The program is invoked from the command line
gendis
You will be prompted for the following inputs:
- Name of output file: This is the file which will be read by
disper. Recommend using disper.d for a name. Limit to 40
characters.
- Sample rate: Actually, not really a rate, this is the sample
interval in seconds when following disper with a synthetic seismogram
computation in waves. It is relevant to the frequency step size (see
next parameter).
- Tmax: This is the maximum time that will be used in the
waves synthetic seismogram
program. It will affect the step size in the frequency domain, and will be
adjusted slightly so that a radix 2 FFT can be employed. That is, in the time
domain, the number of samples will be a power of 2. The frequency increment
will be
, where N=number of samples, and
=the sample interval specified by the above parameter. Also, if you
plan on computing a synthetic seismogram, you probably don't want FFT wrap a
round. Thus, consider your slowest velocity in your model, and the maximum
offset of a synthetic which will be computed in waves. Use a safety
factor of 2 (make tmax 2 times larger than the latest arrival on the far
offset). What ever you decide here, USE THE SAME VALUE LATER IN WAVES.
- Minimum frequency: This is the minimum in Hz. Recommended value
is 1, even if you have a geophone with a higher cut-off frequency in mind.
Choose this and the next parameter to give a wider bandwidth than you think you
will need (unless you are not going to compute a synthetic, and only want a plot
of the dispersion curves). The reason is that the waves program computes
a scaler source moment (wavelet) based on this and the next parameter. The
wavelet will be minimum phase. Too narrow a bandwidth will give a very
ringy synthetic seismogram.
- Maximum frequency: Specify in Hz also. Too high a value may lead
to an unstable computation, where too high depends on the model parameters and
the inherent difficulty in integrating a stiff equation. Too low a value will
lead to a ringy wavelet if you go on to compute a synthetic with program
waves.
- Maximum mode number, modemx: This is the maximum mode to compute.
The limit is 9 modes. For a fundamental only computation, set to 1.
- Step size in depth, deltz: Since the propagator matrix method is
used, there is little value in using too small a step size in computing the
motion-stress vectors. Too small is defined as being a lot of computations in
each layer. For a layer over a half-space, use the upper layer thickness. For
a many layered case, use the smallest layer thickness. The integration is
automatically adjusted to layer boundaries.
There is an option in disper to compute the motion-stress vectors at a
single frequency and plot these with depth. If you are doing that, then a small
step size would be appropriate. This type of computation is usually done after
a dispersion curve computation, since you will need to provide a phase velocity
for the specific frequency of interest.
- Number of control points: This is the number of points in depth
where velocities of P-wave, S-wave and mass density are specified. Linear
interpolation based on the choice of depth step size above will create a layered
model. To create layers larger than the step size, more control points are
needed. For example, consider specifying a single thick layer over a
half-space. This would require 3 points (as shown in the gendis
manpage). One always starts at depth 0, surface of the earth. For a thick
layer over a half-space, the second control point would be at the top of the
next layer, and have material properties equal to the first control point. The
third point would be a very small distance below the second control point, and
give the half-space properties. See the man pages for this example, or later
sections below.
- Layer properties: Actually, these are comma separated quadruplet
entries giving the shear velocity (beta), P-wave velocity (alpha), mass density
(rho), and control depth (tops of layers). You will be prompted until all the
control points have been entered (it loops, and you can't go back).
Subsections
pm
2018-04-08