stfinvfourier_description_usage.txt 1.34 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
# this is <stfinvfourier_description_usage.txt>
# ----------------------------------------------------------------------------
#
# Procedures in the Fourier domain
# --------------------------------
Options and parameters in common for procedures in the Fourier domain:
  fpow2        use power of two for number of coefficients
  fdiv=d       use integer multiple of d for number of coefficients
  fpad=f       padding factor
  tshift=d     delay source correction filter wavelet by d (in seconds)
               in order to expose acausal components

These options define the number of samples N used for the FFT (Fast Fourier
Transform). This number N should be larger than the number of samples M in the
original input time series to avoid wrap-around. If fpow2 is set, N will be
the next power of 2 larger than M*f. Else if fdiv is set, N will be the next
integer multiple of d larger than M*f.

The derived correction filter in some cases can contain acausal components.
This means that the impulse response is non-zero for negative time values.
Since by definition, the impulse response is output for the time interval of
the input data, these acausal components can remain unnoticed. The option
tshift can be used to shift the impulse response as obtained by inverse FFT in
order to expose acausal components.
# ----- END OF stfinvfourier_description_usage.txt -----