Sunteți pe pagina 1din 7

SEIS_PICK: AN INTERACTIVE SEISMIC PICKING ENVIRONMENT FOR

MATLAB

JAMES VERDON, UNIVERSITY OF BRISTOL

© 2012. All rights reserved

This work was produced under the sponsorship of BUMPS (Bristol University
Microseismicity ProjectS): www1.gly.bris.ac.uk/BUMPS/

INTRODUCTION
SEIS_PICK provides an interactive environment for picking seismic waveforms. It
has been developed with the aim of processing 3-component downhole
microseismic data, although is suitable for all 3-component seismic data.
SEIS_PICK displays seismic traces, allowing the user to pick P-wave and S-wave
phases. When phases have been picked, the user can enter spectral analysis to
view the frequency content of each phase, as well as pre-event noise. The user
can also enter hodogram analysis to view the polarization of phases. This is
particularly useful to check that picked P-wave and S-wave phases are polarized
orthogonally, as well as to compute the back-azimuth of the detected event
(using P-wave polarization). SEIS_PICK also provides a number of other signal-
processing facilities, including a band-pass filter, or predictive filter that uses
pre-event noise to create a notch filter. The principle motivation in the
development of SEIS_PICK was a dislike of SAC for the processing of microseismic
data. SAC was designed principally for global seismic analysis. While it has been
successfully used for microseismic event processing, the particular challenges
posed by microseismic data mean that it is not ideal. SEIS_PICK has been designed
SEIS_PICK MANUAL

as a replacement for SAC in microseismic processing, to provide an environment


more in line with typical industry standard microseismic processing workflows.

USING SEIS_PICK
INPUT:
SEIS_PICK is called as a Matlab function. The input to SEIS_PICK can take one of two
forms, either:
[pickmatrix] = seis_pick(traces,dt,nc)
or:
[pickmatrix] = seis_pick(string_argument)
depending on whether the user wishes to read data directly from SAC format
files, or to input Matlab workspace variables.
When the user is inputting variables from the Matlab workspace, traces is a
matrix containing the seismic data. Each column represents one seismic trace. dt
is the sampling rate of the traces. nc is the number of components on each
geophone. The total number of traces must be divisible by nc (i.e., for 8 stations
with three components each, traces must contain 24 columns – 8 geophones x
3 traces each).
Hodogram analysis can be performed on 3 component data. For hodogram
analysis, it is assumed that, in the traces matrix, the order of the traces is ‘East
– North – Vertical’.
Alternatively, SAC format data files can be read. If SAC files are to be read, the
input argument should consist of a single string, which controls the SAC
filenames in the following manner. A number of arguments are defined within
the string, using the format tag=argument. The tags are as follows:
stem= name of the SAC filename stems
numbers= numbered suffixes defining the individual stations
Components= suffixes defining individual components for each station
ncomps= if components is not specified, ncomps specifies the number
of components per station
zeros= set to 1 if numbers below 10 are padded by zeros.
As an example, if the SAC files for a particular event have the naming convention:
event1.1.x, event1.1.y, event1.1.z – event1.n.x(y,z), then string_argument would
be:
string_argument=’stem=event1. numbers=1:n components=xyz’
If the data was 3-component and the naming convention was event1.01,–
event1.n, then string_arguments would be
string_argument=’stem=event1. numbers=1:n zeros=1
ncomps=3’

2/7
SEIS_PICK MANUAL

PICKING:
When SEIS_PICK is called, the picking window displays the seismic data (Figure 1).
As well as the three components for each geophone, the combined energy
envelope is also displayed. This is computed using a Hilbert transform. Where
picks have been made, they are also displayed as vertical dashed lines on the
relevant traces. Where hodogram analysis has been performed, the computed P-
wave azimuths are also displayed.
The user interacts with the picking window using the Matlab ginput function.
Placing the cursor over the relevant trace and pressing the appropriate button
instructs SEIS_PICK to perform the appropriate action. See Table 1 for a full list of
SEIS_PICK options.

Figure 1. Example SEIS_PICK picking window. The 3 components from each geophone are plotted
(red and green = X and Y components, blue = vertical components), as well as the combined energy
envelope (grey). Also plotted are the picks (black = P-wave start, blue = P-wave end, red = S-wave
start, green = S-wave end). Where hodogram analysis has been performed, the P-wave azimuths are
also displayed.

Button Action Notes


p Pick P-wave start
s Pick S-wave start
o Pick P-wave end
a Pick S-wave end
P Delete P-wave start pick
S Delete S-wave start pick
O Delete P-wave end pick
A Delete S-wave end pick
w Wipe picks on one trace
W Wipe picks on all traces

f Predictive filter The predictive filter inverts the frequency spectrum of


the pre-event noise to create a notch filter. To define
the pre-event noise, the predictive filter requires that
the P-wave start be picked. If it is not, no filtering will

3/7
SEIS_PICK MANUAL

occur.
B Bandpass filter Once bandpass has been selected, the user must enter
a length 2 vector defining the low and high pass
corner frequencies.
F Remove filtering When filtering, the unfiltered traces are stored in a
dummy trace. If selected, SEIS_PICK returns to the
original unfiltered trace. The filtered traces are not
saved.

H Enter hodogram analysis See ‘Hodogram Analysis’ section.


Q Enter spectral analysis See ‘Spectral Analysis’ section.
H Perform hodogram analysis Performs hodogram analysis on P-wave windows only
on all stations to compute event azimuth.

C Cut traces Effectively, this tool controls Matlab’s xlim property,


to view a subset of the full trace. This is only an xlim
operation, the trace itself is not cut. When using ‘c’, the
user must press ‘c’ once to define the start of the new
time window, and ‘c’ again to define the end of the
window.
Z Zoom to selected traces Effectively, this tool controls Matlab’s ylim property,
to view one (or more) of the geophones only. As per
the ‘cut’ option, press ‘z’ once to select the topmost
trace to view, and ‘z’ again to select the lowermost
trace to view.
C Undo cut to view full trace Resets Matlab’s xlim property to default settings.
Z Undo zoom to view all traces Resets Matlab’s ylim property to default settings.
V Flip trace order Flips the order in which traces are plotted
(essentially, this is akin to ‘axis ij’). Useful to plot
stations in depth order.

E Save plot of traces to .eps On selecting this option, the user must input the file-
name to which the plot of traces is saved.
L Load pre-existing pickmatrix If an ascii file of picks already exists, ‘L’ loads them.
See ‘Pickmatrix format’ section for details on the
format of a pick matrix file. On selecting this option,
the user is prompted for the file-name of the picks file.
M Save picks to ascii file Saves the picks to an ascii format pick matrix file. On
selecting this option, the user is prompted for the file-
name of the picks file.
m Save picks to SAC file Saves the picks into the SAC file headers (in headers
t0=P-wave start, t1=S-wave start, t2=P-wave end,
t3=S-wave end).
J Toggle help panel Removes/displays a helpful reminder of all SEIS_PICK
options.
X Exit SEIS_PICK
Table 1. SEIS_PICK options in the picking window.

4/7
SEIS_PICK MANUAL

HODOGRAM ANALYSIS:
Selecting ‘h’ on a trace in the picking window sends SEIS_PICK into the hodogram
analysis window (see Figure 2) for the selected trace. SEIS_PICK performs
principle component analysis on the P-wave and S-wave windows to determine
the particle motion. The P-wave particle motion is used to infer the azimuth of
the seismic event. When azimuths from multiple geophones have been
computed, SEIS_PICK uses the Matlab circular statistics package to compute the
average azimuth for the event, as well as 95% confidence limits. These values are
written at the top of the picking window. The S-wave polarization is also
measured, and the angle between P-wave and S-wave particle motion is
calculated. Ideally, they should be orthogonal, i.e., P-S angle = 90°. If there is
substantial deviation from 90°, use caution as the picked phase may in fact be a
converted or reflected wave. The hodogram analysis window replaces the traces
window, displaying the particle motion data, as well as listing the polarization
angles. If either phase has not been assigned both a start pick AND an end pick, it
will not be included in the hodogram analysis, as the phase window is not
defined. If neither phase has been picked, you’ll be left with a nice empty
window. To exit the hodogram analysis window and return to the traces
window, press either ‘x’ or ‘h’. To save the plot as a .eps file, press ‘s’. The file-
name for the plot will be generated automatically as `hodogram_n.eps’, where ‘n’
is the number of the geophone being analyzed.

Figure 2. Example hodogram analysis window. The P-wave (red) and S-wave (green) particle
motions are displayed (horizontal components, X vs vertical, Y vs vertical). Dashed lines show the
average particle motions as computed using principle component analysis. In the textbox
underneath the plots, the P-wave and S-wave polarization azimuths are listed, along with the angle
between P-wave and S-wave particle motions. Ideally, P-waves and S-waves should be orthogonal,
i.e., 90°.
To perform hodogram analysis on all traces simultaneously, pressing ‘H’ in the
picking window will cause SEIS_PICK to consecutively analyse P-wave particle
motion for each picked P-wave window. No hodogram plots are generated
during this analysis.

SPECTRAL ANALYSIS:
Selecting ‘q’ on a trace in the picking window sends SEIS_PICK into the spectral
analysis window (see Figure 3) for the selected trace. SEIS_PICK uses Matlab’s fft

5/7
SEIS_PICK MANUAL

function to compute the frequency content of the pre-event noise, the P-wave
window and the S-wave window. The spectral analysis window displays these
spectra separately for each component. The P-wave spectrum will be plotted in
red, the S-wave spectrum in green, and the pre-event noise in blue. If either
phase has not been assigned both a start pick AND an end pick, it will not be
included in the spectral analysis, as the phase window is not defined.
Additionally, if the P-wave start time has not been picked then the pre-event
noise is not defined, so is not included in the analysis. If neither phase has been
picked, you’ll be left with a nice empty window. To exit the spectral analysis
window, press ‘x’ or ‘q’.

Figure 3. Example spectral analysis window. The frequency spectrum for the P-wave window (red),
S-wave window (green) and pre-event noise (blue) are plotted separately for each component (X, Y
and vertical). In this example, the signal is primarily found on the X and Y components – the signal
does not exceed noise on the vertical component.
Pick matrix Format:
If ‘M’ is selected in the picking window, the picks are exported as a matrix to an
ascii file. The user will be prompted to provide a file-name for this file. The pick
matrix file is headerless, containing a n x 6 matrix (where n is the number of
geophones), and whose columns contain the following:

Column 1 Column 2 Column 3 Column 4 Column 5 Column 6

Geophone number P-wave start S-wave start P-wave end S-wave end Azimuth

If a pre-existing pick matrix file has been generated, it can be loaded into
SEIS_PICK by pressing ‘L’ in the picking window. The user will then be prompted
for the file-name of this input file, which must be a headerless ascii format
containing a pick matrix as defined above. When SEIS_PICK is terminated (by
pressing ‘X’ in the picking window), the pick matrix is also returned as the output
variable from the function.

ACKNOWLEDGEMENTS
SEIS_PICK was developed as part of the BUMPS project, an industry-sponsored
research consortium based at the University of Bristol. Andreas Wuestefeld is
thanked for his work in developing the predictive filter. James Wookey provided
the msac toolbox for reading and writing SAC format files. We also acknowledge

6/7
SEIS_PICK MANUAL

the authors of the Matlab Circular Statistics Toolbox, some functions of which are
used in computing average event P-wave azimuths.

7/7

S-ar putea să vă placă și