Geoscience Reference
In-Depth Information
(MSPT)), and display it in a variety of graphical
formats for visual analysis. The DSP portion of the code
was accomplished on a channel-by-channel basis and
included DC normalization (not with MSPT), window
function application for Fourier transform analysis,
Butterworth-based infinite impulse response (IIR) nar-
rowband reject filtering (to suppress 60 Hz and related
harmonics), and a high-order linear-phase finite impulse
response (FIR) 1Hz high-pass filter for low-frequency
trend removal process. MATLAB nonlinear curve fitting
tools were used to generate the fits to the thresholded
and counted data shown in the following text.
The post-data acquisition signal processing and
analysis sequence is as follows: (1) the data in the BDF
file is read into memory; (2) a single channel was selected
and used to perform spectral analysis of the data to deter-
mine what filters were needed to filter the power system
related and other undesirable sinusoidal spectral content
and low-frequency trends from the data; (3) the needed
filters were designed; (4) the data was filtered to remove
the undesirable sinusoidal spectral content; (5) the data
was then DC normalized; (6) the resulting filtered and
DC normalized data was spectroscopically assessed to
confirm filtering results, and the time series data was
plotted versus voltage; (7) the high-pass filter was applied
to the data to remove all low-frequency trends leaving
only the impulses in the data, and this filtered time series
data was plotted as a function of voltage; and (8) the data
was then passed to a thresholding process where the
signal voltage level was compared to a threshold voltage
level (positive and negative thresholds). Every time
the signal voltage passed a threshold value, a counter
incremented a threshold value bin. This process gener-
ated a list of counts versus threshold that was used to
determine the histogram trend through nonlinear
regression and then the result was plotted. This process
was applied to both drainage and imbibition data.
We sought to visually confirmHaines jumps during the
drainage process. To accomplish that, video recordings of
drainage-based processes show that individual pore
dewatering takes place in a discontinuous process called
Haines jumps (Lu et al., 1994). Figure 5.39 shows, for
example, Haines jumps during the drainage process for
the sand used in these experimental investigations. The
pictures are frame grabs from the 15 frame per second
digital video stream and show that there are places where
there are sudden jumps in the displacement of the menis-
cus during drainage, while there were other areas where
Frame 1
Frame 2
2:55:09
2:55:10
Frame 1
Frame 2
Time notation = m:ss:s/15
Frame rate = 15 fps
6:47:13
6:47:14
Figure 5.39
Haines jumps. Pictures showing instabilities in the
position of the meniscus (outlined by the plain lines) during
drainage. The two pictures show that the meniscus labeled
1 is jumping, while the meniscus labeled 2 is stable. The
opposite situation occurs for the two bottom pictures
(fps stands for frame per second).
the position of the meniscus is stable over the same
period of time. The top two pictures in Figure 5.39 and
then the bottom two pictures in the same figure represent
a single frame step for each pair, or 66.7 ms between
frames. This means that the visualized Haines jumps
occur faster than the frame rate of the digital video
camera.
5.4.3 Discussion
Drainage was begun by opening the control valve and
terminated by closing the control valve (Figure 5.38).
Figure 5.40 is broken down into three different phases,
where the phase preceding the drainage is labeled phase
I, while phase II corresponds to the drainage time of the
experiment, and phase III corresponds to the postdrai-
nage relaxation time. It can be seen in Figure 5.40 that
each phase has different electrical characteristics. The
channel-by-channel DC properties of the records were
normalized to the DC levels in phase I, making most
channel responses near zero in this region. The phase
I region is characterized by relatively stable, near-zero
trends with a low RMS noise.
The drainage region (phase II) shows an expected
positive self-potential trend during water table lowering
(drainage). This low-frequency dynamic behavior is due
