[cig-commits] r11511 - in seismo/3D/automeasure/latex: . figures/japan
alessia at geodynamics.org
alessia at geodynamics.org
Sun Mar 23 07:41:01 PDT 2008
Author: alessia
Date: 2008-03-23 07:41:01 -0700 (Sun, 23 Mar 2008)
New Revision: 11511
Added:
seismo/3D/automeasure/latex/figures/japan/T06_rs-crop.pdf
Modified:
seismo/3D/automeasure/latex/README
seismo/3D/automeasure/latex/figures_paper.tex
seismo/3D/automeasure/latex/flexwin_paper.pdf
seismo/3D/automeasure/latex/results.tex
Log:
Partial re-write and re-organization of the Japan and SCal results sections. Reduced the number of figures and re-ordered them. Added a temporary figure for Japan. Updated README file to reflect current TODO status.
Modified: seismo/3D/automeasure/latex/README
===================================================================
--- seismo/3D/automeasure/latex/README 2008-03-23 03:12:30 UTC (rev 11510)
+++ seismo/3D/automeasure/latex/README 2008-03-23 14:41:01 UTC (rev 11511)
@@ -2,11 +2,25 @@
TODOs:
-* AM:
+* Alessia:
+ finish introduction with description of what is to follow
+ add envelopes to Figure 1
- + read through Carl and Min's results sections closely.
+ + write first draft of discussion
+ + write first draft of conclusion
+ + write abstract
+
+* Min+Carl:
+ + please check the results section to make sure I have not changed your meaning too much with my edits and re-arrangement of figures
+ + please add your time-dependent functions to appendix A
+ + what would you like to see in the discussion section?
+* Min:
+ + you seem to be missing the start of the P-wave on the long-period records. Could you lower the w_E earlier for this period range?
+ + I have combined three of your figures into one (Figure 15) by some rather ugly pdf hacks. Could you please make a decent version of something like this figure?
+
+
+
+
-------------------
Carl Tape, 07-March-2008
Added: seismo/3D/automeasure/latex/figures/japan/T06_rs-crop.pdf
===================================================================
(Binary files differ)
Property changes on: seismo/3D/automeasure/latex/figures/japan/T06_rs-crop.pdf
___________________________________________________________________
Name: svn:mime-type
+ application/octet-stream
Modified: seismo/3D/automeasure/latex/figures_paper.tex
===================================================================
--- seismo/3D/automeasure/latex/figures_paper.tex 2008-03-23 03:12:30 UTC (rev 11510)
+++ seismo/3D/automeasure/latex/figures_paper.tex 2008-03-23 14:41:01 UTC (rev 11511)
@@ -306,9 +306,9 @@
\clearpage
\begin{figure}
%\center
-\includegraphics[width=6in]{figures/japan/200511211536A_T24_rs}
-\caption{\label{fg:200511211536A_T24_rs}
-Summary plots of windowing results for event 200511211536A in Table~\ref{tb:events}, for the period range 24~s to 120~s.
+\includegraphics[width=6in]{figures/japan/200511211536A_T06_rs}
+\caption{\label{fg:200511211536A_T06_rs}
+Summary plots of windowing results for event 200511211536A in Table~\ref{tb:events}, for the period range 6~s to 30~s.
(a)~Map showing paths to each station with at least one measurement window.
(b)-(d)~Histograms of number of windows as a function of normalised cross-correlation $CC$, time-lag $\tau$ and amplitude ratio $\Delta \ln A$.
(e)-(g)~Record sections of selected windows for the vertical, radial and transverse components.
@@ -318,37 +318,47 @@
\clearpage
\begin{figure}
%\center
-\includegraphics[width=6in]{figures/japan/200511211536A_T06_rs}
-\caption{\label{fg:200511211536A_T06_rs}
-Summary plots of windowing results for event 200511211536A in Table~\ref{tb:events}, for the period range 6~s to 30~s.
-Same as Figure~\ref{fg:200511211536A_T24_rs}, only the windowing code has been run using a different set of parameters (Table~\ref{tb:example_params}), so that primarily only the body-wave arrivals are selected.
+\includegraphics[width=6in]{figures/japan/200511211536A_T24_rs}
+\caption{\label{fg:200511211536A_T24_rs}
+Summary plots of windowing results for event 200511211536A in Table~\ref{tb:events}, for the period range 24~s to 120~s.
}
\end{figure}
\clearpage
\begin{figure}
%\center
-\includegraphics[width=6in]{figures/japan/051502B_T06_rs}
-\caption{\label{fg:051502B_T06_rs}
-Summary plots of windowing results for event 051502B in Table~\ref{tb:events}, for the period range 6~s to 30~s.
-Same as Figure~\ref{fg:200511211536A_T06_rs}.
+\includegraphics[width=6in]{figures/japan/T06_rs-crop}
+\caption{\label{fg:T06_rs}
+Summary statistics of windowing results for events 051502B, 200511211536A and 091502B in Table~\ref{tb:events}, for the period range 6~s to 30~s.
}
\end{figure}
-\clearpage
-\begin{figure}
-%\center
-\includegraphics[width=6in]{figures/japan/091502B_T06_rs}
-\caption{\label{fg:091502B_T06_rs}
-Summary plots of windowing results for event 091502B in Table~\ref{tb:events}, for the period range 6~s to 30~s.
-Same as Figure~\ref{fg:200511211536A_T06_rs}.
-}
-\end{figure}
+%
%\clearpage
%\begin{figure}
%%\center
+%\includegraphics[width=6in]{figures/japan/051502B_T06_rs}
+%\caption{\label{fg:051502B_T06_rs}
+%Summary plots of windowing results for event 051502B in Table~\ref{tb:events}, for the period range 6~s to 30~s.
+%Same as Figure~\ref{fg:200511211536A_T06_rs}.
+%}
+%\end{figure}
+%
+%\clearpage
+%\begin{figure}
+%%\center
%\includegraphics[width=6in]{figures/japan/091502B_T06_rs}
+%\caption{\label{fg:091502B_T06_rs}
+%Summary plots of windowing results for event 091502B in Table~\ref{tb:events}, for the period range 6~s to 30~s.
+%Same as Figure~\ref{fg:200511211536A_T06_rs}.
+%}
+%\end{figure}
+
+%\clearpage
+%\begin{figure}
+%%\center
+%\includegraphics[width=6in]{figures/japan/091502B_T06_rs}
%\caption{\label{fg:091502B_T06_rs}
%Summary plots of windowing results for event 051502B in Table~\ref{tb:events},
%for the period range 6~s to 30~s. Same as Figure~\ref{fg:200511211536A_T06_rs).
@@ -358,34 +368,9 @@
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% SOUTHERN CALIFORNIA
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-
\clearpage
\begin{figure}
%\center
-\includegraphics[width=7in]{figures/socal/9983429_T06_rs.pdf}
-\caption{\label{fg:socal_rs_T06}
-Summary plots of windowing results for event 9983429 in Table~\ref{tb:events}, for the period range 6~s to 40~s.
-(a)~Map showing paths to each station with at least one measurement window.
-(b)-(d)~Histograms of number of windows as a function of normalised cross-correlation $CC$, time-lag $\tau$ and amplitude ratio $\Delta \ln A$.
-(e)-(g)~Record sections of selected windows for the vertical, radial and transverse components.
-The two branches observed on the vertical and radial components correspond to the body-wave arrivals and the Rayleigh wave arrivals.
-}
-\end{figure}
-
-\clearpage
-\begin{figure}
-%\center
-\includegraphics[width=7in]{figures/socal/9983429_T02_rs.pdf}
-\caption{\label{fg:socal_rs_T02}
-(THIS FIGURE COULD IN THEORY BE CUT OUT, IF SPACE IS SHORT.)
-Summary plots of windowing results for event 9983429 in Table~\ref{tb:events}, for the period range 2~s to 40~s.
-Same as Figure~\ref{fg:socal_rs_T06}, only the windowing code has been run using a different set of parameters (Table~\ref{tb:example_params}), so that primarily only the body-wave arrivals are selected.
-}
-\end{figure}
-
-\clearpage
-\begin{figure}
-%\center
\includegraphics[width=6in]{figures/socal/9818433_CLC_window.pdf}
\caption{\label{fg:socal_CLC}
Window selection results for event 9818433 from Table~\ref{tb:events} recorded at station CLC.
@@ -416,6 +401,31 @@
\clearpage
\begin{figure}
%\center
+\includegraphics[width=6in]{figures/socal/9983429_T06_rs.pdf}
+\caption{\label{fg:socal_rs_T06}
+Summary plots of windowing results for event 9983429 in Table~\ref{tb:events}, for the period range 6~s to 40~s.
+(a)~Map showing paths to each station with at least one measurement window.
+(b)-(d)~Histograms of number of windows as a function of normalised cross-correlation $CC$, time-lag $\tau$ and amplitude ratio $\Delta \ln A$.
+(e)-(g)~Record sections of selected windows for the vertical, radial and transverse components.
+The two branches observed on the vertical and radial components correspond to the body-wave arrivals and the Rayleigh wave arrivals.
+}
+\end{figure}
+
+%\clearpage
+%\begin{figure}
+%%\center
+%\includegraphics[width=7in]{figures/socal/9983429_T02_rs.pdf}
+%\caption{\label{fg:socal_rs_T02}
+%(THIS FIGURE COULD IN THEORY BE CUT OUT, IF SPACE IS SHORT.)
+%Summary plots of windowing results for event 9983429 in Table~\ref{tb:events}, for the period range 2~s to 40~s.
+%Same as Figure~\ref{fg:socal_rs_T06}, only the windowing code has been run using a different set of parameters (Table~\ref{tb:example_params}), so that primarily only the body-wave arrivals are selected.
+%}
+%\end{figure}
+
+
+\clearpage
+\begin{figure}
+%\center
\includegraphics[width=7in]{figures/socal/9818433_T06_CLC_adj.pdf}
\caption{\label{fg:socal_adj}
Adjoint sources constructed based on the windows picked in Figure~\ref{fg:socal_CLC}d, with the specification of a cross-correlation traveltime measurement. The adjoint sources for this measurement are simply a weighted version of the synthetic velocity traces. The number to the left of each subplot is the $\pm$ height of the $y$-axis. The cross-correlation measurements for traveltime ($\Delta T$) and amplitude ($\Delta \ln A$) are listed above each time window.
Modified: seismo/3D/automeasure/latex/flexwin_paper.pdf
===================================================================
(Binary files differ)
Modified: seismo/3D/automeasure/latex/results.tex
===================================================================
--- seismo/3D/automeasure/latex/results.tex 2008-03-23 03:12:30 UTC (rev 11510)
+++ seismo/3D/automeasure/latex/results.tex 2008-03-23 14:41:01 UTC (rev 11511)
@@ -4,7 +4,7 @@
applied to real data. These examples illustrate the robustness of the
algorithm, as well as its flexibility. We have applied the algorithm to three
tomographic scenarios, with very different geographical extents and distinct period ranges:
-a global tomography ($T = 40 - 200$~s),
+a global tomography ($T = 50 - 150$~s),
a regional tomography of the Japanese subduction zone, down to 700~km ($T = 6 - 120$~s), and
a regional tomography of southern California, down to 60~km ($T = 2 - 40$~s).
For each of these scenarios, we compare
@@ -147,200 +147,138 @@
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Japan scale}
\label{sec:japan}
-We also apply this windowing code to a set of local events within the
-depth range 0--600~km for a tomographic
-study of the Japanese subduction zone.
-The initial model is constructed using a Southeast Asia model
-\citep{LebedevNolet2003} as the background model, with
+Our second scenario is a regional-scale tomographic study of the Japanese subduction zone, using a set of local events within the depth range 0--600~km.
+The lateral dimensions of the domain are
+44$^\circ$(EW)$\times$33$^\circ$(NS) (108--152$^\circ$E
+and 18--51$^\circ$N).
+The initial model is constructed using the Southeast Asia model of
+\cite{LebedevNolet2003} as the background model, with
$P$-wave velocity anomalies added from
a high-resolution Japan $P$-wave model \citep{ZhaoEtal1994}
and $S$-wave velocity anomalies
scaled to $P$ by a factor of 1.5 \citep{ChenEtal2007}.
-
-The lateral dimensions of the entire model domain are
-44$^\circ$(EW)$\times$33$^\circ$(NS) (108--152$^\circ$E
-and 18--51$^\circ$N). Two different crustal
-models are implemented in the mesh: inside the region of the
+Two different crustal
+models are implemented in the spectral-element mesh: inside the region of the
high-resolution model (32--45$^\circ$N, 130--145$^\circ$E
-and down to 500~km), the crust model is derived from
-the arrival time data from local shallow earthquakes \citep{ZhaoEtal1992};
-outside that region, the crustal model is CRUST2.0 \citep{BassinEtal2000}.
+and down to 500~km), the crustal model is derived from
+the arrival times of local shallow earthquakes \citep{ZhaoEtal1992};
+outside this region, the crustal model is CRUST2.0 \citep{BassinEtal2000}.
-We collected more than $200$ events with $M_w$ from 4.5 to 8 that
-occurred between 2000 and 2006. The source location and focal
-mechanism are the Harvard centroid-moment tensor (CMT) solutions.
-There are total 818 stations from three different networks
-(GSN, F-net and Hi-net): 119 stations (GSN and F-net)
+We collected data for more than 200 events with $M_w$ 4.5--8 that
+occurred between 2000 and 2006. The source locations and focal
+mechanisms are the Harvard centroid-moment tensor (CMT) solutions.
+We used a total of 818 stations from three different networks
+(GSN, F-net and Hi-net): the 119 stations of GSN and F-net
provide broadband records,
-whereas 699 Hi-net provides only high-frequency records.
-
+whereas the 699 Hi-net stations provide only high-frequency records.
We use the one-chunk version of spectral-element code to calculate
-the synthetic seismograms accurate at periods of $\sim$6~s and
-longer \citep{ChenEtal2007}. Data and synthetics are processed in
-two period ranges: 6--30~s for all the records and 24--120~s
-for the broadband records.
+synthetic seismograms accurate at periods of $\sim$6~s and
+longer \citep{ChenEtal2007}, and present results for two period ranges: 6--30~s, using all the records, and 24--120~s, using the broadband records only.
-Figure~\ref{fg:ERM_II_051502B}, \ref{fg:KIS_BO_091502B} and
-\ref{fg:SHR_BO_20051121536A} are the examples
-of applying the windowing code on the three-component
-seismograms of three events at
-very different depths (Table~\ref{tb:example_params}):
+Figures~\ref{fg:ERM_II_051502B}, \ref{fg:KIS_BO_091502B} and
+\ref{fg:SHR_BO_20051121536A} show windowing results for three events at
+different depths (Table~\ref{tb:events}):
051502B, 22.4~km deep, Taiwan; 091502B, 589.4~km deep,
Northeastern China; 200511211536A, 155~km deep, Kyushu, Japan.
-
-The windowing code uses different sets of parameters for
-seismograms in $T = 6 - 30$~s and $T = 24 - 120$~s
-(Table~\ref{tb:example_params}). In the period
-range of 24--120~s, the water level is raised after the
-surface-wave arrivals to exclude the later arrivals which are
-not sensitive to the upper mantle structure. In the period
-range of 6--30~s, the water level is raised after the
+We have tuned the windowing algorithm using different sets of parameters for
+the two period ranges
+(see Table~\ref{tb:example_params}). In the period
+range 24--120~s, the water level is raised after the
+surface-wave arrivals to exclude the later arrivals that are
+not sensitive to upper mantle structure. In the period
+range 6--30~s, the water level is raised after the
$S$-wave arrivals to exclude the surface waves,
-as the current crustal model is not sufficient
-to predict the short-period surface
-waves and the corresponding STA/LTA is also very low.
+as the current crustal model is not detailed enough
+to predict the short-period surface waves.
-Figure~\ref{fg:ERM_II_051502B} shows that, for the shallow event beneath Taiwan
-(051502B) recorded by station ERM, the overall fits of data and synthetics
-are not so good in short-period range (6--30~s)
-(Figure~\ref{fg:ERM_II_051502B}b), but are much
-better at long-period range (24--120s) (Figure~\ref{fg:ERM_II_051502B}c).
-For the seismograms in $T = 6 - 30$~s, the synthetics captures only the major $P$-wave
-arrival on the vertical component (Figure~\ref{fg:ERM_II_051502B}b). While
-for the seismograms in $T = 24 - 120$~s, the windowing code picks not only the
-long-period $P$- and $S$-wave arrivals on three components, but also the
-Rayleigh-wave arrival on vertical and radial component and Love-wave
-arrival on transverse component (Figure~\ref{fg:ERM_II_051502B}c). As the depth sensitivity of the surface waves
-are frequency dependent, the good fits of surface waves
-at long periods (24--120~s) indicate structure deeper than MOHO
-is sufficiently good in the starting model for tomography.
+For the shallow event beneath Taiwan
+recorded at station ERM (Figure~\ref{fg:ERM_II_051502B}), the synthetics are a poorer fit to the data in the short-period range (6--30~s), than in the long-period range (24--120s).
+For $T = 6 - 30$~s, the synthetics capture only the major $P$-wave
+arrival on the vertical component (Figure~\ref{fg:ERM_II_051502B}b), which is also the only arrival picked by the windowing algorithm.
+For $T = 24 - 120$~s, the observed and synthetic seismograms are similar in shape, and the algorithm picks not only the long-period $P$- and $S$-wave arrivals, but also the
+Rayleigh-wave on the vertical and radial components and the Love-wave
+on the transverse component (Figure~\ref{fg:ERM_II_051502B}c). As the depth sensitivity of surface waves
+is frequency dependent, the good fits at long periods (24--120~s)
+indicate that the starting model structure below the Moho
+is adequate for tomography.
-Figure~\ref{fg:KIS_BO_091502B} is an example for a deep event (091502B)
-beneath Northeastern China. Compared to the shallow event
-(Figure~\ref{fg:ERM_II_051502B}), the seismograms of this event
-very simple with only two major body-wave arrivals ($P$ and $S$).
-$P$- or $S$-wave windows are picked based on the local level of STA/LTA curve
-and the similarity between the data and synthetics. For example,
-the windowing code did not pick the short-period $S$ arrival
-on the vertical component (Figure~\ref{fg:KIS_BO_091502B}b), as
+Figure~\ref{fg:KIS_BO_091502B} shows an example of window picks for a deep event
+beneath Northeastern China (091502B) recorded at station KIS. Compared to the shallow event above,
+the seismograms from this event are
+very simple, and contain only two major body-wave arrivals ($P$ and $S$). The windowing algorithm's similarity criterion clearly comes into play here, causing it
+not to pick the short-period $S$ arrival
+on the vertical component (Figure~\ref{fg:KIS_BO_091502B}b) as
the distorted $S$-wave waveform of the data is quite
-different from the synthetics shaped like a simple Gaussian.
-On the other hand, the long-period $S$-wave arrival
-(Figure~\ref{fg:KIS_BO_091502B}c) on the
+different from the Gaussian shaped synthetics.
+The long-period $S$-wave arrival on the
same component is selected due to higher data-synthetic
-waveform similarity.
+waveform similarity (Figure~\ref{fg:KIS_BO_091502B}c).
-The intermediate-depth event (20051121536A) has more phases
-showing up on the seismograms recorded by station SHR
-(Figure~\ref{fg:SHR_BO_20051121536A}).
-For example, on the vertical component of the short-period
- seismogram (Figure~\ref{fg:SHR_BO_20051121536A}b), $P$ wave
-arrives around 230~s, immediately followed by $pP$ and $sPn$,
-and $S$ wave comes at around 420~s followed by $sS$ around 470~s and
-$PcP$ around 500~s. In the period range of 6--30~s, the windowing
-code chooses the windows for the $P$, $S$ and $sS$ arrivals
-separately on the vertical
-component and $P$ arrival on the radial component.
-In the period range of 24--120~s, the windows of
-two wavepackets are picked in instead on the vertical
-and radial component, $P+pP+sPn$ and $S+sS+PcP$;
-one wavepacket window of $S+sS$ is selected on
-the transverse component.
-Notice that the surface-wave signals of this
-intermediate-depth event are not as obvious as of the shallow event.
+The records of the intermediate-depth event (20051121536A) recorded by station SHR (Figure~\ref{fg:SHR_BO_20051121536A}) contain more seismic phases
+than the previous two examples.
+On the vertical component of the short-period
+ seismogram (Figure~\ref{fg:SHR_BO_20051121536A}b), the $P$-wave
+arrives at $\sim$230~s, immediately followed by $pP$ and $sPn$,
+and the $S$-wave arrives at $\sim$420~s, followed by $sS$ and
+$PcP$. The windowing algorithm selects separate windows for the $P$, $S$ and $sS$ arrivals
+on the vertical component, and selects only the $P$ arrival on the radial component.
+In the period range 24--120~s, the $P$ and $S$ waves merge with the arrivals that follow them, causing the windowing algorithm to select wavepackets instead of single phases: $P+pP+sPn$ and $S+sS+PcP$ on the vertical and radial components, and $S+sS$ on
+the transverse component. Note that the surface-wave signals of this
+intermediate-depth event are not as clearly defined as those of the shallow event (Figure~\ref{fg:ERM_II_051502B}).
-Figure~\ref{fg:200511211536A_T24_rs} and
-Figure~\ref{fg:200511211536A_T06_rs} show summary plots of the
-window picks for event 200511211536A in two different
-period ranges ($T = 6 - 30$~s and $T = 24 - 120$~s),
-each has a different set of parameters
-(Table~\ref{tb:example_params}). The window record sections
-in both figures highlight the primary phases
-that are picked by the algorithm.
-In the period range of 24--120~s, on the vertical and radial
-component (Figure~\ref{fg:200511211536A_T24_rs}ef), the
-first branch corresponds the wavepacket of $P+pP+sPn$ and the
-second branch corresponds the wavepacket of $S+sS+PcP$;
-the windows on the transverse component capture the
-arrivals of $S+sS$ (Figure~\ref{fg:200511211536A_T24_rs}g).
-The examples of major wavepackets are also shown
-in Figure~\ref{fg:SHR_BO_20051121536A}.
-In the period range of 6--30~s, the windows in the record sections,
-are much narrower compared to the ones in Figure~\ref{fg:200511211536A_T24_rs}. On all three components, the two main branches
-correspond to $P$ and $S$ arrivals. Even on the
-transverse component (Figure~\ref{fg:200511211536A_T06_rs}g),
-the algorithm picks up $P$ arrivals sparsely.
-Of course the number and width of the windows
-for each trace varies with distances.
-We notice that on vertical and radial component
-window record section (Figure~\ref{fg:200511211536A_T06_rs}ef),
-beyond the distance of 14\deg: after the $P$-arrival branch,
-there are two small branches corresponding to $pP$ and $sPn$;
-after the $S$-arrival branch, there is another branch
-corresponding to $sS$. More percentage of measurement windows are
-made in $T = 24 - 120$~s with higher degrees of waveform similarity
-than in $T = 6 - 30$~s, as shown in
-Figure~\ref{fg:200511211536A_T24_rs}b that
-more than two thirds of the total windows have $CC>0.9$, while the
-quite opposite distribution shows in
-Figure~\ref{fg:200511211536A_T06_rs}b.
-$\Delta\tau$ values peak between -5~s to 0~s in both period ranges
-(Figure~\ref{fg:200511211536A_T24_rs}c and
-Figure~\ref{fg:200511211536A_T06_rs}c), indicating that
+Summary plots of window picks for event 200511211536A in the two
+period ranges $T = 6 - 30$~s and $T = 24 - 120$~s are shown in Figures
+Figure~\ref{fg:200511211536A_T06_rs} and~\ref{fg:200511211536A_T24_rs} respectively.
+On the short-period window record sections, the windows picked by the algorithm form two main branches
+that correspond to $P$ and $S$ arrivals. Sparse $P$ arrivals are visible even on the transverse component (Figure~\ref{fg:200511211536A_T06_rs}g).
+The number and width of windows for each trace varies with epicentral distance.
+On the vertical and radial component
+window record sections (Figure~\ref{fg:200511211536A_T06_rs}ef),
+beyond the distance of 14\deg\ and after the $P$-arrival branch
+there are two small branches corresponding to $pP$ and $sPn$, while
+after the $S$-arrival branch there is another branch
+corresponding to $sS$. The summary plot for the 24--120~s period range shows a single branch of windows on the vertical component, that splits up into separate $P$- and $S$-wavepackets at distances greater than 15\deg. The same split is visibile on the radial component, but occurs earlier (around 10\deg), while the transverse component windows form a single branch containing the merged $S+sS$ arrivals.
+
+Comparison of the histograms in Figures~\ref{fg:200511211536A_T06_rs} and~\ref{fg:200511211536A_T24_rs} shows that windows selected on the $T = 24 - 120$~s seismograms tend to have higher degrees of waveform similarity than those selected on the $T = 6 - 30$~s records.
+$\Delta\tau$ values peak between -5~s to 0~s in both period ranges, indicating that
the synthetics are slower than the observed records.
-The particularly large peak at -2~s in
-Figure~\ref{fg:200511211536A_T06_rs}c
-of $\Delta\tau$ distribution possibly corresponds
-to the addition of the large number of Hi-net stations for
-short-period range records, and these stations happened
-to be in the same NE direction with respect to this event.
-The $\Delta\ln A$ distribution peaks at $\Delta\ln A\simeq0$
+The particularly large peak at -2~s in the $\Delta\tau$ distribution
+of Figure~\ref{fg:200511211536A_T06_rs}c is probably due to
+the large number of Hi-net recordings that make up the
+short-period range records.
+The $\Delta\ln A$ distribution peaks at $\Delta\ln A\simeq0$
for $T = 24 - 120$~s (Figure~\ref{fg:200511211536A_T24_rs}d),
-which indicates the amplitude of the synthetics fits the data
-pretty well at long periods. The peak at $\Delta\ln A\simeq-0.3$
-in Figure~\ref{fg:200511211536A_T06_rs}d indicates that on average,
+indicating the amplitude of the synthetics matches the amplitude of the data
+at long periods. The peak at $\Delta\ln A\simeq-0.3$
+in Figure~\ref{fg:200511211536A_T06_rs}d indicates that, on average,
the synthetics overestimate the amplitude of the
-observed waveforms by 30\%. We cannot know at this stage
-if this anomaly in short-period range is due to an
-overestimation of seismic moment of the events,
-or to an underestimation of the attenuation.
+observed waveforms by 30\% at short-periods. We cannot know at this stage
+if this anomaly is due to an overestimation short period energy in the source spectra of the events,
+or to an underestimation of the seismic attenuation.
-
-Figure~\ref{fg:051502B_T06_rs} and Figure~\ref{fg:091502B_T06_rs}
+Figure~\ref{fg:T06_rs}
show summary plots of the
-window picks for the shallow event (051502B) and deep
-event (091502B) in $T = 6 - 30$~s. Notice the very large numbers of
-measurement windows picked due to the addition of more than 600
-Hi-net stations:
-(Figure~\ref{fg:200511211536A_T06_rs}, Figure~\ref{fg:051502B_T06_rs}
-and Figure~\ref{fg:091502B_T06_rs}): 1356 measurement windows
-for event 200511211536A, 1243 for event 051502B and 1880 for
+window picks statistics for the shallow event (051502B), intermediate (200511211536A) and deep
+events (091502B) for the period range $T = 6 - 30$~s. Notice the very large numbers of
+measurement windows picked due to the over 600
+Hi-net stations: 1243 windows for event 051502B, 1356 windows
+for event 200511211536A, and 1880 windows for
event 091502B.
-Comparing all three events at different depth, we can see that
-the degrees of similarity get better with increasing of event
-depth (Figure~\ref{fg:051502B_T06_rs}b,
-Figure~\ref{fg:200511211536A_T06_rs}b and
-Figure~\ref{fg:091502B_T06_rs}b). This implies better estimation
-of mantle structure than the crustal structure
+Comparing the statistics for these three events, we see that
+the degree of similarity $CC$ improves with increasing event
+depth, implies the estimation
+of mantle structure is better than the estimation of crustal structure
in the initial model.
The $\Delta\ln A$ distributions of these three events
-have similar shapes (Figure~\ref{fg:051502B_T06_rs}d,
-Figure~\ref{fg:200511211536A_T06_rs}d and
-Figure~\ref{fg:091502B_T06_rs}d), with peaks in the
-range of -0.5 - -0.3.
-However, the $\Delta\tau$ distributions have very different features
-in Figure~\ref{fg:051502B_T06_rs}c,
-Figure~\ref{fg:200511211536A_T06_rs}c and
-Figure~\ref{fg:091502B_T06_rs}c:
-the shallow event (051502B) has peak $\Delta\tau$ distribution
+have similar shapes, with peaks in the
+range of -0.5 -- -0.3.
+However, the $\Delta\tau$ distributions have very different features:
+the shallow event (051502B) has a large peak
at -10~s and another smaller peak at 8~s; the intermediate-depth
-event (200511211536A) has very peaked $\Delta\tau$ distribution
-with a peak at -2s; the deep event(091502B) has very
-distributed $\Delta\tau$ in the range of -2---10~s.
-
-Again, like what we discussed in the global examples,
-possible explanations for these large average
+event (200511211536A) has very peak at -2~s; the deep event(091502B) has a more
+distributed $\Delta\tau$ in the range -2 to -10~s.
+Possible explanations for these large average
time lags include an origin time error,
and/or an overestimation of the seismic
velocity at the source location.
@@ -349,37 +287,23 @@
\subsection{Southern California scale}
\label{sec:socal}
-We next apply the windowing code to a set of 150 events within southern California.
-We compute synthetic seismograms using the spectral-element method for a regional 3D crustal and upper mantle model for southern California \citep{KomatitschEtal2004}. This model contains three discontinuities:
-%
-\begin{enumerate}
-\item the surface, where the topography is meshed
-\item the basement layer, separating the sedimentary basins from the bedrock
-\item the Moho, separating the crust from the upper mantle
-\end{enumerate}
-%
-The basement surface is essential for simulating the resonance of seismic waves within sedimentary basins, such as the Ventura basin, the L.A. basin, and the Salton trough \citep{KomatitschEtal2004,LovelyEtal2006}. The smooth 3D background velocity model used in \citet{KomatitschEtal2004} was \citet{Hauksson2000}; we use an updated version provided by \citet{LinEtal2007}. The physical domain of the model is approximately 600~km by 500~km at the surface, and extends to a depth of 60~km. Our simulations of seismic waves are numerically accurate down to a period of 1.5~s.
+Our last scenario is a local tomography of Southern California. We apply the windowing algorithm to a set of 150 events within southern California, for which we have computed synthetic seismograms using the spectral-element method and a regional 3D crustal and upper mantle model \citep{KomatitschEtal2004}. This model contains three discontinuities: the surface topography (included in the mesh), the basement layer that separates the sedimentary basins from the bedrock, and the Moho. The basement surface is essential for simulating the resonance of seismic waves within sedimentary basins, such as the Ventura basin, the L.A. basin, and the Salton trough \citep{KomatitschEtal2004,LovelyEtal2006}. The smooth 3D background velocity model used in \citet{KomatitschEtal2004} was \citet{Hauksson2000}; we use an updated version provided by \citet{LinEtal2007}. The physical domain of the model is approximately 600~km by 500~km at the surface, and extends to a depth of 60~km. Our simulations of seismic waves are numerically accurate down to a period of 1.5~s.
The 150 events, with $M_w$ magnitudes between 3.5 and 5.5, were recorded between 1999 and 2007 and constitute a subset of the focal mechanisms presented by \citet{ClintonEtal2006}. The locations and origin times are primarily from \citet{LinEtal2008}, supplemented by the catalog of \citet{ThurberEtal2006} for events near Parkfield and the catalog of \citet{McLarenEtal2008} for events near San Simeon. Each of these regions had $M_w > 6$ earthquakes within our time period of interest: 2004.09.28~$M_w$~6.0 near Parkfield and 2003.12.22~$M_w$~6.5 near San Simeon. Aftershocks of both events are included in the dataset.
-We test the windowing code using two period ranges: 6--40~s and 2--40~s. The parameters we use for the windowing code are listed in Table~\ref{tb:example_params}. Figures~\ref{fg:socal_rs_T06}--\ref{fg:socal_FMP} show examples of the output from the windowing code for the two events listed in Table~\ref{tb:events}: 2002.09.03~$M_w$~4.3 and 2004.02.14~$M_w$~4.6.
+We test the windowing code using two period ranges: 6--40~s and 2--40~s. The parameters we use for the windowing code are listed in Table~\ref{tb:example_params}. Figures~\ref{fg:socal_CLC} and~\ref{fg:socal_FMP} show examples of the output from the windowing algorithm for event 9818433 listed in Table~\ref{tb:events} recorded at two different stations, while Figure~\ref{fg:socal_rs_T06} shows a summary plot for event 9983429 in the 6--40~s period range.
-Figures~\ref{fg:socal_rs_T06} and \ref{fg:socal_rs_T02} show summary plots of the window picks for one event for the two different period ranges, each which has a different set of parameters (Table~\ref{tb:example_params}). The window record section in Figure~\ref{fg:socal_rs_T06} illuminates the primary phases that are picked by the algorithm. On the vertical and radial components, the first branch corresponds to body-wave arrivals, and the second branch corresponds to the Rayleigh-wave arrival. Windows on the transverse component capture the Love wave arrival.
+The windowing algorithm tends to pick five windows on each set of three component longer-period seismograms (Figure~\ref{fg:socal_CLC} and~\ref{fg:socal_rs_T06} are representative examples): on the vertical and radial components the first window corresponds to the body-wave arrival and the second to the Rayleigh wave, while windows on the transverse component capture the Love wave. The shorter-period synthetic seismograms do not agree well with the observed seismograms, especially in the later part of the signal, leading to fewer picked windows. In Figure~\ref{fg:socal_CLC}e, only two windows are selected by the algorithm: a P arrival recorded on the radial component, and the combined S and Love-wave arrival on the transverse component. The P-wave arrival on the vertical component is rejected because the cross-correlation value within the time window did not exceed the specified minimum value of 0.85 (Table~\ref{tb:example_params}).
-Figure~\ref{fg:socal_rs_T02} shows the window picks for the same event as in Figure~\ref{fg:socal_rs_T06}, only for the period range 2--40~s instead of 6--40~s. Because the majority of the synthetic seismograms do not agree well with the observed seismograms, we raise the water level associated with the STA:LTA curve (dashed line in Figure~\ref{fg:socal_CLC}e) so that primarily only body-wave arrivals are picked. The result is that prominent Rayleigh-wave and Love-wave branches in Figure~\ref{fg:socal_rs_T06} are not observed in Figure~\ref{fg:socal_rs_T02}.
+Figure~\ref{fg:socal_FMP} shows results for the same event as Figure~\ref{fg:socal_CLC}, but for a different station, FMP, situated 52~km from the event and within the Los Angeles basin. Comparison of the two figures highlights the characteristic resonance caused by the thick sediments within the basin. This resonance is beautifully captured by the transverse component synthetics (Figure~\ref{fg:socal_FMP}d, record T), thanks to the inclusion of the basement layer in the crustal model\citep{KomatitschEtal2004}. In order to pick such long time windows with substantial frequency-dependent measurement differences, we are forced to lower the minimum cross-correlation value for the entire dataset (0.74 in Table~\ref{tb:example_params}) and increase $c_{4b}$ to capture the slow decay in the STA:LTA curves (Figure~\ref{fg:socal_FMP}d, record T). It is striking that although these arrivals look nothing like the energy packets typical for the global case, the windowing algorithm is still able to determine the proper start and end times for the windows. In Figure~\ref{fg:socal_FMP}e the windowing algorithm selects three short-period body-wave time windows with superb agreement between data and synthetics.
+%Finally, in Figure~\ref{fg:socal_adj} we present one example of a set of adjoint sources for one station for one event. For a cross-correlation traveltime measurement, the adjoint source is simply a weighted version of the synthetic velocity. The synthetic seismograms in the left column of Figure~\ref{fg:socal_adj} are displacement records. The weight for each time window in an adjoint source contains three factors:
+%%
+%\begin{enumerate}
+%\item the cross-correlation traveltime measurement, $\Delta T$;
+%\item the estimated uncertainty associated with the measurement, $\sigma_T$;
+%\item a term representing the size of the synthetic pulse \citep[][Eq.~42]{TrompEtal2005}.
+%\end{enumerate}
+%%
+%The contributions act in a manner such that a large measurement with a low uncertainty estimate for a small pulse will have the largest weight for the adjoint source.
-Figure~\ref{fg:socal_CLC} shows an example of the windows picked for one station for one event for the two period ranges of interest. In (d) the STA:LTA curves capture five primary phases that are revealed in Figure~\ref{fg:socal_rs_T06}. In (e) there are only two windows picked for measurement: a P arrival recorded on the radial component, as well as the combined S and Love-wave arrival on the transverse component. It turns out that the P-wave arrival on the vertical component is rejected because the cross-correlation value within the time window did not exceed the specified minimum value of 0.85 (Table~\ref{tb:example_params}).
-
-Figure~\ref{fg:socal_FMP} is the same as Figure~\ref{fg:socal_CLC}, only for a different station. The station, FMP, is situated 52~km from the event and within the Los Angeles basin. The characteristic resonance of the basin is beautifully captured on the transverse component (Figure~\ref{fg:socal_FMP}d, record T), thanks to the implementation of the basement layer \citep{KomatitschEtal2004}. In order to pick such long time windows with substantial frequency-dependent measurement differences, we are force to lower the minimum cross-correlation value for the entire dataset (0.74 in Table~\ref{tb:example_params}) and increase $c_{4b}$ to capture the slow decay in the STA:LTA curves (Figure~\ref{fg:socal_FMP}d, record T). What is striking is that these arrivals look nothing like energy ``packets'', such as in the global case, but the windowing code is able to suitably determine the proper start and end times for the measurement windows. In (e) the windowing code selects three time windows with superb agreement between data and synthetics.
-
-Finally, in Figure~\ref{fg:socal_adj} we present one example of a set of adjoint sources for one station for one event. For a cross-correlation traveltime measurement, the adjoint source is simply a weighted version of the synthetic velocity. The synthetic seismograms in the left column of Figure~\ref{fg:socal_adj} are displacement records. The weight for each time window in an adjoint source contains three factors:
-%
-\begin{enumerate}
-\item the cross-correlation traveltime measurement, $\Delta T$;
-\item the estimated uncertainty associated with the measurement, $\sigma_T$;
-\item a term representing the size of the synthetic pulse \citep[][Eq.~42]{TrompEtal2005}.
-\end{enumerate}
-%
-The contributions act in a manner such that a large measurement with a low uncertainty estimate for a small pulse will have the largest weight for the adjoint source.
-
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
More information about the cig-commits
mailing list