geopsy.org

It is no longer possible in version 2.2.0 to look at the H/V ratios for individual horizontal components, I mean N/V and E/V ratios?

Marc,

As I want to compute separate H/V for both horizontal components, I try to compute H/V ratio with only one horizontal component. I found that it is not possible any more: when lauching H/V tools box, on an error message tells me that a component is missing for the given station.

Since, it is not possible not only to look at such 'separated' ratios (N/V and E/V) (see my previous post), but also to compute it, what can I do?

I hope you will be able to find out solution!
Thanks

Sylvette

This is true, I removed the automatic output of the N/V and of E/V. It was mainly guided by practical issues. I would like to have a less messy desktop when computing H/V. In most situations, just H/V (with a good average of both horizontal components) is sufficient. For other special cases where you want the ratio in a particular direction, I add an option for combining horizontal components. Three options are currently available:

1. Horizontal energy
2. Azimuth
3. Squared average

Historically, during SESAME project, H=sqrt((N^2+E^2)/2) (option 3) where N and E are real values, they are smoothed spectral amplitudes. By combining them in the real domain, we implicitly assume that E and N are perfectly in phase. In real world, they are never in phase, so we certainly over-estimate H. Moreover, the smoothing add some strong non-linearities, better to introduce it at the last step or our processing. Combining the horizontal components in the complex domain (option 1) has the advantage that the total energy assigned to H takes into account the phase difference between E and N.

Options 1 and 3 are slightly equivalent if we are just interested in the frequency of the peak. On the contrary, if we start looking at the amplitude of the spectral curve, thinks are somehow different. I noticed that sqrt(N^2+E^2) computed in the real domain is about 15% higher than sqrt(N^2+E^2) computed in the complex domain. Hence, the ratio between the curves obtained with options 1 and 3 is around 20% (option 1 gives higher values).

Finally option 2 lets you steer the horizontal components to any direction with a combination of N and E in the complex domain (equivalent to a rotation in time domain). The azimuth is counted clockwise from North (0 is equivalent to N/V and 90 is equivalent to E/V). Another possibility is also the H/V Rotate tool which scans all azimuths from 0 to 180.

Yes, I see better how I can compute N/V and E/V ratios.
Thanks a lot for your explanations!

Excuse me. I still have some problem on N/V and E/V.
I see that you removed the automatic output of the N/V and of E/V. On the other hand, you provided Azimuth (option 2). I think that is Directional energy function written in the wiki, am i right?

I want to calculate N/V and E/V separately, but i cannot only upload N-S and vertical data or E-W and vertical data to calculate H/V. If i upload three data (N-S,E-W,vertical) and use the Directional energy function in 90 degree, will it still use three data to calculate or it will only use E-W and vertical data? If it uses data in three directions to calculate, can it represent E/V reasonably?

These are my questions.
Thanks!

Hi,
If I properly understood your question, if you select direction 90 or 270 deg, you will get E/V. 0 or 180 deg to get N/V. When using 90 deg, the N-S component is multiplied by zero.
As defined by SESAME in 2004, the squared average horizontal spectrum is defined by sqrt((N^2+E^2)/2). If you try to process H/V with a fake null N-S component and if you select "squared average", there will be a factor sqrt(2) difference with the result obtained with "directional energy".