Array Geometries and the Analyzable Quantities
What
Array Geometries Is BIDO Good for?
BIDO's array analysis theory is based on the
assumption of a circular array installed on the ground surface (with sensors at
all locations around a circle). In practice, however, problems arise as to how
many sensors should be installed around the circle and whether or not a sensor
should be installed at the center. To make a shortcut to the conclusion, BIDO
can be applied to arrays of 0, 1, 2, 3 or 5 (or odd numbers of) sensors
around a circle. An array can consist of a maximum number of 6 sensors if one
of them is installed at the center. The range of analyzable quantities differ
for different array geometries and record components.
The following describes the correspondence
relations between the array geometries, record components and the analyzable
quantities. The table lists the names of the quantities that can be analyzed
and of the analysis methods that can be implemented.
Array geometry (Black circles:
sensors) |
|
|
|
|
|
|
|
|
Total number of sensors |
1 |
2 |
3 |
3 |
4 |
5 |
6 |
|
At the center |
1 |
1 |
1 |
0 |
1 |
0 |
1 |
|
Around the circle |
0 |
1 |
2 |
3 |
3 |
5 |
5 |
|
Vertical
motion alone |
Phase velocities of Rayleigh waves |
None |
SPAC |
SPAC H0 |
CCA |
SPAC CCA nc-CCA H0, H1, V |
CCA |
SPAC CCA nc-CCA H0, H1, V |
Phase velocities of Love waves |
None |
None |
None |
None |
None |
None |
None |
|
Other quantities |
PSD |
PSD |
PSD |
PSD |
PSD NSR(V) |
PSD |
PSD NSR(V) |
|
Horizontal motion alone |
Phase velocities of Rayleigh waves |
None |
None |
None |
CCA-R |
CCA-R SPAC-R SPAC+R |
CCA-R |
CCA-R SPAC-R SPAC+R |
Phase velocities of Love waves |
None |
None |
None |
CCA-L |
CCA-L SPAC-L SPAC+L |
CCA-L |
CCA-L SPAC-L SPAC+L |
|
Other quantities |
PSD |
PSD |
PSD |
PSD |
PSD NSR(H) |
PSD |
PSD NSR(H) |
|
Three-component |
Phase velocities of Rayleigh waves |
None |
SPAC |
SPAC H0 |
CCA CCA-R |
SPAC CCA nc-CCA H0, H1, V CCA-R SPAC-R SPAC+R |
CCA CCA-R |
SPAC CCA nc-CCA H0, H1, V CCA-R SPAC-R SPAC+R |
Phase velocities of Love waves |
None |
None |
None |
CCA-L |
CCA-L SPAC-L SPAC+L |
CCA-L |
CCA-L SPAC-L SPAC+L |
|
Other quantities |
PSD H/V |
PSD H/V |
PSD H/V |
PSD H/V |
PSD H/V R/V, R/L NSR(V) NSR(H) |
PSD H/V |
PSD H/V R/V, R/L NSR(V) NSR(H) |
(Notes)
- SPAC: Spatial Autocorrelation method丒(its more generalized
variant is applicable to horizontal motion, but here, the term is used to
denote "the method to analyze phase velocities of Rayleigh waves using
vertical motion)
- CCA: A method to analyze phase velocities of
Rayleigh waves using vertical motion (reference
[1, 3])
- nc-CCA, H0, H1, V: Methods to analyze phase
velocities of Rayleigh waves using vertical motion (reference
[4])
- CCA-R, SPAC-R, SPAC+R: Methods to analyze phase
velocities of Rayleigh waves using horizontal motion (reference [5])
- CCA-L, SPAC-L, SPAC+L: Methods to analyze phase
velocities of Love waves using horizontal motion (reference
[5])
- PSD: Power-spectral densities
- H/V: H/V spectra (Power ratios of horizontal motion
to vertical motion)
- NSR(V): Noise-to-signal ratios of the vertical
component (reference [3])
- NSR(H): Noise-to-signal ratios of the horizontal
components (reference [8])
- R/V: Horizontal-to-vertical amplitude ratios
(NOT power ratios) of the Rayleigh waves (reference
[2])
- R/L: Rayleigh-to-Love power ratios (reference [5, 7])
As shown in the above table, BIDO allows one not
only to analyze array data but also data from single-station measurements,
linear arrays of two stations and L-shaped arrays of three stations.
(It can also cope with huddle tests where more than one seismic sensors are
assigned to a single location). What we had in mind during the early phase of
development was the analysis of circular-array data, but we later came to
realize that, at real measurement sites (especially in urban areas), simpler
array geometries are preferred because of their ease of installation. It is
coming to be known that linear arrays and L-shaped arrays can be applied to the
analysis of vertical motion in a number of cases, and so we decided to
incorporate these into BIDO. The third and fourth columns on the left-hand side
of the table both refer to arrays of three seismic sensors, but the range of
applicable methods differ according to whether one of the sensors is regarded
as assigned to the center of the circular array or whether all of them are
regarded as lying around the periphery of the circle.
It goes without saying,
though, that larger numbers of sensors around the circle produce higher
precision and more stable analysis results. We recommend the installation of
six-sensor arrays, as shown in the right-hand column of the table, when small
arrays (10 m or less in radius) are to be used. Particularly, when using
miniature arrays of around 1 m in radius (see photos in the top page),
six-station arrays are expected to an effective design not only in terms of
high precision but also in terms of efficiency (see One
Approach to Make the Most of the Method's Potential).
Notes on the
Array Analysis of Horizontal Motion
Array analysis of horizontal motion is currently
still on the level of basic research. It has been shown that Love wave
velocities and other properties can be analyzed using an array of five sensors
around a circle and another at its center (references
[5, 7]), but the extent of reliability of the analysis results from
arrays with only three sensors around a circle remains yet to be studied.
BIDO is designed in such a way that the analysis of horizontal motion is
possible even when the array has only three sensors around the circle, but this
feature is intended principally for the developer, and considerable caution is
required in interpreting the analysis output. Caution is also necessary when
S/N ratios are low (noise is large), because this considerably impairs the
reliability of analysis results even when five sensors are installed around the
circle. Unfortunately, there is, so far, no established theoretical rationale
for assessing the reliability of analysis results for horizontal motion on the
basis of S/N ratio estimates.
On Deviations from a "Circular" Array
When using circular arrays with three or more
sensors around a circle, the stations do not necessarily have to be installed
at equal intervals around the circle (reference
[2]), but equidistant arrays are preferred to ensure high precision.
There is no theory of correction for stations that do not lie on the circle
(location errors in the radial direction), so the program sets a default allowance
limit of 5% for such errors (modifiable by program recompilation) and regard
the stations as lying on the circle.