See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/274246373

FuncLab: A MATLAB Interactive Toolbox for

Handling Receiver Function Datasets

Article in Seismological Research Letters · May 2012

DOI: 10.1785/gssrl.83.3.596

CITATIONS READS

16 244

2 authors:

Kevin C. Eagar Matthew J. Fouch

Shell Oil USA Samara/Data
10 PUBLICATIONS 114 CITATIONS 150 PUBLICATIONS 3,357 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Contiguous U.S. lithosphere View project

All content following this page was uploaded by Matthew J. Fouch on 10 August 2016.

The user has requested enhancement of the downloaded file.

 E L E C T R O N I C

S E I S M O L O G I S T
Electronic Seismologist

FuncLab: A MATLAB Interactive Toolbox with receiver function data. We describe the structural build-
ing blocks that make FuncLab an efficient computational sys-
for Handling Receiver Function Datasets tem, features of the primary GUIs, and three additional mod-
ules for commonly employed analysis for regional network and
Kevin C. Eagar1,2 and Matthew J. Fouch1,3
single station studies.
INTRODUCTION OVERVIEW AND DATASET PREPARATION
Receiver functions have become a widely used tool to study
FuncLab is a data management system that works in the
the seismic velocity structure beneath a single station by imag-
MATLAB environment. It is a suite of tools designed to man-
ing impedance boundaries where mode conversions from P
age receiver function data and metadata, visually inspect indi-
to S waves occur. In recent years, the community has seen an
vidual receiver functions and seismograms for quality control,
increase in the amount of broadband seismic data available
and visualize the results of the quality-controlled data. The
through systems such as the IRIS Data Management Center
package also includes a number of add-on modules that handle
(IRIS DMC), requiring new approaches to managing and han-
post-processing analysis and stacking. We use this module-
dling large datasets in computationally efficient ways. Many
based approach to provide the opportunity for future develop-
receiver function studies have focused on tectonic and geody-
ment of new FuncLab add-ons by the community.
namic questions on regional scales of a few hundred square kilo-
Prior to using FuncLab, the user usually computes receiver
meters (e.g., Zandt et al. 1995; Sheehan et al. 1995; Dueker and
functions and performs signal processing functions, where the
Sheehan 1997). These studies typically have utilized data from
choice of parameters is completely user-dependent. Generally
temporary networks operating on time scales of a few months
speaking, a typical receiver function computation utilizes
to a few years. Others have concentrated on global trends, using
three-component broadband seismic data of a recorded earth-
data from increasing numbers stations that have operated for
years or decades (e.g., Andrews and Deuss 2008; Lawrence and
Shearer 2006; Tauzin et al. 2008). Over the last several years, 3000
the seismological community has seen a rapid increase in the
number of deployed passive broadband seismic stations and a
2500
resulting increase in data available for receiver function analysis.
Large initiatives such as EarthScope’s USArray Transportable
# BB Stations

Array enable immediate open access to data and have provided 2000
the community with an order of magnitude more data than in
the previous three decades (Figure 1) based on smaller regional
1500
temporary studies using data from IRIS PASSCAL-style field
experiments (e.g., Suckale et al. 2009; Wilson et al. 2005).
While the computation of receiver functions has been 1000
well studied and is a relatively straightforward and rapid pro-
cess, an efficient method to analyze and further evaluate very
500
large receiver function datasets is not widely available. In this
paper, we present an efficient workflow and MATLAB toolbox
for organizing receiver function data, which we call FuncLab. 0
FuncLab is a package of m-files (MATLAB scripts) written for 1985 1990 1995 2000 2005 2010

the purpose of providing a management system to interface Year

▲▲ Figure 1. Number of broadband seismic stations reporting
1. Arizona State University, School of Earth and Space Exploration,
data to the IRIS Data Management Center (DMC). Data were
Tempe, Arizona, U.S.A.
2. Now at Shell Exploration and Production Company, Houston, Texas gathered on 5 March 2012 using an IRIS SeismiQuery data sum-
3. Now at Department of Terrestrial Magnetism, Carnegie Institution mary query for all broadband stations reporting to the DMC
of Washington, Washington, DC (http://www.iris.edu/SeismiQuery/summaries.htm).

596  Seismological Research Letters  Volume 83, Number 3  May/June 2012 doi: 10.1785/gssrl .83.3.596
quake in the range of 30°–95° distance from the station. The ferences in directory structure. No additional MATLAB tool-
horizontal recordings are rotated along the free surface to the boxes are required, but it is recommended that the Mapping
point in the radial and tangential directions from the backazi- Toolbox be installed in order to use all the plotting capabilities.
muth of the earthquake. A method of source normalization is The package contains a collection of m-files designed around
then performed by deconvolving the vertical seismogram from the concept of a project, or a complete dataset that can be pro-
the radial seismogram. The FuncLab package includes a set of cessed and analyzed collectively.
C-shell scripts that outline the basic preprocessing steps using A FuncLab project contains a set of related imported
the Seismic Analysis Code (SAC) and iterative deconvolution seismic data, files used to organize metadata and store proj-
receiver function computation using a Fortran 77 code devel- ect information, and other files created during data analyses.
oped and distributed by C. Ammon (Ligorria and Ammon Each project in FuncLab is housed in a separate directory with
1999) with parameters used for upper mantle and crustal stud- an organized structure created when the project is initiated.
ies of the Pacific Northwest (Eagar et al. 2010; 2011). FuncLab FuncLab is designed so that the main part of the program is
requires data in SAC format (Goldstein et al. 2003) and auto- responsible for managing the data and metadata within a proj-
matically handles little and big endian byte ordering of the ect and basic visualization of the dataset. Additional analytical
binary files. Currently, other data formats are not supported. tools are not contained in the main program, but instead are
Data must be organized in a formal directory structure incorporated as separate add-on modules that utilize conven-
before working with it in FuncLab. This structure was origi- tions specified in FuncLab to operate on the data.
nally used for data gathered using the Standing Order for
Data (SOD) data request tool (Owens et al. 2004), and there- Main FuncLab GUI
fore FuncLab was designed with this structure in mind. An Once receiver function data are organized with the FuncLab
example SOD recipe is also included on the FuncLab Web conventions described in the FuncLab User Manual, the user
site, http://geophysics.asu.edu/funclab. FuncLab also expects invokes the main FuncLab Graphical User Interface (GUI)
that the seismogram and receiver function files have a specific at the MATLAB command line using the funclab command.
naming convention, which partly arose from the original use The GUI is used from this point forward. The start of a new
of the iterative time-domain deconvolution code (Ligorria and project requires at least one directory containing data that has
Ammon 1999). The radial and transverse receiver functions, been preprocessed as described above and in the User Manual.
as well as the original vertical, radial, and transverse seismo- FuncLab automatically creates the project directory and struc-
grams, are needed to utilize the full functionality of the pack- ture while importing data directories to the project.
age. More details regarding proper setup of data directories and FuncLab saves the metadata for each station-earthquake
SAC file processing, including necessary header information, pair of waveform files in the Project.mat MAT-file. Metadata
can be found in the User Manual. for the records, such as station name, event origin time, time-
The IRIS Data Management Center (DMC) distrib- series sample rate, and SAC file pathname, are stored within
utes a data product providing fully automated computation this file. One of the most common limitations of the MATLAB
of receiver functions for all open broadband seismic stations environment is its inefficiency in storing large amounts of data
(http://www.iris.edu/dms/products/ears/). The purpose of the in memory, since heavy memory usage slows dramatically the
EarthScope Automated Receiver Survey (EARS) (Crotwell performance of computations. Instead of storing the seismic
and Owens 2005) is to provide automated analysis of bulk waveform data from an entire FuncLab project, we optimize
crustal properties using the well-established method of H-κ our algorithms by loading data only when needed for viewing
stacking (Zhu and Kanamori 2000). The DMC now includes or computation. This strategy significantly speeds up processes
a method for downloading all EARS receiver functions on a and enables the development of larger project datasets.
station-by-station basis. We provide a shell script to convert The main FuncLab GUI allows the user to interact with
EARS-formatted files and directory structure to a format read- the entire dataset (Figure 2). A drop-down menu defines the
able by FuncLab, allowing users easy access to receiver func- manner in which receiver function metadata is organized
tion data without computing the time series themselves. The for visualization in the GUI (we refer to these structures as
script includes instructions for direct download of EARS data. “tables,” although FuncLab is not a formal relational database
FuncLab therefore provides a straightforward application to management system). Records common to a particular sta-
customize receiver function analysis that leverages EARS data tion are viewed in “Station Tables.” The stations are then listed
products beyond the automated design of the EARS system. and selectable in the left-hand panel. The records within the
“Station Table” are listed by the event origin time in the cen-
TOOLBOX DESCRIPTION ter panel. The total number of records and number of active
records are listed below the panel. Metadata about the selected
This section describes aspects of the primary FuncLab data record in the center panel are then listed in the right-hand
management system. FuncLab is compatible with MATLAB panel. The user can also view the records organized by event
version 7.1.0.21 and above, but at the time of publication has origin by selecting the “Event Tables” in the drop-down menu.
not been tested beyond version 7.11 (R2010b). At present, Other options in the drop-down menu become available when
FuncLab is not compatible on Windows platforms given dif- using add-on analysis tools discussed below.

Seismological Research Letters  Volume 83, Number 3  May/June 2012  597

▲▲ Figure 2. Black and white screenshot of main FuncLab GUI. Actual FuncLab window is in color. Top-left drop-down box lists the
types of tables used to organize receiver function records. Tables are listed in the left panel. Records in the selected table (highlighted
in left panel) are listed in the center panel. Metadata for the selected record (highlighted in center panel) are shown in the right panel.
Information about number of tables, records, and active records are listed below the left and center panels, respectively. Text dis-
played at the bottom is used to convey short messages to the user about ongoing or finished processes.

Trace Editor New data that have not yet been evaluated through the Trace
FuncLab provides the powerful aspect of organization and effi- Editor are colored red, while data previously viewed in the GUI
ciency through its ability to visualize many data at once. One are colored blue. All records with status “off” are shaded gray.
of the main interactive visualization tools is the Trace Editor Furthermore, right-clicking the waveform brings up a menu to
(Figure 3). The primary function of the Trace Editor is to start another GUI for visualizing the seismograms and radial
evaluate and select waveforms efficiently for further analysis. receiver function changing the record status, and displaying
Selecting the Manual Trace Edit GUI will read in the receiver event and record information useful in determining the appro-
function SAC files of the records in the selected table in the priate status of the record (Figure 4).
left-hand panel. Receiver functions are also normalized to the FuncLab also allows for automated trace editing, which is
largest positive amplitude for each trace. The trace editing GUI currently based on two criteria, and is intended to be a first pass
displays the receiver functions as time series waveforms with through a new receiver function dataset. First, a record’s status
time increasing down the y-axis and each record one unit wide is set to “on” if the maximum amplitude after one pulse width
across the x-axis. from time zero is above a chosen threshold. The pulse width is
Each record’s status can be changed in the GUI, allowing defined as
the user complete flexibility to specify good records that will
be used in future analyses (i.e., stacking and visualization pro- 1
cesses). Records with status “off” are not used in subsequent PW = , (1)
α 2 ln2
analyses but remain part of the dataset and are not deleted
from the project. This approach has the advantage of greater where α is the Gaussian parameter. The second criterion is a
freedom to perform multiple stages of trace editing or easily minimum fit or measure of uncertainty in the receiver func-
change the status of a record at the user’s discretion. The color tion, set in the SAC header (see User Manual). All other records
of the waveform is another indicator of the record’s status. that do not meet these criteria are set to “off.”

598  Seismological Research Letters  Volume 83, Number 3  May/June 2012

▲▲ Figure 3. Black and white screenshot of FuncLab Trace Editing GUI. Actual FuncLab window is in color. Waveform time series
records in a selected table are displayed with colored fill denoting positive amplitudes and dark gray fill denoting negative amplitudes.
Check boxes above waveforms allow the user to change the record status to “on” (selected) or “off” (unselected). Unselected records
are shaded light gray. Scrollbar at the bottom allows the user to scan through all records listed in the table. The “Save Edits” button
saves the record status to the project file.

▲▲ Figure 4. Screenshot of trace editing system. Vertical, radial, and transverse seismograms and the associated radial receiver func-
tion are displayed. Event magnitude, distance, and depth are listed at top left. Status check box allows the user to change the record
status to “on” (checked) or “off” (unchecked). This screen is accessed by right-clicking a waveform in the trace editing GUI.

Seismological Research Letters  Volume 83, Number 3  May/June 2012  599

Data Visualizations horizontal interfaces. It is a back-projection method in which
A number of data plotting functions are available to help guide the receiver function amplitudes are placed in the appropriate
data analyses and are accessible from the main FuncLab GUI. position along the raypath based on a time-to-depth conversion.
Figure 5 shows three examples of the kinds of visualizations These amplitudes are binned at each depth based on the posi-
provided with this package. The Seismograms and Record tion of the ray piercing point and then stacked to enhance the
Section plots (Figure 5A) display the vertical, radial, and trans- signal originating from that approximate position. The stacking
verse seismograms and radial and transverse receiver func- algorithms are described in detail elsewhere (e.g., Dueker and
tions. The RF Moveout Record Section plot displays receiver Sheehan 1997; Eagar et al. 2010, 2011). Here, the focus is on the
function traces of active records vs. ray parameter, and the RF implementation and examples of these algorithms in FuncLab.
Moveout Image plot (Figure 5B) displays stacked records into M-files for this add-on are found in the subdirectory addons/ccp.
ray parameter bins with amplitudes shown by a color scale. RF The CCP GUI contains editable parameters on the left-
Backazimuth Record Section and RF Backazimuth Image plots hand side and functional pushbuttons for data processing on
(Figure 5C) are similar except that records are displayed vs. the right-hand side. The user may set parameters for bootstrap
backazimuth. The Dataset Statistics visualization is not a true stacking, which is a repeating process of randomly resampling
graph, but displays text summarizing various statistics about the data a fixed number of times and linearly stacking receiver
the dataset, such as the number of total active records. Another function amplitudes to produce a mean stack and 95% confi-
data coverage visualization is the Ray Parameter/Backazimuth dence interval (Efron and Tibshirani 1986, 1991). The imaging
Diagram that displays these metadata on a polar plot. Finally, volume is defined from the latitude, longitude, and depth range
two mapping options, a Station Map and a Global Event Map, input. The CCP bin spacing and width (circular radius) is also
are available if the MATLAB Mapping Toolbox is installed. set in the parameters window. The user can specify the receiver
A log file is initiated when the project is first created functions to include in the stacks from ranges of both backa-
within the FuncLab project and maintains information about zimuth and ray parameter. This add-on package also includes
each process performed on the dataset. The log file can be 1-D ray tracing and time-to-depth conversion that uses a cho-
viewed from the main FuncLab GUI but is a text file that can sen 1-D velocity model. Included in the package are the global
be viewed by any text viewer. Dates and times of processes are 1-D models AK135 (Kennett et al. 1995), IASP91 (Kennett
logged, as well as the name of the process and certain prefer- and Engdahl 1991), and PREM (Dziewonski and Anderson
ences set to run the process. The log file should not be edited 1981). Further details on how to define other models are
and is only intended for user guidance. described within the software manual. Also included is a linear
travel time correction algorithm that utilizes a seismic waves-
ADD-ON MODULES peed tomography model to account for 3-D heterogeneities in
the imaging volume. We currently provide only one regional
FuncLab is designed to allow new analysis tools to be built tomography model, published in Roth et al. (2008); however,
around the data and metadata management environment of the the User Manual provides details on how to define other mod-
main FuncLab GUI and project file. Therefore, analyses that els. Finally, a number of visualization tools, including plots of
involve stacking and other processing algorithms are included CCP bin stacks and bin histograms from bootstrapping, are
as add-on modules. The functionality of the main FuncLab accessible from the menu at the top of the CCP GUI.
GUI and trace editing GUI are not dependent upon the add- The “CCP Processes” buttons are ordered to follow an
ons; however, each add-on must contain m-files of a certain established workflow for producing CCP stacked images.
structure, reference the metadata variables in the Project.mat Subsequent processes use the matrices generated during the
file, and reference data files contained in the RAWDATA initial processing steps, so if the status of records changes
directory in order to interact with the main FuncLab GUI. In from trace editing, the CCP Processes workflow needs to be
this first public release of FuncLab, we provide three built-in reinitiated. The user must first perform 1-D ray tracing, which
add-on modules of commonly employed analysis tools, includ- requires the maximum depth of the imaging volume, depth
ing common conversion point stacking, Gaussian-weighted increment spacing, and 1-D velocity model to be set. The cor-
common conversion point stacking, and H-κ stacking. These rection for 3-D heterogeneity is an optional step and requires
tools can be accessed from the main FuncLab GUI and are the user to choose a tomography model, which can be defined
described in more detail in this section. or imported by the user as explained in the User Manual. The
Add-on modules follow a certain form and set of conven- next step in CCP processing is the time-to-depth conversion
tions in order to work with FuncLab. We include a set of tem- of the receiver functions. This process assigns receiver function
plate m-files to help users develop their own add-on FuncLab amplitudes to CCP bins at each depth increment specified by
modules. Additional details regarding add-on development the user. The final step in the workflow is CCP stacking. The
can be found in the User Manual. user may choose a range of backazimuths and ray parameters
over which to stack, although the default is the entire dataset,
Common Conversion Point Stacking from 0 to 360° and 0.04 to 0.08 sec/km, respectively. Each
Common conversion point (CCP) stacking of receiver functions defined range will create a separate CCP bin stack, which
is a method used to image smooth regional topography on sub- allows the user to explore azimuthal and ray angle variations

600  Seismological Research Letters  Volume 83, Number 3  May/June 2012

 (A)

(B)

(C)

▲▲ Figure 5. Examples of data visualization output included in FuncLab. (A) Record section of station XC-OR087 table (data from Eagar
et al. 2011), including (from left to right) the vertical, radial, and transverse time series, followed by the radial and transverse receiver
functions. (B) Stacked receiver function amplitudes in ray parameter bins. Blue denotes positive amplitudes and red denotes negative
amplitudes. Left panels are the radial receiver functions. Right panels are the transverse receiver functions. Top panels show histogram
of receiver functions per bin. (C) Stacked receiver function amplitudes in backazimuth bins. Color and histograms are the same as in (B).

Seismological Research Letters  Volume 83, Number 3  May/June 2012  601

that may indicate complexities such as anisotropy or dipping seismic stations. The increase in the amount of data available to
layers. The drop-down menu in the main FuncLab GUI then the seismological community has presented new challenges in
allows the user to view records of CCP bins in the same man- dealing with large datasets. Here we present a new GUI-based
ner as receiver function records. MATLAB toolbox for organizing and analyzing receiver func-
tion data. The main FuncLab package provides a management
Gaussian-Weighted Common Conversion Point Stacking system around which analysis tools may be written as add-on
Also included is an enhancement to the CCP add-on that uti- modules. We provide three add-on modules to enable com-
lizes a Gaussian weighting scheme as developed and described mon analysis techniques, including common conversion point
in Eagar et al. (2011). Gaussian-weighted common conversion stacking, Gaussian-weighted common conversion point stack-
point (GCCP) stacking modifies basic CCP stacking by weight- ing, and H-κ stacking. A conversion script is included to enable
ing the receiver functions within the stacks by the distance of users to request receiver functions directly from the IRIS Data
the piercing point from the image point, which is the same as Management Center. This user-friendly environment will
the center point of the CCP bin. The weight is determined from allow a range of researchers, from students new to seismic data
a 2-D Gaussian function whose one standard deviation width is analysis to seasoned scientists, to obtain useful results from
chosen appropriately by the array aperture and/or Fresnel zone receiver function analyses quickly and easily.
at the imaging target. The GCCP GUI is quite similar to the
CCP GUI with only a few minor differences, such as the choice ACCESSING FuncLab
of the Gaussian width and weighting function.
The FuncLab distribution is maintained at http://geophysics.
H-κ Stacking asu.edu/funclab/. The MATLAB codes are packaged in a tar file
H-κ stacking is a grid searching method that stacks receiver for download. We also provide a User Manual for quickly get-
function amplitudes for a given seismic station at predicted ting started using the codes. Links to other supporting sources
arrival times of crustal phases based on the Moho depth (H) of information and data can also be found at the Web site.
and Vp/Vs ratio (κ) of the crust (Zhu and Kanamori 2000). The
P-to-S conversion at the Moho (Ps) and reverberations within ACKNOWLEDGMENTS
the crust (PpPs and PpSs+PsPs) are used to reduce the inher-
ent tradeoffs between H and κ (Zandt and Ammon 1995). We thank Rick Aster and Gary Pavlis for organizing extremely
Additional details of this method can be found elsewhere (e.g., helpful MATLAB tutorials on receiver functions for the 2006
Zhu and Kanamori 2000; Eagar et al. 2011). IRIS/EarthScope Imaging Science Workshop at Washington
The H-κ Stacking Analysis GUI is similar to the other University, which first inspired KCE to deal with this prob-
add-ons described here and includes parameters that control lem in MATLAB. We would also like to thank Mike Thorne
the stacking on the left-hand side and processing functions for his original SACLAB codes (http://web.utah.edu/thorne/
on the right-hand side. Two options are provided for stacking software.html) for importing SAC files into MATLAB, many
and error analysis, selected from the “Error” drop-down menu. of which were modified for use with FuncLab. Discussions with
One option is a bootstrap method much like that described in John West provided valuable insight into proper documentation
the section on CCP stacking, which requires the user to set and programming style. Many thanks go to Philip Crotwell
the number of times to resample and the percentage of data (U. South Carolina), as well as Manochehr Bahavar and Chad
to resample. The other option is a linear stack, where error is Trabant (IRIS Data Management Center), for extensive help
determined from the contour of one standard deviation below in providing EarthScope Automated Receiver Survey (EARS)
the maximum value in the H-κ grid (Eaton et al. 2006). Several data products that are directly compatible with FuncLab. Beta
parameters that affect the resulting stack, such as the weights testers spent many valuable hours testing the code while in
of the Ps, PpPs, and PpSs+PsPs phases, and the assumed Vp, can development; we would like to acknowledge those who par-
also be defined. The time, depth, and Vp/Vs parameters only ticipated, especially Eun-Sun Chong, Jonathan Delph, Chelsea
affect the grid-search range and not necessarily the first-order Allison, Maggie Benoit, Kelsey Brunner, and Eli Raymond. This
result. As with CCP stacking, the user can specify the receiver research was supported by National Science Foundation awards
functions included in the stack by a range of backazimuths EAR-0548288 (MJF EarthScope CAREER grant) and EAR-
and ray parameters. The results of these procedures are saved as 0507248 (MJF Continental Dynamics High Lava Plains grant).
separate MAT-files for each stack and store the amplitudes in
the H-κ grid, the grid space axes, and the list of events that are REFERENCES
included in the stack. Several visualizations to aid the user in
the analysis are included in the package. Andrews, J., and A. Deuss (2008). Detailed nature of the 660 km
region of the mantle from global receiver function data. Journal of
Geophysical Research 113; doi: 10.1029/2007JB005111.
CONCLUSION Crotwell, H. P., and T. J. Owens (2005). Automated receiver function
processing. Seismological Research Letters 76, 702–708.
Receiver functions have become a commonly used imaging Dueker, K. G., and A. F. Sheehan (1997). Mantle discontinuity struc-
technique for studying sharp seismic velocity contrasts beneath ture from midpoint stacks of converted P to S waves across the

602  Seismological Research Letters  Volume 83, Number 3  May/June 2012

 Yellowstone hotspot track. Journal of Geophysical Research 102, Roth, J. B., M. J. Fouch, D. E. James, and R. W. Carlson (2008).
8,313–8,327. Three-dimensional seismic velocity structure of the north-
Dziewonski, A. M., and D. L. Anderson (1981). Preliminary reference western United States. Geophysical Research Letters 35;
Earth model. Physics of the Earth and Planetary Interiors 25, 297– doi:10.1029/2008GL034669.
356. Sheehan, A. F., G. A. Abers, C. H. Jones, and A. L. Lerner-Lam (1995).
Goldstein, P., D. Dodge, M. Firpo, and L. Minner (2003). SAC2000: Crustal thickness variations across the Colorado Rocky Mountains
Signal processing and analysis tools for seismologists and engi- from teleseismic receiver functions. Journal of Geophysical Research
neers. In The IASPEI International Handbook of Earthquake and 100, 20,391–20,404.
Engineering Seismology, ed. W. H. K. Lee, H. Kanamori, P. C. Suckale, J., S. Rondenay, M. Sachpazi, M. Charalampakis, A. Hosa,
Jennings, and C. Kisslinger, 1,613–1,614. Boston and Amsterdam: and L. H. Royden (2009). High-resolution seismic imaging of
Academic Press. the western Hellenic subduction zone using teleseismic scat-
Eagar, K. C., M. J. Fouch, and D. E. James (2010). Receiver function tered waves. Geophysical Journal International 178, 775–791; doi:
imaging of upper mantle complexity beneath the Pacific Northwest, 10.1111/j.1365-1246X.2009.04170.x.
United States. Earth and Planetary Science Letters; doi:10.1016/j. Tauzin, B., E. Debayle, and G. Wittlinger (2008). The mantle transition
epsl .2010.1006.1015. zone as seen by global Pds phases: No clear evidence for a thin tran-
Eagar, K. C., M. J. Fouch, D. E. James, and R. W. Carlson (2011). Crustal sition zone beneath hotspots. Journal of Geophysical Research 113;
structure beneath the High Lava Plains of eastern Oregon and doi:10.1029/2007JB005364.
surrounding regions from receiver function analysis. Journal of Wilson, D., R. Aster, J. Ni, S. Grand, M. West, W. Gao, W. S. Baldridge,
Geophysical Research 116; doi:10.1029/2010JB007795. and S. Semken (2005). Imaging the seismic structure of the crust
Eaton, D. W., S. Dineva, and R. Mereu (2006). Crustal thickness and Vp/ and upper mantle beneath the Great Plains, Rio Grande Rift, and
Vs variations in the Grenville orogen (Ontario, Canada) from analy- Colorado Plateau using receiver functions. Journal of Geophysical
sis of teleseismic receiver functions. Tectonophysics 420, 223–238. Research 110; doi:10.1029/2004JB003492.
Efron, B., and R. Tibshirani (1986). Bootstrap methods for standard Zandt, G., and C. J. Ammon (1995). Continental crust composition
errors, confidence intervals, and other measures of statistical accu- constrained by measurements of crustal Poisson’s ratio. Nature 374,
racy. Statistical Science 1, 54–77. 152–154.
Efron, B., and R. Tibshirani (1991). Statistical data analysis in the com- Zandt, G., S. C. Myers, and T. C. Wallace (1995). Crust and mantle
puter age. Science 253, 390–395. structure across the Basin and Range–Colorado Plateau boundary
Kennett, B. L. N., and E. R. Engdahl (1991). Traveltimes for global at 37°N latitude and implications for Cenozoic extensional mecha-
earthquake location and phase identification. Geophysical Journal nism. Journal of Geophysical Research 100, 10,529–10,548.
International 105, 429–465. Zhu, L., and H. Kanamori (2000). Moho depth variation in south-
Kennett, B. L. N., E. R. Engdahl, and R. Buland (1995). Constraints ern California from teleseismic receiver functions. Journal of
on seismic velocities in the Earth from travel times. Geophysical Geophysical Research 105, 2,969–2,980.
Journal International 122, 108–124.
Lawrence, J. F., and P. M. Shearer (2006). A global study of transition Department of Terrestrial Magnetism
zone thickness using receiver functions. Journal of Geophysical
Research 111; doi: 10.1029/2005JB003973. Carnegie Institution of Washington
Ligorria, J. P., and C. J. Ammon (1999). Iterative deconvolution and 5241 Broad Branch Road, NW
receiver-function estimation. Bulletin of the Seismological Society of Washington, DC 20015 U.S.A.
America 89, 1,395–1,400. fouch@dtm.ciw.edu
Owens, T. J., H. P. Crotwell, C. Groves, and P. Oliver-Paul (2004). (M. F.)
SOD: Standing Order for Data. Seismological Research Letters 75,
515–520.
Kevin C. Eagar and Matthew J. Fouch

Seismological Research Letters  Volume 83, Number 3  May/June 2012  603

View publication stats