Difference between revisions of "Geopsy: H/V and Spectrum Toolboxes: Processing Tab"

From GeopsyWiki
Jump to navigation Jump to search
Line 13: Line 13:
 
This area, common for [[H/V_spectral_ratio|H/V]] and [[Spectral_amplitudes|Spectrum]] computings, is divided in three different parts with three different objectives :
 
This area, common for [[H/V_spectral_ratio|H/V]] and [[Spectral_amplitudes|Spectrum]] computings, is divided in three different parts with three different objectives :
  
*fixing which smoothing will be applied to the windows (2 first lines).
+
=== Smoothing ===
 +
Fixing which smoothing will be applied to the windows (2 first lines).
 
During the computing of [[H/V_spectral_ratio|H/V]] or [[Spectral_amplitudes|Spectrum]], the Fourier spectra can be smoothed (it is strongly recommanded) in the goal to clarify the global aspect of the curves.
 
During the computing of [[H/V_spectral_ratio|H/V]] or [[Spectral_amplitudes|Spectrum]], the Fourier spectra can be smoothed (it is strongly recommanded) in the goal to clarify the global aspect of the curves.
  
Line 31: Line 32:
  
  
*fixing the use or not of a [[Geopsy:_Taper|cosine taper]] on the individual window. In the goal to minimize the border effect due to the extraction of a window and so improve its properties in the frequency domain, it is strongly recommended to use a taper to avoid the creation of false frequencies.
+
=== Taper ===
 +
Fixing the use or not of a [[Geopsy:_Taper|cosine taper]] on the individual window.
 +
 
 +
In the goal to minimize the border effect due to the extraction of a window and so improve its properties in the frequency domain, it is strongly recommended to use a taper to avoid the creation of false frequencies.
 +
 
 +
 
 +
=== Filter ===
 +
Fixing the use or not of a [[Geopsy:_Filter|filter]] on the whole signal.
 +
 
 +
This option allows to filter the signal before the extraction of the time windows.
 +
 
 +
To the contrary of the [[Geopsy: Waveform Menu|Waveform menu filter]] there is only a "High pass" filter allowed, rejecting frequencies below a given value
 +
 
 +
Be carefull with this option. As the signal is high pass filtered, the results below the choosen frequency will display a lake of data.
 +
If the studied frequency is close to 1 Hz and it is applied this filter at 2 Hz, the results will lead to bad data in the frequency of interest
 +
and then to flaky interpretation.
 +
 
 +
 
 +
 
 +
[[Image:HV_SPEC_toolb_process_param.png|thumb|right|300px|''Processing'' tab giving access to smoothing, parameters (for windows) and how is processed the H/V (available only for [[H/V_spectral_ratio|H/V]] computing)]]
  
 
== ''Horizontal components'' ==
 
== ''Horizontal components'' ==

Revision as of 15:29, 10 March 2010

Processing tab giving access to smoothing, parameters (for windows) and how is processed the H/V (available only for H/V computing)

Here are described the different possibilities offered in this tab. This tab is divided in two areas:

-Parameters area

-Horizontal components

Parameters area

Processing tab giving access to smoothing, parameters (for windows) and how is processed the H/V (available only for H/V computing)

This area, common for H/V and Spectrum computings, is divided in three different parts with three different objectives :

Smoothing

Fixing which smoothing will be applied to the windows (2 first lines). During the computing of H/V or Spectrum, the Fourier spectra can be smoothed (it is strongly recommanded) in the goal to clarify the global aspect of the curves.

For Spectrum, each Fourier spectrum (coming from a window) is smoothed and then the averaged curve is computed.

For H/V, the horizontal Fourier spectra (NS and EW, coming from a synchronous window) are first added and then the smoothing is applied on the merged horizontal Fourier spectrum and on the vertical Fourier spectrum.


There are 4 possibilities for the smoothing: Process param smoothingtypes.png

  1. no smoothing. This can be sometime usefull to have a rough idea about Fourier spectra;
  2. Konno and Ohmachi smoothing. This smoothing use a constant bandwidth in a logarithmic scale and is strongly recommanded because this smoothing function preserves the different number of points at low and high frequency. This smoothing is controled by a smoothing constant varying in-between 0 and 100. A constant of 0 gives a very strong smoothing, when a constant of 100 a very soft smoothing, more details;
  3. Constant smoothing. This smoothing function has a triangular shape centered on the current frequency and its width is equal to "Band width" given by the user. This band width varies from 0 (soft smoothing) to 100 (strong smoothing);
  4. Proportional smoothing. This smoothing function has a triangular shape and its width depends upon the current frequency. The half width is defined by percentage*frequency. The value of "percentage" cannot be greater or equal to 100% (strong smoothing).


Whatever the smoothing method used, even if the impact of the smoothing is minimized, the result will be different from a processing without smoothing.


Taper

Fixing the use or not of a cosine taper on the individual window.

In the goal to minimize the border effect due to the extraction of a window and so improve its properties in the frequency domain, it is strongly recommended to use a taper to avoid the creation of false frequencies.


Filter

Fixing the use or not of a filter on the whole signal.

This option allows to filter the signal before the extraction of the time windows.

To the contrary of the Waveform menu filter there is only a "High pass" filter allowed, rejecting frequencies below a given value

Be carefull with this option. As the signal is high pass filtered, the results below the choosen frequency will display a lake of data.
If the studied frequency is close to 1 Hz and it is applied this filter at 2 Hz, the results will lead to bad data in the frequency of interest
and then to flaky interpretation.


Processing tab giving access to smoothing, parameters (for windows) and how is processed the H/V (available only for H/V computing)

Horizontal components

that indicate how the H/V is computed


Konno K. and T. Ohmachi, 1998. Ground motion characteristics estimated from spectral ratio between horizontal and vertical components of microtremors. Bull. seism. Soc. Am., 88-1, 228-241