefficients calculated using the velocity profiles obtained above. Having found that 
 96 
 
 
Brocher VP/VS scaling relationships are compatible with the data, we use density-VS 
relationships to construct density profiles for the ACP. Together with crustal 
parameters of VP = 6.7 km/s and density = 2.69 g/cm3 from NGA East (Drieling et 
al., 2016), we use our VP, VS, and density profiles to calculate reflection and 
transmission coefficients for PPbs and the SS reverberation (r0) at each depth using 
expressions from Aki and Richards (1980). For thin portions of the basin (< 0.5km), 
the calculated ?? reflection coefficients have similar amplitude as the measured ?r0 at 
each station (Figure 3.17A), suggesting that the S-velocity profiles obtained are not 
only consistent with amplitude measurements not used in the inversion, but also that 
they may be expected to yield reasonable amplitude predictions when used in wave 
simulations. For thicker basins (> 0.5 km), measured amplitudes tend to be somewhat 
higher than those predicted by our velocity profiles, suggesting that there may be 
multiple small layers of sediment in the deeper portion of the basin. The relatively 
large variability in measured r0 likely reflects variability in the lithology of the 
bedrock amalgamated in past collisions and accretionary events (Wagner et al., 
2018). The PPbs reflection coefficient is related to the ???! reflection at the free 
surface multiplied by the ?? reflection coefficient at the basin/bedrock interface. We 
find that the ??! ? ?? amplitude predicted by our ACP velocity profiles fits th e 
measured PPbs amplitudes (Figure 3.17B), except for thinnest basins (< 0.15km), 
where the PPs phase gets amplified from merging with the Ps phase. The agreement 
between measured and predicted PPbs amplitudes provides independent validation of 
our P-velocity profile, and supports our finding that Brocher scaling relationships are 
appropriate for used in the ACP. 
 97 
 
 
A. B.
SS amplitude PPbs amplitude
1
-r0 0.3 PPbs amplitude
0.8 SS PP0*PS
0.6 0.2
0.4
0.1
0.2
0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
Basin Thickness (km) Basin Thickness (km)
 
Figure 3.17 ? Amplitude measurements from PRFs compared to calculated 
 reflection/transmission coefficients for the sediment velocity models obtained 
 in this study using Aki and Richards (1980). Sediment density calculated 
 using the Brocher (2005) relationships; crustal parameters are obtained from 
 Drieling et al. (2016). A. r0 measurements (blue circles) compared to 
 calculated ?? reflection coefficients (black line) for a range of basin thickness. 
 B.  Measurements of PPbs (pink circles) compared to calculated ??! ? ? ?? 
 reflection coefficients (black line) for a range of basin thickness.   
 
3.6 Discussion of ACP results 
In large regions like the Atlantic Coastal Plain, where broadband seismic stations 
have been installed but regional seismicity is rare, obtaining basin structure 
constraints using more traditional methods developed for regions with an abundance 
of seismicity may prove challenging. High-frequency P-to-S receiver functions 
provide an underutilized approach for constraining the basin fundamental frequency 
and velocity profiles using teleseismic events.   
 
Traditional SSR or HVSR techniques to measure site-specific basin properties rely on 
local earthquakes or the ambient noise field. The lack of widespread seismicity on the 
 98 
 
SS amplitude
PPbs amplitude
 
ACP, requires that passive seismic techniques are used. While ambient noise 
techniques have shown great promise in improving data coverage in largely aseismic 
areas, the accuracy of the retrieved Green's functions depends on assumptions about 
the isotropic distribution of noise sources and stationarity of the signal (e.g. Yao et al.  
2009; Tsai, 2009). In the ACP, the dominant source of the seismic noise is the 
Atlantic Coast (e.g. Koper and Burlacu, 2015), resulting in a highly anisotropic noise 
wavefield. This anisotropy can introduce bias into phase and group velocity 
measurements made from ambient noise cross-correlations.  
 
Because receiver functions are calculated from teleseismic events, and do not rely on 
local earthquakes or ambient noise, existing broadband station deployments (such as 
USArray) can be used to characterize site-specific basin fundamental frequency and 
regional velocity profiles across large geographic areas without relying on local 
earthquake events or ambient noise measurements. When the velocity contrast is 
significant, and the sedimentary layer is thick (greater than 0.1 km), phase arrivals are 
large and visible on the mean PRF with arrival times that increase, as expected, with 
increasing sediment thickness. However, relatively low frequency bandpass corners 
of 0.5 - 1 Hz typically used in PRF studies of crustal structure are insufficiently high 
to distinguish between and interpret different phase arrivals (particularly Pbs and 
PPbs) when sediment is less than 1 km thick, as is the case across much of the ACP 
(see Figure 3.18). Here, we show that much higher frequency signals can be reliably 
extracted from teleseismic PRFs, enabling the characterization of sediments as thin as 
0.1 km. In regions where broadband stations have not been deployed or coverage is 
 99 
 
 
insufficient, receiver functions can be calculated from nodal array deployments (e.g. 
Ward and Lin, 2017; Liu et al. 2018), which have optimal sensitivity in the frequency 
range of interest. Indeed, the ease of deployment and low cost of nodal deployment 
also enable targeted, detailed mapping of sedimentary basins based on PRF 
technique.  
 
Figure 3.18 ? Synthetic PRFs calculated for a 0.5 km thick sediment In order to 
 distinguish between the Pbs and PPbs phase arrivals, High frequency PRF 
 (with high corner frequency of 4Hz) must be used.  
 
PRFs are most often used to map interfaces within the deep crust and underlying 
mantle. However, interpreting PRF signals from deeper structures often requires 
accounting for signals due to sedimentary basins (Chapter 2), yielding useful 
constraints on sediment structure as a byproduct. S-reverberation time (and basin 
fundamental frequency) must be estimated when constructing shallow-layer 
reverberation removal filters, and Ps phases must be identified when applying 
shallow-layer corrections to H-k stacks. Therefore, velocity profiles can be calculated 
whenever PRFs are used in regions with significant sedimentary cover. While 
 100 
 
 
requiring more labor than traditional SSR or HVSR techniques, high frequency PRFs 
may be useful in constraining basin properties over large areas especially in regions 
with low or intermittent seismic activity.  
3.7 Conclusions 
Receiver function analysis of teleseismic events recorded at broadband seismic 
stations from USArray places constraints on basin fundamental frequency and 
velocity profiles. We constrain these basin properties across the Atlantic Coastal Plain 
from phase arrivals on high-frequency (4 Hz) P-to-S receiver functions. From 
measurements of the two-way S-wave travel time on the receiver function, the basin 
fundamental frequency is mapped across the ACP. While shallow sediment structure 
may result in even higher fundamental frequencies, the basin fundamental frequencies 
presented in this work are consistent with basin thickness and results from applying 
more traditional techniques to teleseismic data. In addition to providing site-specific 
fundamental frequencies, the relationship of TSs (or fundamental frequency) and basin 
thickness may provide a method to estimate the expected fundamental frequency at 
any basin thickness for thick sedimentary basins.   The amplitudes measured on the 
RFs match expected patterns with increasing basin thickness. ACP and ME strata 
used this this study share a similar high velocity profile for VS until about 2.5 km 
thick, implying that above this depth, they share similar basin structure.  The VS 
profile increases as the square of depth, while VP profiles for the ACP and ME are fit 
well using by the Brocher regression fit. The receiver function amplitudes at the 
phase arrival times validate the high velocity VP and VS profiles. 
   
 101 
 
 
Chapter 4: Crustal and Lithospheric Structure in the Southeastern US 
using Receiver Function Common Conversion Point (CCP) Stacking  
 
4.1 Introduction  
The Southeastern United States (SEUS) lies on the passive eastern North American 
margin and preserves a long geologic record of continental collision, rifting, and 
volcanism. The multiple episodes of continental collisions have left a chaotic 
assortment of accreted terranes and faults. The lithospheric structure reflecting these 
accreted terranes documents the evolution of the SEUS, where lithospheric thickening 
indicates the locations of past compressional or collisional events and lithospheric 
thinning suggests a past extensional or rifting events. Constraining the geographic 
extents of these lithospheric variations will help clarify the tectonic history of the 
region. 
  
The dense deployment of the USArray Transportable array seismic network allowed 
for seismic imaging of the often-neglected passive margin with higher resolution than 
ever before. Despite the numerous studies that investigate the lithospheric structure of 
the SEUS (e.g. Parker et al. 2016; Biryol et al. 2016; Hopper et al. 2017; Lynner and 
Porrit, 2017; Soto-Cordero et al., 2018; Wagner et al., 2018) questions about the 
underlying large-scale structures remain. For example, some studies have found 
evidence for a low velocity ?mid-lithospheric discontinuity? (MLD) beneath the 
Appalachian Mountains (Abt et al. 2010; Long et al., 2017), while others suggest that 
the MLD and LAB coincide in depth in this region (Liu and Gao, 2018). While the 
existence of a low velocity discontinuity at 80-120 km has been well documented in 
 102 
 
 
continental cratons, mechanisms controlling this MLD have remained elusive with 
explanations spanning the rheological, chemical, and mechanical realm. 
  
Even the base of the lithosphere-asthenosphere boundary (LAB) is not well 
constrained across the SEUS. While most receiver function studies find that the LAB 
beneath the Piedmont shallows to 100-120 km depth (Wagner et al. 2018, Hopper 
and Fischer, 2018), recent magnetotelluric studies suggest the lithosphere in the same 
region can be no thinner than 200 km (Murphy and Egbert, 2017). 
  
The multiple continental collisions undergone in the SEUS left a complex crustal 
structure, complicating the imaging of deeper structure. SRF CCP stacking has shown 
evidence of multiple crustal layers of crustal emplacement on Laurentian basement 
during past collisional events, indicating that past Proterozoic tectonics in the region 
may have been analogous to modern tectonics. Seismic imaging of complicated fault 
structure in the Blue Ridge beneath seismically active zones in Tennessee has pointed 
to additional layering at mid-crustal depths trending both south-southeast 
(Bobyarchick, 1998; Culotta et al., 1990, Wagner 2012) and north-northwest (Culotta 
et al., 1990; Wagner 2012, Powell et al., 2016). These mid-crustal layers may be 
evidence of crustal emplacement during a separate orogeny and may be related to the 
underlying seismicity. 
  
Despite its distance from active margin tectonics, the SEUS contains multiple zones 
of seismic activity capable of producing damaging earthquakes. Because large 
earthquakes in this region occur infrequently, little is known about the mechanisms 
 103 
 
 
controlling the locations and magnitudes of these events. The seismogenic zones are 
dispersed throughout the SEUS and include the central Virginia seismic zone 
(CVSZ), the Eastern Tennessee seismic zone (ETSZ), and the South Carolina seismic 
Zone (SCSZ) (Figure 4.1). Previous studies have suggested that stresses that control 
intra-plate seismogenic zones are correlated with boundaries between accreted 
terranes and variations in deeper earth structure. Specifically, these seismogenic 
zones have been found to correlate with abrupt changes in thickness of the tectonic 
plate (lithosphere) (Biryol et al., 2016), a weak intra-lithospheric layer (Thybo et al., 
2000), and sharp changes in thickness of the continental crust (Soto-Cordero et al., 
2017). We explore these hypotheses by systematically mapping variations in 
lithospheric structure using CCP stacks beneath regions of seismicity and seismic 
quiescence, looking for structural patterns that may highlight which mechanisms play 
a role in controlling the location of passive margin seismicity. 
  
Receiver functions have proven extremely useful in constraining crustal and 
lithospheric variations, as they are particularly sensitive to sharp changes in the 
elastic parameters. P-to-s receiver functions (PRFs) isolate S-wave conversions 
produced across impedance contrasts beneath a seismic station, and the relative phase 
arrival times can be used to constrain the deeper structure. Similarly S-to-p receiver 
functions (SRFs) isolate P-wave conversions produced across impedance contrasts. 
PRFs contain higher frequency energy are thus preferred to constrain structure 
variations since they are able to provide higher vertical and horizontal resolution. 
However, reverberations produced within low-velocity shallow layers, such as 
sedimentary basins, can contaminate traditional PRFs (Yeck et al. 2013, Yu et al., 
 104 
 
 
2015, Schulte-Peklum et al., 2017). In the SEUS, sedimentary basins cover most of 
the region, making interpretations of lithospheric structure from PRFs nearly 
impossible. We remove large amplitude sedimentary reverberations by applying a 
resonance removal filter and creating reverberation-removed PRFs (e.g. Yu et al., 
2015; Cunningham and Lekic, 2019). Common conversion point (CCP) stacking of 
receiver functions is a useful tool to image lithospheric structure variations across an 
array (e.g. Dueker and Sheehan 1997; Gilber et al. 2003; Frasetto et al., 2010). CCP 
stacking uses a velocity model to migrate the relative arrival times of converted 
phases to depth and lateral locations beneath the station. We calculate reverberation-
removed PRFs CCP stacks across the SEUS to look for structure variations at 
relatively high resolution. SRFs are not contaminated by sediment reverberations in 
the same way are PRFs; therefore, we calculate complimentary SRF CCP stacks 
which provide a relatively low resolution view of lithospheric structure across the 
region. Using both the sediment removed PRF and SRF CCP stacks, we constrain the 
crustal and lithospheric structure beneath the SEUS. We focus our investigations on 
identifying crustal and lithospheric thickness variations, any evidence of accreted 
terranes preserved as intra-lithospheric or intra-crustal layering, and identifying 
correlations between lithospheric structure variations and seismicity. 
4.3 Receiver Function Data for Common Conversion Point (CCP) 
Stacking 
For this study, we calculate PRFs and SRFs for station in the Southeastern United 
States. Locations of 236 stations from the USArray Temporary Array between 30 - 
42.25N and 86.5 - 73.5W are shown in Figure 4.2. 
 105 
 
 
  
Data is collected from the IRIS DMC for earthquakes before January 1, 2018 with 
Mw >5.5 and epicentral distances between 30 and 90 for PRFs and between 55 and 
80 for SRFs (Wilson et al., 2006). The events are quality controlled, requiring a 
minimum Z-to-R cross correlation of 0.3 and a maximum difference of 25 seconds 
between observed and predicted arrival times. The RFs are calculated following the 
automated procedure detailed in Abt et al. (2010), where the waveforms are projected 
onto the P-SV-SH system using the free surface transform matrix and finding the 
surface velocity that minimizes the amplitude of the parent wave (P for PRFs and S 
for SRFs) on the daughter component. A fourth order band-pass filter is applied to 
waveforms of 0.03-1Hz (for 1Hz PRFs), 0.03-4Hz (for 4Hz PRFs) and 0.03-0.5Hz 
(for SRFs). Using the time-domain deconvolution of Ligorria and Ammon (1999), we 
calculated the RFs with Gaussian half-amplitude half-width of  ~0.12 s to calculate 
the 4Hz PRFs, of  ~0.5 s to calculate the 1Hz PRFs, and of  ~1 s to calculate the 
SRFs. We find a median of 93 PRFs and between 80 SRFs per station (Figure 4.2). 
We choose the time domain deconvolution over frequency domain deconvolution 
technique as the damping or regularization parameter added to stabilize the 
frequency-domain deconvolution results in less-detailed RFs with stronger side-lobes 
(Lekic and Fischer, 2017). 
 106 
 
 
A B
# of PRFs at each Station # of SRFs at each Station
450 400
  42oN   42oN
350
  39oN   39oN 250
250
  36oN   36oN
150
150
  33oN 100   33oN 100
50 50
  30oN   30o N 87o
0
W  84oW  81oW  78oW  75oW   87
oW  84oW  81oW  78oW  75oW
 
Figure 4.1- The locations of USArray TA stations used in this study colored by the 
 number of P-to-s (A) and S-to-p (B) receiver functions calculated at each 
 station. Dark red circles are stations that have been deployed for a longer time. 
 
4.4 Common Conversion Point (CCP) stacking 
4.4.1 CCP stacking overview 
When considering the structure beneath an array, RFs are often combined using 
common conversion Point (CCP) stacking to create a 3-D image of the subsurface. 
CCP stacking takes advantage of the range of source events and receivers, projecting 
individual RF amplitudes to depth along their ray path and backazimuth using a 
velocity model and stacking them in volumetric bins, which suppresses noise present 
on individual RF traces. The velocity model used to migrate the RFs from time to 
 107 
 
 
depth can be a general 1-D reference model such as ak-135 (Kennett et al., 1995) or 
can be modified to include measured crustal values obtained from H- stacking (Zhu 
and Kanamori, 2000). The velocity model we use in this study is based on crustal 
values obtained from SRTC H-k-VP stacking of PRFs and SRFs (discussed in section 
3.2 below).   
 
CCP stacks are calculated following the method detailed in the supplement of Lekic 
and Fischer (2011), where the stack at each depth (?!) and latitude-longitude location 
is calculated as the weighted average of the individual RFs at that point. The 
weighting factor (?) depends on the cubic spline at the horizontal distance (?) of each 
migrated ray from the CCP stack point. The horizontal distance of each migrated ray 
at depth (?) is normalized by the knot spacing of the cubic spline ?(??), which 
increases as a function of depth written in terms of the wavelength (?) at each discrete 
depth (?!):  
 
4.1  ? ? ? ??? =  ? ! (! + ? )! ? ? ! . 
! ! ! !
The weighting factor ? of each individual RF contributing to the CCP stack at a given 
point is calculated as: 
3
?!
3
? ?! + 1, ? ? ? ? ? ?? <  ?1
4 2
4.2  ? ? ??(?) = 1 (2? ?)!, ? ? ? ? ? ? ? ?1 < ? < 2 
4
0 ?, ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? > 2
 
 108 
 
 
where ? =  ??/ ???. Therefore, if the migrated RF is spatially close to a given location 
in the subsurface volume, the weighting factor is close to 1, while at distances of  ! 
!
the weighting factor will tend toward 0. The weighting factor ? ?(?) has the effect of 
horizontally smoothing physical structure through the zero offset Fresnel zone half-
width; as depth increases, the Fresnel zone half-width (and effectively the smoothing 
length scale) increases. The benefit to using this type of weighting or smoothing 
method is that it that it reduces the possibility of over-smoothing and losing sub-
Fresnel zone structure (Lekic and Fischer, 2017). 
 
PRFs contain higher frequencies than SRFs and so near the station the horizontal 
distance becomes too small; for this reason PRF CCP stacks have a ?? set to 50 for all 
depths and will demonstrate both higher vertical and horizontal resolution than SRFs. 
PRFs will be able to image sharp lateral variations better than SRFs, but are more 
sensitive to contamination by shallow layer multiples. For this reason, we calculate 
and compare the results from both PRF and SRF CCP stacks through the SEUS.   
 
The CCP stacking volume for both PRFs and SRFs is discretized by 0.5km in depth, 
and 0.25? in latitude and longitude. The results of CCP stacking are shown as both N-
S and W-E cross-sections through the stack volume (Figure 4.3).  
 109 
 
 
Common Conversion points and cross sections 
  42oN A E J O T
20
  39oN
15
  36oN
10
  33oN 5
  30o 1
 N 87oW  84oW  81oW  78oW  75oW
 
Figure 4.2- Locations of CCP stacking points (blue) and stations (red triangles).  
 Cross sections shown are calculated along vertical and horizontal lines in the 
 gird. Each cross section number indicates a W-E cross section, while letters 
 indicate a N-S cross-section. 
 
 110 
 
 
4.4.2 Regional Velocity Model with SRTC H- ?-VP Stacking 
Common conversion point (CCP) stacking requires that the individual RFs be 
migrated from time to depth using a velocity model of the subsurface. While global 
models such as ak-135 can be used, errors in the velocity structure lead to errors in 
depth in the CCP stack (Hopper et al., 2017). To alleviate errors associated with an 
incorrect velocity model, we calculate a velocity model for the SEUS based on crustal 
thickness and velocity parameters obtained at each station through an H-?-VP stacking 
technique. 
 
Traditional H-? stacking uses predicted arrival times of converted phases on the PRFs 
to constrain the velocity beneath the seismic station assuming an average crustal P-
wave velocity (Zhu and Kanamori, 2000). However, an incorrect assumption of 
average crustal VP for a region can lead to error of up to 10 km in crustal thickness 
(H) and 0.4 in VP/VS ratio (?) (Figure 4.4). Because both PRFs and SRFs have some 
sensitivity to average crustal P-wave velocity (VP), combining them into H-?-VP will 
constrain all three crustal parameters (e.g. Rychert and Harmon, 2016).  
060Z Crustal thickness vs VP
40
35
30
25
5.2 5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7
VP (km/s) 
060Z VP/VS vs VP
2
1.9
1.8
1.7
1.6
1.5
5.2 5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7
VP (km/s)  
Figure 4.3- The velocity assumption in H- stacking is extremely important and can 
 significantly change inferred crustal properties. Results from traditional H-k 
 111 
 
VP/VS H  (km)
 
 stacks calculated for station 060Z on Atlantic Coastal Plain sediments. Top. 
 As the assumed average crustal VP velocity used to calculate the stacks 
 increases, the crustal thickness estimate increases until a new maximum is 
 inferred (at 6.5 km/s). Bottom. As the assumed average crustal VP increases, 
 the VP/VS ratio (or ??) decreases, until again a new maximum is inferred. 
 
In regions with a low-velocity sediment layer, such as the Atlantic Coastal Plain 
(ACP), crustal signal in PRFs is often contaminated by sediment multiples (Yeck et 
al., 2013; Yu et al., 2015; Schulte-Pelkum et al., 2017). For this reason, we calculate 
sediment removed (SR) PRFs for stations with sediment reverberation contamination. 
The SR PRFs are calculated following the method detailed in Chapter 2, where a 
resonance removal filter is constructed to remove the oscillatory reverberation signal. 
The filter is constructed using measurements of the two-way travel time of the S-
wave in the sediment (??) made from the autocorrelation of the mean PRF. To correct 
for time delays introduced from a slow sedimentary layers, the arrival time of the 
PPbs phase (???) made on the high-frequency (4Hz) PRF as it depends on both VP 
and VS of the sedimentary layer. Measurements of ?? and ??? are made at each 
station with sediment reverberations (Figure 4.5).  
 112 
 
 
TSs (s) TPPbs (s)
3.5
  42oN   42oN 1.8
3
1.6
  39oN 2.5   39oN 1.4
2 1.2
  36oN   36o1.5 N
1
0.8
  33o 1N   3 o
0.6
3 N
0.5 0.4
0.2
  30oN o o 0   30
o
 N 87 W  84oW  81oW  78oW  75oW  72 W   87
oW 0  84o oW  81oW  78oW  75oW  72 W
Figure 4.4- Measurements of sediment phase arrival times at each station.The two 
 way S-wave travel time in sediment (left) and the arrival time of the PPbs 
 phase (right) are shown. In regions with thicker sediment, both TSs (??) and 
 TPPbs (???) are large, while in regions with thinner sediment TSs and TPPbs are 
 small. Stations in black do not demonstrate a sedimentary signal in the 
 receiver functions.   
 
When the SR PRFs and sediment time-delay corrections to each phase are used to 
calculate a stack over a range of input H, ? and VP values, we call this sediment 
removed time-corrected (SRTC) H-?-VP triple stack. The equation for a (SRTC) H-?-
VP triple stack (?) using the predicted Pms, PPms, and PSms+PmsPms phase arrival 
times (?!!!, ??!!!!, ??!!!! + ?!!!!!! respectively) for ? number of PRFs computed a 
station ?!(?) and the predicted Smp phase arrival time (?!!!) for ?! number of SRFs 
computed at the same station ?!! ? is written as:  
 
 113 
 
 
4.3  ? ? ? ?? ?, ?,?! = !!!!?!?! ?!!! + ?? ? ??? + ?!?! ?!!!! + ??? ?
?!?
!! !
! ?!!!! + ?!!!!!! +?? +  ? !!! ?! ?!!! + ?? ? ???  + 
 
where ?!, ?!, and ?!are the relative weights of the three PRF phases (set to 0.7 0.3 
and 0.1 respectively).  
 
We calculate the triple stack over a range of H, k and VP values where H ranges from 
20 to 60 km and ? from 1.5 to 2. In chapter 2, we found that even by combining 
sediment removed time corrected PRFs, SRFs, and the post-critical reflection at the 
Moho (SsPmp) the VP obtained from the triple stack is not well constrained. 
Therefore, we restrict or search for average crustal VP to values within +/-0.2 km/s of 
values predicted by Crust2.0 (Bassin et al., 2000). The SRTC H-?-VP triple stacks 
provide an estimate of the crustal thickness (H), VP/VS (?) and crustal VP beneath 
each station (Figure 4.6). The velocity model used to calculate the CCP stacks is then 
updated to include the new crustal parameters. The new velocity model should help 
constrain the correct depths to layers in the CCP stacks.  
 114 
 
 
A B C
H (km)  VP (km/s)
60 2 6.8
o o o
  42 N   42 N   42 N
1.9 6.6
o 50 o o
  39 N   39 N   39 N
1.8 6.4
o o o
  3 406 N   36 N   36 N
1.7 6.2
o o o
  33 N 30   33 N   33 N
1.6 6
o o o
  30 N   320 0 N
  30 N
o
  87 W o o o o
o o o 1.5 o
W   87 W o o o o
o 5.8
  84 W  81 W  78 W  75 W  72   
o o o o
84 W     81 W  78 W  75 W  72 W
87 W  84 W  81 W  78 W  75 W  72 W
Figure 4.5- Values of crustal H, ??, and VP for each station in our study area obtained 
 from H- ?? -VP stacking method (described in text). These parameters are used 
 to calculate a velocity model for the migration of RFs from time to depth in 
 the CCP stack 
4.4.3 Sediment Removed PRF CCP stacks and SRF CCP stacks 
We calculate SR PRF CCP stacks for the SEUS using the velocity model produced 
using our H- -VP stacking approach. CCP stacks should help mitigate the effects of 
sediment reverberations, as the near vertical reverberations are not expected to 
constructively stack at Moho depths when individual RFs are migrated along ray path 
and backazimuths. However, near the edge points of a seismic array, too few RFs 
from different source and receiver paths may cause reverberations to become visible. 
In addition, if sediment reverberations directly overprint the P-to-s wave arrival from 
the Moho, CCP stacking will not be able to constrain crustal and lithospheric 
structure and sediment reverberations may still be visible in the stack (Figure 4.7A). 
 115 
 
 
 
  42oN
  39oN
  36oN
  33oN
  30oN
  87oW  84oW  81oW  78oW  75oW  72
oW
A H Traditonal PRF CCP stack
0
40
80 Moho
120 Sed. Reverberations
160
200
27 127 227 327 427 527 627 727 827 927 1027 1127
Distance Along Profile (km)
0 0.05 0.1 0.15 0.2 0.25
RF amplitude
B SR PRF CCP stack with HkVP
0
40
80 Moho Sed. Reverberations
120 removed
160
200 27 127 227 327 427 527 627 727 827 927 1027 1127
Distance Along Profile (km) Moho Reverberations
0 0.05 0.1 0.15 0.2 0.25
RF amplitude
SRF CCP stack with HkVP
C
0
40
80 Moho
120
160
200 27 127 227 327 427 527 627 727 827 927 1027 1127
Distance Along Profile (km)
0 0.05 0.1 0.15 0.2
RF amplitude
 
Figure 4.6- Results of CCP stacking for cross-section H (location shown in top panel 
 of figure). Red indicates positive RF amplitude associated with a velocity 
 increase with depth while blue indicates negative RF amplitude associated 
 with a velocity decrease with depth. A. In traditional PRF CCP stacking we 
 116 
 
Depth (km) Depth (km) Depth (km)
 
 see large amplitude sediment reverberations that create a Moho step at about 
 327 km along the profile B. Sediment removed (SR) PRF CCP stacks remove 
 much of the sediment reverberation contamination in the regions highlighted 
 and show a flat Moho at 327 km along the profile. C. SRF CCP stacks show a 
 similar Moho profile to SR PRF CCP stacks, but have lower resolution. 
 
 
To improve upon traditional CCP stacking techniques, we use sediment removed 
(SR) PRFs calculated in section 4.2. Before the resonance removal filter is applied, 
the mean PRF at two example stations on ACP sediment show large amplitude 
oscillatory signal (Gray line ? Figure 4.8). After the resonance removal is applied, a 
small positive amplitude peak emerges between 4-5 seconds, which is likely to be the 
Pms arrival from the Moho. When the SR PRFs are used in CCP stacking, the 
oscillatory pattern seen on the edges of the stack and beneath the ACP sediments is 
reduced (Figure 4.7B). Although the Pms phase amplitude appears small and weak on 
the individual RFs, when migrated to depth, the Pms phases coherently stack together 
producing a strong Moho signal. 
 
Figure 4.7- Examples of the resonance filter applied to PRFs. Gray is the mean of the 
 PRFs before the resonance removal filter is applied, red line shows the mean 
 of the PRFs after the sediment is removed. A. Station W61A in New Bern, 
 NC. Before the sediment is removed, no clear Moho phase arrival (Pms) exists 
 in the RF. After the sediment is removed, a small amplitude peak appears at 
 ~5s which is likely the Moho arrival. B.  Station S60A in Water view, VA. 
 Again, before the sediment is removed, no clear Moho phase arrival (Pms) 
 117 
 
 
 exists in the RF. After the sediment is removed, a evidence of a Moho arrival 
 appears between 4-5 s. 
 
Additionally, below the Moho on the SR PRF CCP stacks, additional positive and 
negative phases appear that are not seen in the traditional PRF CCP stacks. These 
phases are not in fact lithospheric structure, but instead are the Moho-related 
multiples. Therefore, we restrict our interpretation to depths shallower than the PpPs 
Moho phase 
  
At greater depths, we look to SRF CCP stacks to constrain the lithospheric structure 
since Moho reverberations do not contaminate SRFs (Figure 4.7C).  SRF CCP stack 
shows a broad crust in at the same depths as the SR PRF CCP stacks. The broad 
structure is due to the lower frequency content of S-waves, which explains the 
decrease in vertical resolution. No strong negative amplitude in the lithosphere is 
visible on the SRF CCP stacks, which means that the lithosphere asthenosphere 
boundary in the SEUS either extends much deeper than previous seismic studies 
suggest, or that the boundary between the lithosphere and asthenosphere in beneath 
the SEUS is not a sharp impedance contrast, but rather a gradational boundary. 
4.5 Results and Observations 
Using the sediment removed PRF and SRF common conversion point stacks, we map 
the vertical and lateral variations in lithospheric structure beneath the Southeastern 
United States. Specifically, we investigate evidence of past collisional events and 
inherited structures and look for correlations between deeper structure and seismicity 
in the SEUS. In this section, we present selected cross-sections required for 
 118 
 
 
discussion; however, cross-sections through the total CCP stacking volume for the 
entire SEUS are shown in Appendix. Letters starting with A in the west and V in the 
east note the location of north to south (N-S) cross-sections. Numbers starting with 1 
in the south and 23 in the north note the locations of east to west (E-W) cross-
sections. 
4.5.1 Crustal Thickness  
When the CCP stacks are plotted in cross-section, the most notable feature is the large 
amplitude positive phase between 30-60 km depth (e.g. Figure 4.9). A positive phase 
implies a velocity increase with depth. The positive phase seen on our CCP stacks is 
correlated with the overlying topography (shown on cross-sections with 10x vertical 
exaggeration), where high topographic regions parallel increases in the depth of the 
positive phase. Based on the depth, amplitude and depth correlation with expectations 
from isostasy, we interpret the positive phase between 20-60 km to be the boundary 
between the crust and the mantle (Moho). 
 119 
 
 
 
  42oN
  39oN
  36oN
  33oN
  30oN
  87oW  o o o oW  72o84 W  81 W  78 W  75 W
22 SR PRF CCP SRF CCP
Central Lowland Appalachain Plateaus Valley and Ridge Central Lowland Appalachain Plateaus Valley and Ridge
RT AB RT AB
Moho
Distance Along Profile (km) Distance Along Profile (km)
RF amplitude RF amplitude
 
Figure 4.8 - Figure 9- W-E Cross-section 22 showing a strong positive (red) arrival 
 between 35 and 45 km depth. Right is the sediment removed (SR) PRF CCP 
 stack and left is the SRF CCP stack. Stacks below show the variance of the RF 
 amplitude at each point in the CCP stack. We interpret this positive Labels in 
 black are features from Figure 1 that the cross sections pass through. RT= 
 Rome trough, GF= Grenville front, SGR = South Georgia rift, ETSZ = eastern 
 Tennessee seismic zone, BFZ = Brevard fault zone, CPS= central Piedmont 
 suture, NCSQ = North Carolina seismic quiescence, AB = Appalachian basin, 
 SCSZ = South Carolina seismic zone, SSZ = Suwanee suture zone, CVSZ = 
 central Virginia seismic zone. Colored labels above each cross section denote 
 the physiographic province that the cross sections pass through. CL = Central 
 Lowland; AP = Appalachian Plateau, V&R = Valley and Ridge, BR = Blue 
 Ridge, ACP = Atlantic Coastal Plain, LP = Interior Lower Plateau. 
 
The Moho depth is picked beneath each CCP cross-section, (both N-S and E-W). This 
allows for a birds eye view of the Moho depth (Figure 4.10). We find regions of very 
thick crust beneath the valley and ridge, Blue Ridge, and Appalachian Plateau 
physiographic provinces (Fenneman, 1928) with thinner crust closer to the Atlantic 
 120 
 
Depth (km)
GF
GF
 
Coast. The sharp crustal thickness change on the western portion of our map seems to 
follow a similar trend as the predicted Grenville Front (GF). When compared to our 
previous estimates of crustal thickness from H-k-VP triple stacks (Figure 4.6), the 
crustal thickness from CCP stacking is smoother and deeper. The crustal thickness 
from CCP stacks is smoother due to the lateral smoothing that arises from stacking 
the SR PRFs along their ray paths and backazimuths. We find the thickest crust 
beneath the Eastern Tennessee Seismic Zone (ETSZ) of around 63 km. While a 63 
km crust is reasonable in active continent-continent orogenic regions such as 
Himalaya (Vera Schulte Pelkum et al., 2005) it is an unusually thick estimate for 
crustal thickness in the SEUS. 
 
Figure 4.9- Left: Crustal thickness/Moho depth from SR PRF CCP stacks. Contour 
 figure of Moho depths shows a deep (~65km) Moho beneath the Blue Ridge 
 and Appalachian Plateaus (~60 km) that shallows towards the coast (~25km). 
 Right: Moho depth from CCP stacks are picked in each horizontal and vertical 
 direction cross section, where Moho depth is interpolated at each CCP 
 stacking point (white dotted line) in-between user picks (black stars). These 
 interpolated values are averaged from both N-S and E-W cross sections to 
 create the crustal thickness map from CCP stacks.  
 
 121 
 
 
When compared to the aeromagnetic data and gravity anomaly data (Figure 4.11), 
We find that the sharp change in crust thickness between the Piedmont and Blue 
Ridge seems to agree well with the observed sharp change in gravitational anomaly 
feature ?A? while the two thicker crust in southern Ohio an northeastern Tennessee 
may agree with to gravitational features ?B? and ?C? in Figure 4.11B (Knucks , 
1999; Stein et al., 2014; black line ).  When compared to the magnetic data of the 
region, we find that the NY-AL magnetic lineament (Steltenpohl et al., 2010 black 
line? Figure 4.11A) is mirrors the predicted edge of the Grenville Front (GF) in the 
northwest portion of our study, although, we do not find evidence of the GF further 
south along the New York- Alabama (NY-AL) magnetic lineament.  
 122 
 
 
A.
SSZ 
B.
B
C
 
Figure 4.10- A.  Aeromagnetic maps, where reds are aeromagnetic highs and blues 
are areomagnetics lows.  The black line is the NY-AL lineament which is likely due 
to a buried strike slip fault (figure modified from Steltenpohl et al., 2010) B. Bouger 
gravitational anomalies in the Southeastern US. Figure created from USGS gravity 
anomaly map of the conterminous United States (Knucks, 1999). Features ?A? 
correspond to a sharp change in crustal thickness, Features ?B? and ?C? seems to 
agree with increased crustal thickness measurements from CCP stacks in this study.   
 
 
 123 
 
A
CPS 
NY-AL 
 
We investigate the anomalously thick crust in a W-E cross-section near the maximum 
crustal thickness estimate (Figure 4.12). Beneath the Blue Ridge (BR) and the ETSZ, 
multiple positive phase arrivals are seen between 40-60 km. Following the dominant 
positive phase in the east of the stack at about 37 km (beneath the ACP), it appears to 
begin dipping westward beneath the BR and continues to dip to about 60 km beneath 
the ETSZ and the Brevard fault zone (BFZ). The crustal thickness of ~60 km agrees 
well with the positive phase in the SRF CCP stacks, providing some confidence that 
the Moho phase is not a result of constructive interference of uncorrected 
reverberations in the SR PRF CCP stacks. Similarly, a mid-crustal positive phase at 
about 20 km (too shallow to be the crust in this region) remains flat beneath the ACP 
and then appears to dip westward beneath the BR, reaching about 40-45 km ? near 
expected Moho depths in this region. The ?double-Moho? structure explains the 
discrepancy between H-k-VP stacking and CCP stacking and accounts for the very 
thick crust observed in this region. Interestingly, a W-E cross section through 
Maryland (near the northern edge of our CCP stacking volume) shows an intra-crustal 
layer dipping eastward beneath the Valley and Ridge (V&R) and BR (Figure 4.13) 
This shallow eastward dipping structure is more characteristic of the intra-crustal 
layering we would expect to arise from inherited structures accreting on top of the 
Laurentian basement with current subduction directions (Rivers, 2014). 
 
 124 
 
 
  42oN
  39oN
  36oN
  33oN
  30oN
  87oW  o84oW  81oW  78oW  75oW  72 W
10 SR PRF CCP SRF CCP
V&R BR Piedmont ACP V&R BR Piedmont ACP
ETSZ&BFZ NCSZQ ETSZ&BFZ NCSZQ
0
40
80
23 223 323 423 623 723 823 923 23 223 323 423 623 723 823 923
Distance Along Profile (km) Distance Along Profile (km)
0 0
RF amplitude RF amplitude
10 SR PRF CCP Variance SRF CCP Variance
0
40
80
23 223 323 423 623 723 823 923 23 223 323 423 623 723 823 923
Distance Along Profile (km) Distance Along Profile (km)
0 0
RF amplitude RF amplitude
 
Figure 4.11- W-E Cross-section 10, which passes through the maximum crustal 
 thickness estimate in the Blue Ridge. Evidence of multiple crustal layers 
 dipping sharply westward can be seen beneath the ETSZ and BFZ. The intra-
 crustal layer is at ~10 km beneath the CPS and deepens to ~40 km beneath the 
 ETSZ, while the Moho phase in the same region begins at ~40km depth 
 beneath the CPS and dips to ~60 km beneath the ETSZ and BFZ. The SRF 
 CCP stacks also show a deep ~60 km Moho beneath the ETSZ and BFZ. A 
 flat intra-crustal layer can also be seen beneath the NCSQ at about 15 km 
 depth. . Stacks below show the variance of the RF amplitude at each point in 
 the CCP stack. See caption in Figure 9 for description of labels and 
 abbreviations. 
 
 
 
  
 125 
 
Depth (km) Depth (km)
CPS
CPS
 
  42oN
  39oN
  36oN
  33oN
  30oN
  87oW  84o
o
W  81oW  78oW  75oW  72 W
19 SR PRF CCP SRF CCP
Central Lowland AP V&R BR Piedmont Central Lowland AP V&R BR Piedmont
RT AB RT AB
0
40
80
22 122 222 322 422 522 622 722 822 22 122 222 322 422 522 622 722 822
Distance Along Profile (km) Distance Along Profile (km)
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2
RF amplitude RF amplitude
19 SR PRF CCP Variance SRF CCP Variance
0 0
20
40
60
80
100
Distance Along Profile (km) Distance Along Profile (km)
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
RF amplitude RF amplitude
 
Figure 4.12- W-E Cross section 19 that shows an eastward dipping intra-crustal layer 
 beneath the V&R and BR provinces in the SR PRF CCP stack. This intra-
 crustal layer appears at ~15 km beneath the Appalachian basin and dips 
 eastward to ~20 km beneath the Piedmont. Stacks below show the variance of 
 the RF amplitude at each point in the CCP stack. See caption in Figure 9 for 
 description of labels and abbreviations. 
 
To better resolve the extent and dip direction of intra-crustal layering, we map all 
locations where intra-crustal layering appears and note the direction of dip as west, 
flat, or east (Figure 4.14). We find that at the eastern/ northeastern edge of our study 
region, a shallowly dipping interface across which velocity increases with depth 
seems to coincide with the location of the Grenville Front (GF) and where crustal 
thickness sharply increases.  In the southern edge of our study region, a flat to 
 126 
 
Depth (km)
Depth (km)
GF
GF
 
shallowly eastward-dipping intra-crustal layer is present. The current mapped location 
of the SSZ is just about 50 km south of this region,, and therefore we interpret the 
intra-crustal layering in this region to be evidence of the location SSZ. 
  
We see one additional east-dipping intra-crustal layer in the northeast portion of our 
study that is not readily explained. We see two zones of sharp westward dipping 
intra-crustal layering: In the south, the change from flat to sharp agrees well with the 
locations of the ETSZ and the Brevard fault zone (BFZ), but in the north, the 
westward dipping seems to coincide with the central Virginia seismic zone (CVSZ). 
However, we see no evidence for intra-crustal layering beneath the South Carolina 
seismic zone (SCSZ). 
 
 127 
 
 
Figure 4.13- Dip direction of intra-crustal layering at each CCP stacking latitude and 
 longitude.  Points indicate locations in the CCP stacking grid that show intra-
 crustal layering. Red points indicate locations where a westward dipping intra-
 crustal layer is seen, purple points indicate the locations where a flat intra 
 crustal layer is seen, and light blue indicates the locations where an eastward 
 dipping intra-crustal layer is identified. Lavender shaded regions note the 
 location of seismic zones in the SEUS . The brown line in the western portion 
 of the study notes the predicted location of the Grenville Front  (GF), and the 
 green dotted lines indicate possible locations of suture zones based on 
 interpretations of intra-crustal layering (see text) . In the south, the green line 
 is interpreted as the Suwannee suture zone (SSZ) where in the north the green 
 dotted line indicates a suture of unknown origin, but may be the suture 
 between the Laurentian and Amazonian crust. 
4.5.2 Relationship between Structure and Seismicity  
The apparent correlation of westward dipping intra-crustal layering in two of the three 
seismic zones is surprising. Here we investigate the underlying structure in cross 
section view in order to shed light on what other crustal and lithospheric processes 
might explain the location of seismicity in the SEUS. 
  
The N-S trending cross section (D) in Figure 4.15 crosses through the main portion 
of the ETSZ. Starting beneath the Central Piedmont Suture (CPS), the Suwannee 
Suture Zone, and the Piedmont physiographic province, a strong mid-crustal 
discontinuity can be seen at ~10-20 km depth with an underlying Moho 40 km 
deeper. The mid-crustal discontinuity dips northward beneath the ETSZ to ~40 km 
thick while the deeper Moho layer dips northward to ~65 km thick. The Suwannee 
suture zone (SSZ) is also visible in this cross section. In the SP CCP stack, there is a 
southward-dipping strong intra-crustal phase. While a similar positive phase is visible 
the SR PRF CCP stack, it is not resolvable. What is clear is that beneath the SSZ the 
signal from the Moho changes and becomes weaker in both SR PRF and SP CCP 
 128 
 
 
stacks. We see no coherent negative phase in the lithosphere, implying that there is no 
MLD and that the LAB signal is weak or deeper than our CCP stacks. 
 
  42oN
  39oN
  36oN
  33oN
  30oN
  87oW  o84oW  81oW  78oW  75oW  72 W
D SR PRF CCP SRF CCP
CL AP BR V&R Piedmont ACP CL AP BR V&R Piedmont ACP
RT ETSZ BFZ SGR RT ETSZ BFZ SGR
0
40
80
120
160
200
27 127 227 327 427 527 627 727 827 927 1027 1127 27 127 227 327 427 527 627 727 827 927 1027 1127
Distance Along Profile (km) Distance Along Profile (km)
0 0
RF amplitude RF amplitude
D SR PRF CCP Variance SRF CCP Variance
0
40
80
120
160
200
27 127 227 327 427 527 627 727 827 927 1027 1127 27 127 227 327 427 527 627 727 827 927 1027 1127
Distance Along Profile (km) Distance Along Profile (km)
0 0
RF amplitude RF amplitude
 
Figure 4.14- N-S cross section D which crosses through the main seismicity in the 
 ETSZ. Here a strong mid-crustal discontinuity is seen dipping northward from 
 ~10 km beneath the CPS/SSZ to ~20 km beneath the ETSZ and BFZ. In the 
 same region, there is a sharp change in the crustal thickness from ~40 km to 
 ~60 km beneath the BFZ/ETSZ. Stacks below show the variance of the RF 
 amplitude at each point in the CCP stack. See caption in Figure 9 for 
 description of labels and abbreviations. 
 
 
 129 
 
Depth (km)
Depth (km)
GF
CPS/
SSZ
GF
CPS/
SSZ
 
A shallower and weaker westward-dipping intra-crustal layer is also seen beneath the 
Piedmont and the Central Virginia Seismic Zone (CVSZ-Figure 15). Here the W-E 
cross section (16) shows an intra-crustal layer that starts at around 15 km beneath the 
ACP and only deepens to 22-23 km beneath the BR. The more obvious change 
beneath the CVSZ is in the Moho depth itself. Beneath the ACP to the east, the Moho 
depth is not well resolved, but is likely around 25 km; directly below the Piedmont 
and the CVSZ, the Moho appears to deepen abruptly to around 45 km. The change in 
crustal thickness happens over a short lateral distance of about 100 km. As in the 
previous cross-section, there is no obvious evidence of a negative phase indicating the 
presence of a seismically resolvable MLD or LAB.  
 130 
 
 
  42oN
  39oN
  36oN
  33oN
  30oN
  87oW  84oW  81o o oW  72
o
W  78 W  75 W
16 SR PRF CCP SRF CCP
LP AP V&R Piedmont ACP LP AP V&R Piedmont ACP
RT AB CVSZ RT AB CVSZ
0
40
80
120
160
200
22 122 222 322 422 622 722 822 922 22 122 222 322 422 622 722 822 922
Distance Along Profile (km) Distance Along Profile (km)
0 0
RF amplitude RF amplitude
16 SR PRF CCP Variance SRF CCP Variance
0
40
80
120
160
200
22 122 222 322 422 622 722 822 922 22 122 222 322 422 622 722 822 922
Distance Along Profile (km) Distance Along Profile (km)
0 0
RF amplitude RF amplitude
 
Figure 4.15- W-E cross-section 16 through the CVSZ. Beneath the CVSZ we see an 
 intra-crustal layer dipping slightly westward from around 15 km beneath the 
 ACP and only deepens to 22-23 km beneath the BR. Directly below the CVSZ 
 there is a change in crustal thickness from ~25 km to ~45 km over a later 
 distance of about 100 km. Stacks below show the variance of the RF 
 amplitude at each point in the CCP stack. See caption in Figure 9 for 
 description of labels and abbreviations 
 
Unlike the two other seismogenic zones in the SEUS, There is no westward dipping 
mid-crustal layer present beneath the South Carolina seismic zone (SCSZ- Figure 
17). Instead we see a shallow slightly southeastward dipping or flat layer beneath the 
CPS and which continues beneath the SCSZ (see also Appendix: cross-section 6). 
There is, however, a fairly sharp change in the crustal thickness beneath the SCSZ 
 131 
 
Depth (km) Depth (km)
GF
BR
GF
BR
 
seen in the N-S cross section (H) where the Moho shallows from near 40 km to about 
25 km directly beneath the SCSZ and over a horizontal distance of approximately 50 
km. 
 
  42oN
  39oN
  36oN
  33oN
  30oN
  87oW  o o84 W  81oW  78oW  75oW  72 W
16 SR PRF CCP SRF CCP
LP AP V&R Piedmont ACP LP AP V&R Piedmont ACP
RT AB CVSZ RT AB CVSZ
0
40
80
120
160
200
22 122 222 322 422 622 722 822 922 22 122 222 322 422 622 722 822 922
Distance Along Profile (km) Distance Along Profile (km)
0 0
RF amplitude RF amplitude
16 SR PRF CCP Variance SRF CCP Variance
0
40
80
120
160
200
22 122 222 322 422 622 722 822 922 22 122 222 322 422 622 722 822 922
Distance Along Profile (km) Distance Along Profile (km)
0 0
RF amplitude RF amplitude
 
Figure 4.16- N-S cross section H through the SCSZ. In this cross section the Moho 
 appears fairly flat beneath the CL, AB, BR, and V&R. The only large change 
 in the Moho depth occurs almost directly below the SCSZ in the Piedmont. 
 Here the Moho changes from ~42 km to 30 km.  Unlike Figures 15 & 16, no 
 westward dipping intra-crustal layering is seen. Stacks below show the 
 variance of the RF amplitude at each point in the CCP stack. See caption in 
 Figure 9 for description of labels and abbreviations 
 
 132 
 
Depth (km) Depth (km)
GF
BR
GF
BR
 
Biryol et al. (2016) describe the region with little seismicity in North Carolina as the 
North Carolina seismic quiescence (NCSQ ? Figure 18). The W-E cross-section (10) 
shows the structure beneath the ETSZ as well as the NQSZ, allowing for direct 
comparison. Again, at the edge of the W-E cross-section, we see the double Moho 
feature near the ETSZ with a westward dip. However, beneath the region of NCSQ 
the Moho remains remarkably flat remaining at a constant depth of 40km from the 
CPS to the edge of the NCSQ. 
  42oN
  39oN
  36oN
  33oN
  30oN
  87oW  84oW  81oW  78oW  75oW  72
oW
10 SR PRF CCP SRF CCP
V&R BR Piedmont ACP V&R BR Piedmont ACP
ETSZ&BFZ NCSZQ ETSZ&BFZ NCSZQ
23 123 223 323 423 523 623 723 823 923 23 123 223 323 423 523 623 723 823 923
Distance Along Profile (km) Distance Along Profile (km)
RF amplitude RF amplitude
10 SR PRF CCP Variance SRF CCP Variance
23 123 223 323 423 523 623 723 823 923 23 123 223 323 423 523 623 723 823 923
Distance Along Profile (km) Distance Along Profile (km)
RF amplitude RF amplitude
 
Figure 4.17- W-E Cross-section 10 that passes through the NCSQ. Beneath the 
 NCSQ in the ACP we see both a flat Moho at 40km and no strong evidence 
 for intra-crustal layering. Stacks below show the variance of the RF amplitude 
 133 
 
Depth (km) Depth (km)
CPS
CPS
 
 at each point in the CCP stack. See caption in Figure 9 for description of 
 labels and abbreviations 
  
 4.6 Discussion  
4.6.1 The Extent of the Grenville Front in the SEUS 
Recently, it has been debated whether the edge of deformation associated with the 
Grenville orogeny could be seen in the United States (Stein et al., 2018). The 
deformation edge of the Grenville orogeny, called the Grenville front (GF), marks the 
extent of continental collision that made up the supercontinent Rodinia. Although the 
location of the GF has been associated in southern Canada with an eastward-dipping 
reflector through the GLIMPCE and LITHOPROBE projects (White et al., 2000), its 
location in the United States has remained much more elusive. While some studies 
suggest that what is interpreted as the GF actually corresponds to the edge of the 
magnetic anomaly beneath the mid-continent rift through Minnesota and Wisconsin 
(Stein et al. 2018); others associate the boundary to a dipping negative velocity 
discontinuity consisting of shear zones through the center of Ohio (Long et al., 2019). 
  
In this study, we find evidence for low angle eastward mid-crustal layering in 
Western Ohio and Northern Kentucky that flattens out at ~15 km beneath the 
Appalachian Plateau. We interpret the western edge of the eastern mid-crustal dipping 
structure to be evidence of crustal decollements created during the collision of 
Laurentia to create the supercontinent Rodinia. This north-west edge of the Grenville 
Front that we constrain agrees with the COCORP project and MAGIC deployment, 
 134 
 
 
which image a 50km dipping zone to 25-30 km depth (Culotta et al. 1990; Long et 
al., 2019). The vergence direction of subduction during the Grenville orogeny is still 
debated (Rivers, 1997; Hynes and Rivers, 2010; Boyce et al. 2016); however, if the 
subduction is N-NW, then the eastward dipping layering we see near the edge of the 
predicted GF may be evidence of subsequent crustal shortening during the formation 
of Pangea instead of crustal emplacement during crustal accretion during the 
formation of Rodinia.  
  
We do not find evidence of southernmost extent of the GF past Kentucky, which may 
suggest that our study region does not extend far enough westward to capture the 
edge of the deformation in the south. Because receiver functions image present-day 
structure, they cannot constrain the timing of layering; we cannot say for sure that the 
intra-crustal structure we see is the GF. Indeed, it is possible that the mid-crustal 
layering is a remnant feature inherited from one or more of the other continental 
collisions that make up the formation of the SEUS. However, the dipping layer we 
associate with the GF seems to disappear beneath the Rome Trough. Between 1.3-1 
Bya, the Grenville province experienced major granitoid intrusions, which would 
underlie the RT at the time of its failed rifting between 0.72-0.68 Bya (Whitmeyer and 
Karlstrom 2007) and likely alter the mid-crust, preventing imaging of mid-crustal 
layering prior to 1.3 Bya. Therefore, we may be able to infer that the western intra-
crustal layering must be related to a suture that occurred before 1.3-1 Gya.  
 
We do find another example of eastward dipping intra-crustal structure a few hundred 
kilometers away to the east, at the eastern edge of the Rome Trough (RT Appendix 
 135 
 
 
cross-sections 18-22). While this eastward dipping crustal structure extend to a 
similar depth and share a eastward dip direction, it appears distinct from the western 
Ohio GF. The location of the dipping intra-crustal layer observed begins at the 
eastward extent of the RT and Grenville province and at the edge of the rift boundary 
that marks the location of the breakup of Rodinia. Therefore, the crustal layering 
we observe east of the RT is likely evidence of a different suture zone, perhaps 
between Laurentian and Amazonian crust (Rivers 2014), Taconian arcs, or Avalon 
accreted terranes (Hatcher, 2010).   The presence of these multiple mid-crustal layers 
is consistent with the interpretation that the tectonics of crustal emplacement that we 
observe today in other locations around the world, were operating in the SEUS in the 
past. (Long et al., 2019; Hopper et al., 2017) 
 
 4.6.2 Evidence of the Suwanee Suture 
In addition to the GF, we also image crustal layering in the SEUS similar to that 
found by Hopper et al. (2017). Hopper et al. (2017) discuss the likelihood of 
shallowly-eastward-dipping layer as being the Suwannee Suture Zone. The Suwannee 
Suture zone marks the boundary between Laurentian and Gondwanan rocks during 
the formation of the supercontinent Pangea. Here, we map the extent of the 
southeastward dipping feature from our CCP stacks (Figure 14), and find that it 
agrees well with predicted locations of the suture within ~50km. In the SRFs, the 
dipping structure associated with the Suwannee Suture Zone continues into the South 
Georgia Rift, but this feature is not seen on the SR PRF CCP stacks. One possible 
explanation for this is that the PRFs might not be able to capture the impedance 
 136 
 
 
contrast if it occurs gradually over tens of kilometers, while the lower frequency 
SRFs can see it clearly. 
 
4.6.3 Westward dipping ?double Moho? 
Beneath the Blue Ridge and Piedmont in southeastern Tennessee and northern 
Georgia, we find evidence for a westward dipping intra-crustal layer that produces 
strong signatures in the SR PRF CCP stacks. This intra-crustal layer starts at ~10 km 
beneath the Piedmont near the CPS and dips down to ~45 km beneath the eastern 
Tennessee seismic zone 400 km to the west. The underlying Moho starts between 43-
40 km and dips downward to nearly 65 km in the same region. The COCORP line 
imaged a west-dipping layer in northernmost Alabama where the expectation was to 
image the edge of the southeastward-dipping Grenville Front (Culotta et al., 1990). If 
we assume that the subduction direction during the Grenville Orogeny is S-SE, the 
explanation for the westward dipping intra-crustal proposed in the CORCORP study 
is that the subduction polarity switched beneath this area with the accretion of a 
colliding arc terrane. Arc passive margin collision events can account for crustal 
growth and a subduction polarity reversal (Clift et al., 2003).   However, this active 
source study did not image a deeper crustal boundary below.  
 
Wagner et al. (2012) identified an intra-crustal layer and deeper Moho with the same 
characteristics we observe at the border of South Carolina, North Carolina, and 
Tennessee, where they interpret this feature to be a ?double-Moho?. A ?double-
Moho? could be caused by lithospheric downwelling (Zandt et al., 2004), partial 
 137 
 
 
eclogitization of the lower crust (Schulte-Pelkum et al. 2005), and/or wedge tectonics 
(Hansen and Dueker, 2009). It has been suggested that the only reliable seismic 
observable criterion for lower crustal eclogites is a double Moho observed near a 
Moho uplift that would be unrealistic if the Moho is interpreted as the base of a crust 
(Mjelde et al., 2013). While we observe the shallow layer extending up, too shallow 
to be interpreted as the Moho, the fact that it extends to ~10 km depth means that this 
cannot be explained by a boundary between a partially eclogitized lower crust alone. 
Wagner et al. (2012) interpret that an indenter (either the Carolina terrane or from 
crustal shortening) is thrust into the Laurentian crust, pushing some of it down where 
it could become partially ecologized, which is consistent with the isostasy in the 
region. However, in that study, they observe a Moho-hole, a region where only the 
deeper Moho exists, which is not what we observe in the CCP stacks. 
 
A deep >60 km Moho is observed beneath the Himalaya, overlain by a shallow 
interface that has been interpreted as a double-Moho (Schulte-Peklum et al. 2005). In 
the modern Himalayan continent-continent collision, the interpretation of Schulte-
Peklum et al. (2005) of the double Moho is that the shallow feature at ~15-20 km is 
occurring due to a strongly anisotropic layer at the decollment between the Indian and 
the Himalaya plates. At depths of 45-55 km, the shallower feature is interpreted as the 
boundary of partial eclogitization of the lower crust. If this interpretation were 
relevant to the SUES, it too, like the COCORP interpretation, would require a 
subduction polarity change. 
 
A subduction polarity change can occur with the accretion of an island arc, such as a 
 138 
 
 
central gneiss island arc accretion during the Precambrian, as proposed in Culotta et 
al. (1990). The double Moho we observe is consistent with a shear zone near the 
surface and partial eclogization of the lower crust observed in modern continental 
collisions. Given the extent of the sharply westward dipping feature we observe in the 
southern portion of Figure 14, Appendix C, we interpret that at shallow depths, the 
shallow intra-crustal layer is likely a remnant top of the shear zone created during 
subduction of an accreted island arc, while at deeper depths, it represents partial 
eclogization of the lower crust. Further study could provide insight into the 
anisotropic characteristics of the top layer, and determine whether it is likely to be 
related to shear at shallow depths.   
 
4.6.4 Correlation of Structure to SEUS Seismicity  
While two of the three seismic zones show correlations to westward-dipping intra-
crustal layers, we find all three show a relationship between sharp changes in crustal 
thickness and seismicity in contrast to the flat crust beneath the region of NCSQ. This 
observation agrees with that of Soto-Cordero et al. (2017), who argue that sharp 
crustal thickness changes can focus strain and control the location of seismicity in the 
Mid-Atlantic. 
  
This does not mean that changes in the underlying lithospheric structure beneath 
regions of seismicity does not also play a role in controlling the location of seismicity 
or that the model proposed by Biryol et al (2016) is incorrect. With CCP stacking, we 
only image the upper 200 km of the lithosphere. The lithospheric drip observed in the 
 139 
 
 
Biryol et al. (2016) study between 200-400 km.  Unlike lithospheric drip, if MLDs 
were related to seismicity in the SEUS, we would expect to find evidence supporting 
their existence, at least in the SRF CCP stacks. The lack of evidence supporting the 
relationship between mid-lithospheric discontinuities and seismicity on the east coast 
is telling in-and-of itself. We should be able to see MLDs if they exist, at least on the 
SP CCP stacks. The fact that we see no relationship between MLDs and seismicity 
(and no evidence of pervasive MLDs in general), leads us to conclude that MLDs are 
not related to intraplate seismicity on the east coast. 
 
4.6.5 On the lack of a lithosphere asthenosphere boundary or intra-
lithospheric layering  
The boundary between the lithosphere and asthenosphere should be associated with a 
negative phase that would imply a velocity decrease with depth. Unlike the strong 
positive Moho phase, no strong and coherent negative phase is visible in our CCP 
stacks. The lack of visible LAB beneath the SEUS using SR PRF and SRF CCP 
stacks implies that either the LAB across the SEUS occurs at depths greater than 200 
km, in agreement with magnetotelluric studies, or that the velocity decrease at the 
base of the lithosphere occurs over 40 km in depth, and therefore is too gradual to be 
imaged even by low frequency SRFs (Ford et al., 2010). Both a deep LAB and a 
diffuse LAB boundary would be consistent with old continental lithosphere and shear 
wave observations beneath the SEUS (Wagner et al. 2018). Therefore, this is our 
preferred interpretation of lack of clear LAB-related signatures in our CCP stacks. 
 140 
 
 
 
4.7 Conclusions 
Using sediment removed PRFs and SRFs, we constrain complex features in the 
lithosphere and attempt to place them in the context of the tectonic evolution of the 
SEUS. We find that the crustal thickness in the SEUS ranges from 25 km near the 
Atlantic coast to about 65 km beneath the Blue Ridge. The 65 km crust we interpret is 
surprisingly deep, but reflects the more complex structure we see in intra-crustal 
layering.  
 
We find evidence for both eastward and westward dipping intra crustal layering 
through the SEUS. We interpret the eastward dipping intra crustal layers as evidence 
of crustal emplacement marking the suture zone of past accretionary events. In the 
south, we map the Suwannee suture zone, where the Gondwanan crust is emplaced on 
Laurentian basement. In the northwest, we interpret the dipping intra-crustal layer to 
mark the Grenville Front, and in the northeast, that it marks a separate suture zone 
along the rift of Rodinia of unknown origin.  
 
We interpret the westward dipping intra-crustal layering beneath the ETSZ and BFZ 
in Tennessee to reflect a suture of arc accretion that changes the polarity of 
subduction to a westward dipping regime.  The ?double Moho? feature likely reflects 
similar processes observed in the Himalaya today, where the top boundary represents 
a shear zone at the detachment boundary which transitions to a partially ecologic 
lower crust at depth.  
 
 141 
 
 
We find a correlation between sharp changes in crustal structure and seismogenic 
zones in the SEUS and no correlation between seismogenic zones and lithospheric 
structure. This finding is in agreement with Soto-Cordero et al. (2018), which 
suggests that crustal structure and topography play a large role in dictating the 
locations of intraplate earthquakes.  
 
Finally, we find no evidence of a sharp lithosphere-asthenosphere boundary or 
evidence for mid-lithospheric discontinuities beneath the SEUS, which we interpret to 
suggest that the lithosphere-asthenosphere transition beneath the SEUS is diffuse, 
occurring over at least 40 km in depth.  
 
Currently, this study relies only on the USArray TA deployment; however, there are 
smaller-scale dense deployments in the SEUS that may help add higher resolution to 
some interesting areas. In addition, developments in receiver function migration 
techniques beyond the relatively simple CCP stacking used in this study should allow 
for better imaging of sharp lateral discontinuities. Constraining sharp lateral 
discontinuities of the Moho beneath seismic zones may further illuminate what 
crustal conditions control the locations of intraplate seismicity.  
 
  
 142 
 
 
Summary 
 
The goal of this project was to understand underlying structure of the Southeastern 
United States and its relationship to seismic hazards posed by ongoing intra-plate 
seismicity. Since the recent damaging 2011 Mw = 5.5 Mineral, VA earthquake, many 
studies have sought to understand the relationship between passive margin 
earthquakes and the underlying mechanisms that control them. Thanks to the large-
scale deployment of the EarthScope USArray Transportable Array and focused 
MAGIC and SESAME seismic deployments, seismologists have been able to image 
the subsurface of this region with significantly higher resolution than ever before. 
 
Despite the numerous recent studies of the SEUS, constraining the basic structure of 
the lithosphere has proven difficult. Receiver functions are perhaps the most useful 
seismic tool to image discontinuities beneath a seismic station such as the depth to 
basement, Moho, lithosphere-asthenosphere boundary, and any intra-lithospheric 
layering. However, while SRFs can yield interpretable images of these structures, 
their low frequency content results in relatively low resolution. On the other hand, 
interpreting deeper structure with high-frequency PRFs is challenging as signal from 
deeper structures can become overprinted by multiples and reverberations from the 
sedimentary basins. This problem is significant as over 50% of this region is overlain 
by sediment including the Appalachian basin, Mississippi embayment, and Atlantic 
coastal plain. 
 
To tackle this challenge and enable high-resolution PRF imaging of lithospheric 
structure across the SEUS, we develop a method to suppress sediment reverberations 
 143 
 
 
in PRFs using a resonance removal filter. We also leverage complimentary SRFs and 
SsPmp RFs that are not so affected by sedimentary reverberations, and use them to 
construct H-k-VP triple stacks. This method enables use to constrain the crustal 
thickness, VP/VS ratio, and average crustal P-wave velocity in regions where sediment 
reverberations can lead to incorrect interpretations of the crustal properties (Chapter 
2). 
 
We find that the parameters used to remove sediment reverberations are also useful to 
constrain important parameters of ground motion during an earthquake for seismic 
hazard analysis. Because earthquakes on the East Coast are generally smaller in 
magnitude and occur more infrequently, fewer studies have been done that constrain 
the basin structure. By using measurements of basin phase arrivals on high frequency 
PRFs, we are able to constrain both the site fundamental frequency and average 
velocity profiles of the SEUS.  The fundamental frequency is important for 
understanding how large building and structures will be affected by ground shaking 
as matching resonant frequencies between structures and basin can be particularly 
damaging.  We validate our method by comparing against more traditional methods 
of estimating basin fundamental frequency. In addition we provide equations of VP 
and VS profiles to 2.5 km depth for the Atlantic Coastal Plain (Chapter 3). 
 
Finally, we apply the methodology developed in Chapter 2 to the SEUS to better 
understand its history of inherited structures and relationship to intraplate seismicity. 
By constructing common conversion point stacks of both RR PRFs and SRFs, we 
map the lithospheric structure across the SEUS. We find no evidence of a strong LAB 
 144 
 
 
or intra-lithospheric layering to 200 km. We do find evidence for multiple intra-
crustal layers that are likely related to inherited structures from past continental 
collision events, but that have no correlation with regions of seismicity. We do, 
however, note that directly beneath the eastern Tennessee seismic zone, central 
Virginia seismic zone, and South Carolina seismic zone a sharp change in Moho 
depth occurs. In contrast, beneath the region of North Carolina seismic quiescence the 
Moho depth remains remarkably flat and consistent (Chapter 4). 
 
The obvious avenues of future work for this research would be to apply the new 
methods of RR H-k-VP stacking and fundamental frequency from high frequency 
PRFs methods to other regions where sediment reverberations limit study of 
lithospheric structure. Specific regions of interest could include other areas with 
intraplate seismicity, such as the New Madrid seismic zone in southern Illinois and 
Missouri to investigate if a sharp crustal thickness exists beneath.  
 
Below the crust in the CCP stacks calculated for the SEUS, we see evidence of Moho 
multiples coincident with expect MLD/LAB arrivals. Using the timing of these 
reverberations can help improve constraints on crustal structure in CCP stacks, since 
the arrival depths (and times) of Moho multiples in the CCP stacks depend on crustal 
thickness and also crustal Vp and Vs, ray parameter and back azimuth (e.g. Jenkins et 
al. 2018).  
 
Calculating P-to-s receiver functions from Ocean Bottom Seismometers (OBS) has 
proven to be even more challenging as water-wave reverberations overprint P-to-s 
 145 
 
 
arrivals and noise on the horizontal components can introduce artifacts into the 
receiver functions (Bostock and Trehu 2012, Janiszewski and Abers 2015). New 
techniques for removing OBS noise can reduce these undesirable effects, though 
typically only at long periods (Ruan et al 2014, Bell et al 2015). Additionally, ocean 
margin settings contain thick low velocity sedimentary layers along the Cascadia 
subduction zone (1-2 km) and on the offshore Eastern North American Margin 
(ENAM 5-10 km). This low velocity sediment may be responsible for significant 
reverberations that obscure interpretations made from P-to-s receiver functions and 
overprint crustal and lithospheric arrivals. Extending methodologies developed above, 
one could develop a systematic method of calculating sediment-removed P-to-s 
receiver functions for otherwise challenging OBS stations to answer outstanding 
questions about the sediment and lithospheric structure on both the Cascadia 
subduction zone and ENAM. 
 
In addition, this research motivates future investigations using high frequency 
receiver functions to understand the very near surface. Previously, it has been 
suggested that constraining the S-wave velocity structure in the sediment could be 
done using body wave polarization analysis and the frequency-dependence of 
apparent incidence angle of teleseismic P waves. However, when sediment 
reverberations contaminate the high frequency RF signal, the apparent incidence 
angle method fails to provide useful constraints on the sediment structure, instead 
producing velocities that consistently over and underestimate the actual velocity 
structure. If the SR removal filter successfully removes the reverberations whilst 
 146 
 
 
preserving the direct sediment conversions, then it is possible the apparent incidence 
method would be useful. Producing a site-specific velocity structure using SR and the 
apparent incidence angle method would provide a more detailed model of the 
sediment velocity structure for hazard analysis. 
 
 
 147 
 
 
Appendices 
 
 
Appendix A: Table A.1 
All station measurements of stations used in Chapter 3 that have been determined to 
lie on the ACP (according to Fenneman physiographic provinces of the US, 1947) 
including: Depth to basement, two-way S-wave travel time in the basement TSs, and 
amplitude r0, travel time of the PPbs phase TPPbs, and amplitude (PPbs), and the basin 
fundamental frequency F0 at each station. 
 
Station Depth to 
name basement (km) TSs (s) r0 TPPbs (s) PPbs F0 
153A 0 0.47 0.58 0.2 0.93 1.06 
154A 0.157 1.28 0.39 0.55 1.17 0.39 
155A 0.278 1.3 0.5 0.6 1.13 0.38 
156A 0.395 1.69 0.37 0.75 0.64 0.3 
157A 0.464 1.62 0.48 0.9 0.56 0.31 
158A 0.701 1.79 0.53 0.96 0.58 0.28 
251A 0.061 1.32 0.41 0.66 1.51 0.38 
252A 0.233 1.65 0.51 0.75 0.99 0.3 
253A 0.279 1.62 0.52 0.85 0.65 0.31 
254A 0.766 2.36 0.45 1.05 0.58 0.21 
255A 0.796 1.95 0.51 0.85 0.52 0.26 
256A 0.758 2.15 0.52 0.97 0.29 0.23 
257A 0.877 1.9 0.41 0.8 0.75 0.26 
351A 1.667 2.48 0.34 1.2 0.31 0.2 
352A 1.275 2.27 0.36 1.15 0.82 0.22 
353A 1.606 2.5 0.4 1.36 0.43 0.2 
355A 1.358 1.88 0.3 0.85 0.65 0.27 
356A 1.172 2.09 0.38 1.09 0.38 0.24 
357A 1.168 2.09 0.48 0.9 0.3 0.24 
452A 2.19 2.82 0.32 1.35 0.39 0.18 
453A 2.112 3.1 0.14 0.9 0.18 0.16 
454A 2.484 1.35 0.39 0.75 0.17 0.37 
455A 2.27 2.38 0.12 0.65 0.16 0.21 
456A 1.519 1.37 0.17 0.9 0.26 0.36 
457A 1.592 2.06 0.26 0.34 0.07 0.24 
553A 3.237 1.65 0.36 0.8 0.18 0.3 
554A 4.378 1.32 0.41 0.65 0.16 0.38 
 148 
 
 
555A 3.958 1.09 0.41 0.5 0.15 0.46 
CBN 0.334 0.63 0.69 0.15 1.52 0.79 
CNNC 0.059 0.52 0.68 0.1 1.31 0.96 
GWDE 1.521 1.92 0.37 0.91 0.6 0.26 
N62A 0.257 0.5 0.63 0.1 1.08 1.01 
NHSC 0.451 1.42 0.5 0.7 0.8 0.35 
O61A 0.567 0.77 0.66 0.25 1.27 0.65 
P61A 1.309 1.88 0.54 0.69 0.41 0.27 
Q59A 0.524 1.23 0.51 0.55 0.82 0.41 
Q60A 1.07 1.88 0.55 0.8 0.64 0.27 
Q61A 1.875 2.11 0.48 1.05 0.45 0.24 
R59A 0.461 1.12 0.41 0.5 0.83 0.45 
R60A 0.718 2.11 0.47 0.85 0.56 0.24 
R61A 2.134 2.68 0.2 0.91 0.11 0.19 
S59A 0.371 0.52 0.67 0.2 1.06 0.96 
S60A 0.595 1.53 0.52 0.65 1.05 0.33 
S61A 1.795 2.59 0.34 0.8 0.19 0.19 
T60A 0.494 1.19 0.56 0.55 0.8 0.42 
TIGA 1.427 2.36 0.37 1.2 0.37 0.21 
U60A 0.137 0.7 0.72 0.2 1.29 0.71 
U61A 0.751 2.41 0.49 0.85 0.59 0.21 
V60A 0.174 0.84 0.73 0.25 1.01 0.59 
V61A 0.657 2.18 0.51 0.75 0.46 0.23 
V62A 1.496 2.59 0.45 0.61 0.36 0.19 
W60A 0.13 0.79 0.73 0.3 0.94 0.63 
W61A 0.471 1.6 0.38 0.5 0.87 0.31 
X59A 0.082 0.68 0.78 0.1 1.23 0.74 
X60A 0.151 0.82 0.62 0.35 1.08 0.61 
Y57A 0.007 0.5 0.71 0.1 0.68 1.01 
Y58A 0.151 0.98 0.64 0.45 0.88 0.51 
Y59A 0.213 1.03 0.53 0.4 1.02 0.49 
Y60A 0.288 1.03 0.55 0.45 0.88 0.49 
Z55A 0 0.66 0.64 0.15 1.46 0.76 
Z56A 0.081 0.86 0.69 0.35 1.15 0.58 
Z57A 0.226 1.23 0.48 0.5 0.89 0.41 
Z58A 0.412 1.35 0.54 0.65 0.62 0.37 
Z59A 0.522 1.3 0.52 0.55 1.04 0.38 
 
 
 
 149 
 
 
 
Appendix B: CCP Figures for Chapter 4 
Appendix B contains all additional SR PRF and SPRF CCP stacking cross-sections 
for Chapter 4. Cross Sections are either W-E trending and are given numbers that 
correspond to their S-N location or are N-S trending and are given letters that 
correspond to their W-E location. Each set of 4 cross sections contains a map (top left 
corner) with cross section locations and directions for reference. 
 
Key for figures: labels in black are features which the cross sections pass through, 
and locations can be found in Figure 4.1 in Chapter 4.  Lines below labels indicate the 
distance on the cross section that each cross-section passes through. RT= Rome 
trough, GF= Grenville front, SGR = South Georgia rift, ETSZ = Eatern Tennessee 
seismic zone, BFZ = Brevard fault zone, CPS= central Piedmont suture, NCSQ = 
North Carolina seismic quiescence, AB = Appalachian basin, SCSZ = South Carolina 
seismic zone, SSZ = Suwanee suture zone, CVSZ = central Virginia seismic zone  
 150 
 
 
 151 
 
 
 152 
 
 
 153 
 
 
 154 
 
 
 155 
 
 
 
 156 
 
 
 157 
 
 
 
 158 
 
 
 159 
 
 
 
 
 160 
 
 
 
 
 161 
 
 
 
 
 
 
 
 162 
 
 
Bibliography 
 
Abrahamson, N., Silva, W., 2008. Summary of the Abrahamson &amp; Silva NGA 
Ground-Motion Relations. Earthq. Spectra 24, 67?97. 
https://doi.org/10.1193/1.2924360 
Aki, K., Richards, P.G., 1980. Quantitative Seismology, second eid. ed. University 
Science Books. 
Allmendinger, R.W., Serpa, L., Smith, R.B., n.d. Cenozoic and Mesozoic structure of 
the eastern Basin and Range province, Utah, from COCORP seismic-reflection 
data. https://doi.org/10.1130/0091-7613(1983)11<532:CAMSOT>2.0.CO;2 
American Association of Petroleum Geologists, 1967. Basement map of North 
America between latitudes 24? and 60? N. Basement Rock Proj. Committee, U.S 
Geol. Surv. map67001271. 
Audet, P., Bostock, M.G., Christensen, N.I., Peacock, S.M., 2009. Seismic evidence 
for overpressured subducted oceanic crust and megathrust fault sealing. Nature 
457, 76?78. https://doi.org/10.1038/nature07650 
B. Herrmann, R., 1979. Surface wave focal mechanisms for eastern North American 
earthquakes with tectonic implications. J. Geophys. Res. 84, 3543?3552. 
https://doi.org/10.1029/JB084iB07p03543 
Baise, L.G., Kaklamanos, J., Berry, B.M., Thompson, E.M., 2016. Soil amplification 
with a strong impedance contrast: Boston, Massachusetts. Eng. Geol. 202, 1?13. 
https://doi.org/10.1016/J.ENGGEO.2015.12.016 
Bassin, C., Laske, G., Mastern, G., 2000. The Current limits of Resolution for Surface 
Wave Tomography in North America, in: EOS Trans AGU. 
Biryol, C.B., Lee, S.J., Lees, J.M., Shore, M.J., 2018. Lithospheric structure of an 
incipient rift basin: Results from receiver function analysis of Bransfield Strait, 
NW Antarctic Peninsula. Polar Sci. 0?1. 
https://doi.org/10.1016/j.polar.2018.02.003 
Biryol, C.B., Wagner, L.S., Fischer, K.M., Hawman, R.B., 2016. Relationship 
between observed upper mantle structures and recent tectonic activity across the 
Southeastern United States. J. Geophys. Res. Solid Earth 121, 3393?3414. 
https://doi.org/10.1002/2015JB012698 
 163 
 
 
Bodin, P., Smith, K., Horton, S., Hwang, H., 2001. Microtremor observations of deep 
sediment resonance in metropolitan Memphis, Tennessee. Eng. Geol. 62, 159?
168. https://doi.org/10.1016/S0013-7952(01)00058-8 
Borcherdt, R.., 1970. Effects of Local Geology on Ground Motion Near San 
Francisco Bay. Bull. Seismol. Soc. Am. 60, 29?61. 
https://doi.org/10.2307/1779766 
Bostock, M.G., 1998. Mantle stratigraphy and evolution of the Slave province. J. 
Geophys. Res. 103, 21183. https://doi.org/10.1029/98JB01069 
Bostock, M.G., 1996. Ps conversions from the upper mantle transition zone beneath 
the Canadian landmass. J. Geophys. Res. Earth 101, 8393?8402. 
Bostock, M.G., Kumar, M.R., 2010. Bias in seismic estimates of crustal properties. 
Geophys. J. Int. 182, 403?407. https://doi.org/10.1111/j.1365-
246X.2010.04623.x 
Bracewell, R., 1968. The Fourier Transform and Its Applications, 2nd editio. ed, 
Integers. McGraw-Hill, New York. https://doi.org/10.1119/1.1973431 
Brocher, T.M., 2005. Empirical relations between elastic wavespeeds and density in 
the Earth?s crust. Bull. Seismol. Soc. Am. 95, 2081?2092. 
https://doi.org/10.1785/0120050077 
Chai, C., Ammon, C.J., Maceira, M., Herrmann, R.B., 2015. Inverting interpolated 
receiver functions with surface wave dispersion and gravity: Application to the 
western U.S. and adjacent Canada and Mexico. Geophys. Res. Lett. 42, 4359?
4366. https://doi.org/10.1002/2015GL063733 
Chapman, C.H., 2003. Yet another elastic plane-wave, layer-matrix algorithm. 
Geophys. J. Int. 154, 212?223. https://doi.org/10.1046/j.1365-
246X.2003.01958.x 
Chapman, D.M.F., Godin, O.A., 2001. Dispersion of interface waves in sediments 
with power-law shear speed profiles. II. Experimental observations and seismo-
acoustic inversions. J. Acoust. Soc. Am. 110, 1908?1916. 
https://doi.org/10.1121/1.1401739 
Chen, K.-C., Chiu, J.-M., Yang, Y.-T., 1996. Shear-wave velocity of the sedimetnary 
basin in the upper Mississippi embayment using S-to-P converted waves. Bull. 
Seismol. Soc. Am. 86, 848?856. 
Christensen, N.I., 1996. Poisson?s ratio and crustal seismology. J. Geophys. Res. 
Solid Earth 101, 3139?3156. https://doi.org/10.1029/95JB03446 
 164 
 
 
Chulick, G.S., Mooney, W.D., 2002. Seismic structure of the crust and uppermost 
mantle of north America and adjacent oceanic basins: A synthesis. Bull. 
Seismol. Soc. Am. 92, 2478?2492. https://doi.org/10.1785/0120010188 
Clark, S.H.B., n.d. Location of the Southern Appalachians Birth of the Mountains 
The Geologic Story of the Southern Appalachian Mountains. 
Clayton, R.W., Wiggins, R.A., 1976. Source shape estimation and devoncolution of 
teleseismic body waves. Geophys. J. R. Astron. Soc. 151?177. 
https://doi.org/https://doi.org/10.1111/j.1365-246X.1976.tb01267.x 
Clowes, R.M., Brandon, M.T., Green, A.G., Yorath, C.J., Sutherland Brown, A., 
Kanasewich, E.R., Spencer, C., 1987. LITHOPROBE-southern Vancouver 
Island: Cainozoic subduction complex image by deep seismic reflections. Can. J. 
Earth Sci. 24, 31?51. https://doi.org/10.1139/e87-004 
COOK, F.A., 1988. Proterozoic thin-skinned thrust and fold belt beneath the Interior 
Platform in northwest Canada. Geol. Soc. Am. Bull. 100, 877?890. 
https://doi.org/10.1130/0016-7606(1988)100<0877:PTSTAF>2.3.CO;2 
Cook, F.A., Brown, L.D., Kaufman, S., Oliver, J.E., Petersen, T.A., 1981. COCORP 
seismic profiling of the Appalachian orogen beneath the coastal plain of 
Georgia. Geol. Soc. Am. Bull. 92, 738?748. https://doi.org/10.1130/0016-
7606(1981)92<738:CSPOTA>2.0.CO;2 
Crotwell, H. Philip and Owens, T.J. (Department of G.S.U. of S.C., 2005. Automated 
Receiver Function Processing [WWW Document]. Seismol. Res. Lett. 
https://doi.org/10.17611/DP/EARS.1. 
Culotta, R.C., Pratt, T., Oliver, J., 1990. A tale of two sutures: COCORP?s deep 
seismic surveys of the Grenville province in the eastern U.S. midcontinent. 
Geology 18, 646. https://doi.org/10.1130/0091-
7613(1990)018<0646:ATOTSC>2.3.CO;2 
Delgado, J., L?pez Casado, C., Giner, J., Est?vez, A., Cuenca, A., Molina, S., 2000. 
Microtremors as a Geophysical Exploration Tool: Applications and Limitations. 
Pure Appl. Geophys. 157, 1445?1462. https://doi.org/10.1007/PL00001128 
DiPietro, J.A., n.d. Geology and landscape evolution??: general principles applied to 
the United States. 
Dreiling, J., Isken, M.P., Mooney, W.D., Park, M., Chapman, M.C., Godbee, R.W., 
2014. PACIFIC EARTHQUAKE ENGINEERING NGA-East Regionalization 
Report??: Comparison of Four Crustal Regions within Central and Eastern North 
America using Waveform Acceleration Response Virginia Polytechnic Institute 
and State University. 
 165 
 
 
Dueker, K.G., Sheehan, A.F., 1997. Mantle discontinuity structure from midpoint 
stacks of converted P to S waves across the Yellowstone hotspot track. J. 
Geophys. Res. Solid Earth 102, 8313?8327. https://doi.org/10.1029/96JB03857 
Earle, P.S., Shearer, P.M., 1994. Characterization of global seismograms using an 
automatic-picking algorithm. Bull. Seismol. Soc. Am. 84, 366?376. 
Eaton, D.W., Dineva, S., Mereu, R., 2006. Crustal thickness and VP/VS variations in 
the Grenville orogen (Ontario, Canada) from analysis of teleseismic receiver 
functions. Tectonophysics 420, 223?238. 
https://doi.org/10.1016/j.tecto.2006.01.023 
Fairbanks, C.D., Andrus, R.D., Camp, W.M., Wright, W.B., 2008. Dynamic Periods 
and Building Damage at Charleston, South Carolina During the 1886 
Earthquake. Earthq. Spectra 24, 867?888. https://doi.org/10.1193/1.2980346 
Farra, V., Vinnik, L., 2000. Upper mantle stratification by P and S receiver functions. 
Geophys. J. Int. 141, 699?712. https://doi.org/10.1046/j.1365-
246x.2000.00118.x 
Fenneman, N.M., Johnson, D.W., 1946. Physiographic divisions of the conterminous 
U.S. 
Fischer, K.M., Ford, H.A., Abt, D.L., Rychert, C.A., 2010. The Lithosphere-
Asthenosphere Boundary. Annu. Rev. Earth Planet. Sci. 38, 551?575. 
https://doi.org/10.1146/annurev-earth-040809-152438 
Fischer, K.M., Salvati, L.A., Hough, S.E., Gonzalez, E., Nelsen, C.E., Roth, E.G., 
2009. Bulletin of the Seismological Society of America . J. Geol. 21, 288?288. 
https://doi.org/10.1086/622062 
Flores, J., Novaro, O., Seligman, T.H., 1987. Possible resonance effect in the 
distribution of earthquake damage in Mexico City. Nature 326, 783?785. 
https://doi.org/10.1038/326783a0 
Frankel, A.D., Carver, D.L., Williams, R.A., 2002. Nonlinear and Linear Site 
Response and Basin Effects in Seattle for the M 6.8 Nisqually, Washington, 
Earthquake. Bull. Seismol. Soc. Am. 92, 2090?2109. 
https://doi.org/10.1785/0120010254 
Frassetto, A., Zandt, G., Gilbert, H., Owens, T.J., Jones, C.H., 2010. Improved 
imaging with phase-weighted common conversion point stacks of receiver 
functions. Geophys. J. Int. 182, no?no. https://doi.org/10.1111/j.1365-
246X.2010.04617.x 
 166 
 
 
French, S.W., Fischer, K.M., Syracuse, E.M., Wysession, M.E., 2009. Crustal 
structure beneath the Florida-to-Edmonton broadband seismometer array. 
Geophys. Res. Lett. 36, L08309. https://doi.org/10.1029/2008GL036331 
Hansen, S., Dueker, K., n.d. P-and S-Wave Receiver Function Images of Crustal 
Imbrication beneath the Cheyenne Belt in Southeast Wyoming. 
https://doi.org/10.1785/0120080168 
Hatcher, R.D., 2010. The Appalachian orogen: A brief Summary. Geol. Soc. Am. 
Memoir 206. https://doi.org/10.1130/2010.1206(01) 
Hopper, E., Fischer, K.M., Wagner, L.S., Hawman, R.B., 2017. Reconstructing the 
end of the Appalachian orogeny. Geology 45, 15?18. 
https://doi.org/10.1130/G38453.1 
Joyner, W.B., Warrick, R.E., Fumal, T.E., 1981. The effect of quaternary alluvium on 
strong ground motion in the Coyote Lake, California, earthquake of 1979. Bull. 
Seismol. Soc. Am. 71, 1333?1349. 
Juli?, J., Ammon, C.J., Herrmann, R.B., Correig, A.M., 2000. Joint inversion of 
receiver function and surface wave dispersion observations. Geophys. J. Int. 143, 
99?112. https://doi.org/10.1046/j.1365-246X.2000.00217.x 
Kang, D., Yu, C., Ning, J., Chen, W., 2016. Simultaneous Determination of Crustal 
Thickness and P Wavespeed by Virtual Deep Seismic Sounding (VDSS). 
Seismol. Res. Lett. 87, 1104?1111. https://doi.org/10.1785/0220160056 
Kawakatsu, H., Kumar, P., Takei, Y., Shinohara, M., Kanazawa, T., Araki, E., 
Suyehiro, K., 2009. Seismic evidence for sharp lithosphere-asthenosphere 
boundaries of oceanic plates. Science 324, 499?502. 
https://doi.org/10.1126/science.1169499 
Kennett, B.L.N., 1991. The Removal of Free Surface Interactions From Three?
Component Seismograms. Geophys. J. Int. 104, 153?154. 
https://doi.org/10.1111/j.1365-246X.1991.tb02501.x 
Kennett, B.L.N., Engdahl, E.R., Buland, R., 1995. Constraints on seismic velocities in 
the Earth from traveltimes. Geophys. J. Int. 122, 108?124. 
https://doi.org/10.1111/j.1365-246X.1995.tb03540.x 
KIKUCHI, M., KANAMORI, H., 1982. Inversion of complex body waves. Bull. 
Seismol. Soc. Am. 72, 491?506. 
Kolb, J.M., Leki , V., 2014. Receiver function deconvolution using transdimensional 
hierarchical Bayesian inference. Geophys. J. Int. 197, 1719?1735. 
https://doi.org/10.1093/gji/ggu079 
 167 
 
 
Koper, K.D., Burlacu, R., 2015. The fine structure of double-frequency microseisms 
recorded by seismometers in North America. J. Geophys. Res. Solid Earth 120, 
1677?1691. https://doi.org/10.1002/2014JB011820 
Kumar, M.R., Bostock, M.G., 2008. Extraction of absolute P velocity from receiver 
functions. Geophys. J. Int. 175, 515?519. https://doi.org/10.1111/j.1365-
246X.2008.03963.x 
Kumar, M.R., Saul, J., Sarkar, D., Kind, R., Shukla, A.K., 2001. Crustal structure of 
the Indian shield: New constraints from teleseismic receiver functions. Geophys. 
Res. Lett. 28, 1339?1342. https://doi.org/10.1029/2000GL012310 
Lachet, C., Hatzfeld, D., Bard, P.-Y., Theodulidis, N., Papaioannou, C., Savvaidis, 
A., 1996. Site Effects and Microzonation in the City of Thessaloniki (Greece) 
Comparison of Different Approaches, Bulletin of the Seismological Society of 
America. 
Langston, C.A., 1979. Structure under Mount Rainier, Washington, inferred from 
teleseismic body waves. J. Geophys. Res. 84, 4749. 
https://doi.org/10.1029/JB084iB09p04749 
Langston, C.A., Horton, S.P., 2014. Three-dimensional seismic-velocity model for 
the unconsolidated mississippi embayment sediments from H/V ambient noise 
measurements. Bull. Seismol. Soc. Am. 104, 2349?2358. 
https://doi.org/10.1785/0120140026 
Laske, G., Mastern, G., Ma, Z., Pasyanos, M., 2013. CRUST 1.0: An Updated Global 
Model of Earth?s Crust. 
Lawrence, D.P., Hoffman, C.W., 1993. Geology of Basement Rocks Beneath the 
North Carolina Coastal Plain. 
Lawrence, J.F., Shearer, P.M., 2006. Constraining seismic velocity and density for the 
mantle transition zone with reflected and transmitted waveforms. Geochemistry, 
Geophys. Geosystems 7. https://doi.org/10.1029/2006GC001339 
Lawrence, J.F., Wiens, D.A., 2004. Combined receiver-function and surface wave 
phase-velocity inversion using a niching genetic algorithm: Application to 
Patagonia. Bull. Seismol. Soc. Am. 94, 977?987. 
https://doi.org/10.1785/0120030172 
Leahy, G.M., Saltzer, R.L., Schmedes, J., 2012. Imaging the shallow crust with 
teleseismic receiver functions. Geophys. J. Int. 191, 627?636. 
https://doi.org/10.1111/j.1365-246X.2012.05615.x 
 168 
 
 
Leki?, V., Fischer, K.M., 2017. Interpreting spatially stacked Sp receiver functions. 
Geophys. J. Int. 210, 874?886. https://doi.org/10.1093/gji/ggx206 
Lermo, J., Chavez-Garcia, F., 1994. Are Microtremors useful in site response 
evaluation? Bull. Seismol. Soc. Am. 84, 1350?1364. 
Levander, A., Miller, M.S., 2012. Evolutionary aspects of lithosphere discontinuity 
structure in the Western U.S. Geochemistry, Geophys. Geosystems 13, n/a?n/a. 
https://doi.org/10.1029/2012GC004056 
Levin, V., Park, J., 1997. P-SH conversions in a flat-layered medium with anisotropy 
of arbitrary orientation. Geophys. J. Int. 131, 253?266. 
https://doi.org/10.1111/j.1365-246X.1997.tb01220.x 
Li, A., 2002. Crust and upper mantle discontinuity structure beneath eastern North 
America. J. Geophys. Res. 107, 2100. https://doi.org/10.1029/2001JB000190 
Ligorria, J.P., Ammon, C.J., 1999. Iterative deconvolution and receiver-function 
estimation. Bull. Seismol. Soc. Am. 89, 1395?1400. 
Liu, G., Persaud, P., Clayton, R.W., n.d. Structure of the Northern Los Angeles 
Basins Revealed in Teleseismic Receiver Functions from Short-Term Nodal 
Seismic Arrays. https://doi.org/10.1785/0220180071 
Liu, H.-P., Hu, Y., Dorman, J., Chang, T.-S., Chiu, J.-M., 1997. Upper Mississippi 
embayment shallow seismic velocities measured in situ. Eng. Geol. 46, 313?330. 
https://doi.org/10.1016/S0013-7952(97)00009-4 
Liu, L., Gao, S.S., 2018. Lithospheric layering beneath the contiguous United States 
constrained by S-to-P receiver functions. Earth Planet. Sci. Lett. 495, 79?86. 
https://doi.org/10.1016/J.EPSL.2018.05.012 
Long, M.D., Benoit, M.H., Aragon, J.C., King, S.D., 2019. Seismic imaging of mid-
crustal structure beneath central and eastern North America: Possibly the elusive 
Grenville deformation? Geology 47, 371?374. https://doi.org/10.1130/G46077.1 
Long, M.D., Ford, H.A., Abrahams, L., Wirth, E.A., 2017. The seismic signature of 
lithospheric deformation beneath eastern North America due to Grenville and 
Appalachian orogenesis. Lithosphere 9, 987?1001. 
https://doi.org/10.1130/L660.1 
Luzi, L., Puglia, R., Pacor, F., Gallipoli, M.R., Bindi, D., Mucciarelli, M., 2011. 
Proposal for a soil classification based on parameters alternative or 
complementary to Vs,30. Bull. Earthq. Eng. 9, 1877?1898. 
https://doi.org/10.1007/s10518-011-9274-2 
 169 
 
 
Lynner, C., Porritt, R.W., 2017. Crustal structure across the eastern North American 
margin from ambient noise tomography. Geophys. Res. Lett. 44, 6651?6657. 
https://doi.org/10.1002/2017GL073500 
Mendecki, M.J., Bieta, B., Mycka, M., 2014. Determination of the resonance 
frequency ? thickness relation based on the ambient seismic noise records from 
Upper Silesia Coal Basin. Contemp.Trends.Geosci 3, 41?51. 
https://doi.org/10.2478/ctg-2014-0011 
Mooney, W.D., Andrews, M.C., Ginzburg, A., Peters, D.A., Hamilton, R.M., 1983. 
Crustal structure of the northern mississippi embayment and a comparison with 
other continental rift zones. Dev. Geotecton. 19, 327?348. 
https://doi.org/10.1016/B978-0-444-42198-2.50025-0 
Murphy, B.S., Egbert, G.D., 2017. Electrical conductivity structure of southeastern 
North America: Implications for lithospheric architecture and Appalachian 
topographic rejuvenation. Earth Planet. Sci. Lett. 462, 66?75. 
https://doi.org/10.1016/J.EPSL.2017.01.009 
Musacchio, G., Mooney, W.D., Luetgert, J.H., Christensen, N.I., 1997. Composition 
of the crust in the Grenville and Appalachian Provinces of North America 
inferred from V P /V S ratios. J. Geophys. Res. Solid Earth 102, 15225?15241. 
https://doi.org/10.1029/96JB03737 
Nakamura, Y., 2000. CLEAR IDENTIFICATION OF FUNDAMENTAL IDEA OF 
NAKAMURA?S TECHNIQUE AND ITS APPLICATIONS. 12WCEE. 
Nakamura, Y., 1989. A Method for Dynamic Characteristics Estimation of 
Subsurface using Microtremor on the Ground Surface. Q. Rep. Railw. Tech. 
Res. Insitute 30. 
Narayan, J.P., Kumar, R., n.d. Spatial spectral amplification of basin-transduced 
Rayleigh waves 3D Kinematic strong ground motion simulation View project 
Spatial spectral amplification of basin-transduced Rayleigh waves. 
https://doi.org/10.1007/s11069-013-0917-2 
Olugboji, T.M., Park, J., 2016. Crustal anisotropy beneath Pacific Ocean-Islands from 
harmonic decomposition of receiver functions. Geochemistry, Geophys. 
Geosystems 17, 810?832. https://doi.org/10.1002/2015GC006166 
Owens, T.J., Zandt, G., 1997. Implications of crustal property variations for models 
of Tibetan plateau evolution. Nature 387, 37?43. 
https://doi.org/10.1038/387037a0 
Owens, T.J., Zandt, G., Taylor, S.R., 1984. Seismic evidence for an ancient rift 
beneath the Cumberland Plateau, Tennessee: A detailed analysis of broadband 
 170 
 
 
teleseismic P waveforms. J. Geophys. Res. 89, 7783?7795. 
https://doi.org/https://doi.org/10.1029/JB089iB09p07783 
Parker, E.H., Hawman, R.B., Fischer, K.M., Wagner, L.S., 2016. Estimating crustal 
thickness using SsPmp in regions covered by low-velocity sediments: Imaging 
the Moho beneath the Southeastern Suture of the Appalachian Margin 
Experiment (SESAME) array, SE Atlantic Coastal Plain. Geophys. Res. Lett. 43, 
9627?9635. https://doi.org/10.1002/2016GL070103 
Parker, E.H., Hawman, R.B., Fischer, K.M., Wagner, L.S., 2013. Crustal evolution 
across the southern Appalachians: Initial results from the SESAME broadband 
array. Geophys. Res. Lett. 40, 3853?3857. https://doi.org/10.1002/grl.50761 
Parolai, S., Picozzi, M., Richwalski, S.M., Milkereit, C., 2005. Joint inversion of 
phase velocity dispersion and H/ /V ratio curves from seismic noise recordings 
using a genetic algorithm, considering higher modes. 
https://doi.org/10.1029/2004GL021115 
Perron, V., G?lis, C., Froment, B., Hollender, F., Bard, P.-Y., Cultrera, G., Cushing, 
E.M., 2018. Can broad-band earthquake site responses be predicted by the 
ambient noise spectral ratio? Insight from observations at two sedimentary 
basins. Geophys. J. Int. 215, 1442?1454. https://doi.org/10.1093/gji/ggy355 
Pesce, K.A., 2010. Comparison of Receiver Function Deconvolution Techniques. 
Massachusetts Insitute of Technology. 
Petersen, T.A., Brown, L.D., Cook, F.A., Kaufman, S., Oliver, J.E., 1984. Structure 
of the Riddleville Basin from COCORP Seismic Data and Implications for 
Reactivation Tectonics. J. Geol. 92, 261?271. https://doi.org/10.1086/628859 
Pitilakis, K., Riga, E., Anastasiadis, A., 2013. New code site classification, 
amplification factors and normalized response spectra based on a worldwide 
ground-motion database. Bull. Earthq. Eng. 11, 925?966. 
https://doi.org/10.1007/s10518-013-9429-4 
Pratt, T.-L., Brocher, T.-M., Weaver, C.-S., Miller, K.-C., Trhu, A.-M., Creager, K.-
C., Crosson, R.-S., Snelson, C.-M., 2003. Amplification of seismic waves by the 
Seattle basin, northwestern U.S. Bull. Seism. Soc. Am. 93, 533?545. 
Pratt, T.L., 2018. Characterizing and Imaging Sedimentary Strata Using Depth?
Converted Spectral Ratios: An Example from the Atlantic Coastal Plain of the 
Eastern United States. Bull. Seismol. Soc. Am. 108, 2801?2815. 
https://doi.org/10.1785/0120180046 
Pratt, T.L., Wright Horton Jr, J., Mu?oz, J., Hough, S.E., Chapman, M.C., Guney 
Olgun, C., n.d. Amplification of Earthquake Ground Motions in Washington, 
 171 
 
 
DC, and Implications for Hazard Assessments in Central and Eastern North 
America. https://doi.org/10.1002/2017GL075517 
Ramesh, D.S., Kind, R., Yuan, X., 2002. Receiver function analysis of the North 
American crust and upper mantle. Geophys. J. Int. 150, 91?108. 
https://doi.org/10.1046/j.1365-246X.2002.01697.x 
Reeves, Z., Leki?, V., Schmerr, N., Kohler, M., Weeraratne, D., 2015. Lithospheric 
structure across the California Continental Borderland from receiver functions. 
Geochemistry, Geophys. Geosystems 16, 246?266. 
https://doi.org/10.1002/2014GC005617 
Rivers, T., 2014. Tectonic Setting and Evolution of the Grenville Orogen: An 
Assement of Progress over the last 40 years. Geosci. Canada 42, 77?124. 
Ross, Gerald, M., Milkereit, B., Eaton, D., White, D., Kanasewich, E.R., Burianyk, 
M., n.d. Paleoproterozoic collisional orogen beneath the Western Canada 
Sedimentary Basin imaged by Lithoprobe crustal seismic-reflection data 
Electrical conduction View project Simulating hydraulic fracturing using Finite-
Discrete Element Methods FDEM View projec. https://doi.org/10.1130/0091-
7613(1995)023<0195:PCOBTW>2.3.CO;2 
Rychert, C.A., Harmon, N., 2016. Stacked P-to-S and S-to-P receiver functions 
determination of crustal thickness, Vp, and Vs: The H-V stacking method. 
Geophys. Res. Lett. 43, 1487?1494. https://doi.org/10.1002/2015GL067010 
Rychert, C.A., Rondenay, S., Fischer, K.M., 2007. P-to-S and S-to-P imaging of a 
sharp lithosphere-asthenosphere boundary beneath eastern North America. J. 
Geophys. Res. Solid Earth 112, 1?21. https://doi.org/10.1029/2006JB004619 
Schimmel, M., Paulssen, H., 1997. Noise reduction and detection of weak, coherent 
signals through phase-weighted stacks. Geophys. J. Int. 130, 497?505. 
https://doi.org/10.1111/j.1365-246X.1997.tb05664.x 
Schulte-Pelkum, V., Mahan, K.H., Shen, W., Stachnik, J.C., 2017. The distribution 
and composition of high-velocity lower crust across the continental U.S.: 
Comparison of seismic and xenolith data and implications for lithospheric 
dynamics and history. Tectonics 36, 1455?1496. 
https://doi.org/10.1002/2017TC004480 
Schulte-Pelkum, V., Monsalve, G., Sheehan, A., Pandey, M.R., Sapkota, S., Bilham, 
R., Wu, F., 2005. Imaging the Indian subcontinent beneath the Himalaya. Nature 
435, 1222?1225. https://doi.org/10.1038/nature03678 
 172 
 
 
Seismological Society of America., C.A., 1911. Bulletin of the Seismological Society 
of America., Bulletin of the Seismological Society of America. Seismological 
Society of America. 
Seismological Society of America., E.H., Hough, S.E., Jacob, K.H., 1911. Bulletin of 
the Seismological Society of America., Bulletin of the Seismological Society of 
America. Seismological Society of America. 
Shearer, P.M., 2009. Introduction to Seismology, 2nd edition. ed. Cambridge 
University Press, Cambridge. 
Shearer, P.M., Orcutt, J.A., 1987. Surface and near-surface effects on seismic 
waves?theory and borehole seismometer results. Bull. Seismol. Soc. Am. 77, 
1168?1196. 
Sheehan, A.F., Abers, G.A., Jones, C.H., Lerner-Lam, A.L., 1995. Crustal thickness 
variations across the Colorado Rocky Mountains from teleseismic receiver 
functions. J. Geophys. Res. 100, 20391. https://doi.org/10.1029/95JB01966 
Sheehan, A.F., Dueker, K., Gilbert, H.J., Sheehan, A.F., Dueker, K.G., Molnar, P., 
Gilbert, H.J., Dueker, K.G., Molnar, P., 2003. Receiver functions in the western 
United States, with implications for upper mantle structure and dynamics 
Tsunami Data Assimilation Without a Dense Observation Network View project 
Integrated Geological and Geophysical Studies of the Western U.S. and adjacent 
Mexico and Canada View project Receiver functions in the western United 
States, with implications for upper mantle structure and dynamics. Artic. J. 
Geophys. Res. Atmos. 108, 2229. https://doi.org/10.1029/2001JB001194 
Shen, W., Ritzwoller, M.H., Schulte-Pelkum, V., 2013. A 3-D model of the crust and 
uppermost mantle beneath the Central and Western US by joint inversion of 
receiver functions and surface wave dispersion. J. Geophys. Res. Solid Earth 
118, 262?276. https://doi.org/10.1029/2012JB009602 
Soto?Cordero, L., Meltzer, A., Stachnik, J.C., 2018. Crustal Structure, Intraplate 
Seismicity, and Seismic Hazard in the Mid?Atlantic United States. Seismol. Res. 
Lett. 89, 241?252. https://doi.org/10.1785/0220170084 
Stein, C.A., Stein, S., Elling, R., Keller, G.R., Kley, J., 2018. Is the ?Grenville Front? 
in the central United States really the Midcontinent Rift? GSA Today 4?10. 
https://doi.org/10.1130/GSATG357A.1 
Stephenson, W.J., Asten, M.W., Odum, J.K., Frankel, A.D., 2019. Shear?Wave 
Velocity in the Seattle Basin to 2 km Depth Characterized with the krSPAC 
Microtremor Array Method: Insights for Urban Basin?Scale Imaging. Seismol. 
Res. Lett. https://doi.org/10.1785/0220180194 
 173 
 
 
Thybo, H., Perchu?, E., Zhou, S., 2000. Intraplate earthquakes and a seismically 
defined lateral transition in the upper mantle. Geophys. Res. Lett. 27, 3953?
3956. https://doi.org/10.1029/2000GL011636 
Tian, X., Chen, Y., Tseng, T.L., Klemperer, S.L., Thybo, H., Liu, Z., Xu, T., Liang, 
X., Bai, Z., Zhang, X., Si, S., Sun, C., Lan, H., Wang, E., Teng, J., 2015. Weakly 
coupled lithospheric extension in southern Tibet. Earth Planet. Sci. Lett. 430, 
171?177. https://doi.org/10.1016/j.epsl.2015.08.025 
Tsai, V.C., 2009. On establishing the accuracy of noise tomography travel-time 
measurements in a realistic medium. Geophys. J. Int. 178, 1555?1564. 
https://doi.org/10.1111/j.1365-246X.2009.04239.x 
Tsai, V.C., Atiganyanun, S., 2014. Green?s functions for surface waves in a generic 
velocity structure. Bull. Seismol. Soc. Am. 104, 2573?2578. 
https://doi.org/10.1785/0120140121 
van Der Baan, M., 2009. The origin of SH-wave resonance frequencies in 
sedimentary layers, in: Geophysical Journal International. Wiley/Blackwell 
(10.1111), pp. 1587?1596. https://doi.org/10.1111/j.1365-246X.2009.04245.x 
Vinnik, L.P., 1977. Detection of waves converted from P to SV in the mantle. Phys. 
Earth Planet. Inter. 15, 39?45. https://doi.org/10.1016/0031-9201(77)90008-5 
Wagner, L.S., Fischer, K.M., Hawman, R., Hopper, E., Howell, D., 2018. The relative 
roles of inheritance and long-term passive margin lithospheric evolution on the 
modern structure and tectonic activity in the southeastern United States. 
Geosphere 14, 1385?1410. https://doi.org/10.1130/GES01593.1 
Wagner, L.S., Stewart, K., Metcalf, K., 2012. Crustal-scale shortening structures 
beneath the Blue Ridge Mountains, North Carolina, USA. Lithosphere 4, 242?
256. https://doi.org/10.1130/L184.1 
Ward, K.M., Lin, F.-C., n.d. On the Viability of Using Autonomous Three-
Component Nodal Geophones to Calculate Teleseismic Ps Receiver Functions 
with an Application to Old Faithful, Yellowstone. 
https://doi.org/10.1785/0220170051 
Warren, D.H., Healy, J.H., Jackson, W.H., 1966. Crustal seismic measurements in 
southern Mississippi. J. Geophys. Res. 71, 3437?3458. 
https://doi.org/10.1029/JZ071i014p03437 
Wells, D., Egan, J.A., Murphy, D.G., Paret, T., 2015. Ground shaking and structural 
response of the Washington Monument during the 2011 Mineral, Virginia, 
earthquake. pp. 199?233. https://doi.org/10.1130/2015.2509(12) 
 174 
 
 
White, D.J., Lucas, S.B., Hajnal, Z., Green, A.G., Lewry, J.F., Weber, W., Bailes, 
A.H., Syme, E.C., Ashton, K., n.d. Paleo-Proterozoic thick-skinned tectonics: 
Lithoprobe seismic reflection results from the eastern Trans-Hudson Orogenl. 
Wiener, N., 1964. Extrapolation, interpolation, and smoothing of stationary time 
series with engineering applications. Technology Press of the Massachusetts 
Institute of Technology. 
Wilson, D.C., Angus, D.A., Ni, J.F., Grand, S.P., 2006. Constraints on the 
interpretation of S-to- P receiver functions. Geophys. J. Int. 165, 969?980. 
https://doi.org/10.1111/j.1365-246X.2006.02981.x 
W?lbern, I., R?mpker, G., 2017. Limitations of H-? stacking: ambiguous results 
caused by crustal layering. J. Seismol. 21, 221?235. 
https://doi.org/10.1007/s10950-016-9599-z 
Yantis, B.R., Costain, J.K., Ackerman, H.D., 1983. A reflection seismic study near 
Charleston, South Carolina. U.S. Geol. Surv. Prossfesional Pap. 1313-G. 
Yao, H., Campman, X., de Hoop, M. V., van der Hilst, R.D., 2009. Estimation of 
surface wave Green?s functions from correlation of direct waves, coda waves, 
and ambient noise in SE Tibet. Phys. Earth Planet. Inter. 177, 1?11. 
https://doi.org/10.1016/J.PEPI.2009.07.002 
Yassminh, R., Gallegos, A., Sandvol, E., Ni, J., 2019. Investigation of the Regional 
Site Response in the Central and Eastern United States. Bull. Seismol. Soc. Am. 
https://doi.org/10.1785/0120180230 
Yeck, W.L., Sheehan, A.F., Schulte-Pelkum, V., 2013. Sequential H-  Stacking to 
Obtain Accurate Crustal Thicknesses beneath Sedimentary Basins. Bull. 
Seismol. Soc. Am. 103, 2142?2150. https://doi.org/10.1785/0120120290 
Yilar, E., Baise, L.G., Ebel, J.E., 2017. Using H/V measurements to determine depth 
to bedrock and Vs30 in Boston, Massachusetts. Eng. Geol. 217, 12?22. 
https://doi.org/10.1016/J.ENGGEO.2016.12.002 
Yu, C.-Q., Chen, W.-P., van der Hilst, R.D., 2013. Removing source-side scattering 
for virtual deep seismic sounding (VDSS). Geophys. J. Int. 195, 1932?1941. 
https://doi.org/10.1093/gji/ggt359 
Yu, C.Q., Chen, W.P., Ning, J.Y., Tao, K., Tseng, T.L., Fang, X.D., John Chen, Y., 
van der Hilst, R.D., 2012. Thick crust beneath the Ordos plateau: Implications 
for instability of the North China craton. Earth Planet. Sci. Lett. 357-358, 366?
375. https://doi.org/10.1016/j.epsl.2012.09.027 
 175 
 
 
Yu, Y., Song, J., Liu, K.H., Gao, S.S., 2015. Determining crustal structure beneath 
seismic stations overlying a low-velocity sedimentary layer using receiver 
functions. J. Geophys. Res. Solid Earth 120, 3208?3218. 
https://doi.org/10.1002/2014JB011610 
Yuan, H., Dueker, K., Stachnik, J., 2010. Crustal structure and thickness along the 
Yellowstone hot spot track: Evidence for lower crustal outflow from beneath the 
eastern Snake River Plain. Geochemistry, Geophys. Geosystems 11, 1?14. 
https://doi.org/10.1029/2009GC002787 
Zandt, G., Gilbert, H., Owens, T.J., Ducea, M., Saleeby, J., Jones, C.H., 2004. Active 
foundering of a continental arc root beneath the southern Sierra Nevada in 
California. Nature 431, 41?46. https://doi.org/10.1038/nature02847 
Zelt, B.C., Ellis, R.M., 1999. Receiver-function studies in the Trans-Hudson Orogen , 
Saskatchewan. Can. J. Earth Sci. 603, 585?603. 
https://doi.org/https://doi.org/10.1139/e98-109 
Zhao, J.X., Xu, H., 2013. A Comparison of VS 30  and Site Period as Site?Effect 
Parameters in Response Spectral Ground?Motion Prediction Equations. Bull. 
Seismol. Soc. Am. 103, 1?18. https://doi.org/10.1785/0120110251 
Zhu, L., Kanamori, H., 2000. Moho depth variation in southern California from 
teleseismic receiver functions, J. Geophys. Res 105, 2969?2980. 
https://doi.org/10.1029/1999JB900322 
 
 
 
 
 
 
 176