*.HV file from HV tool

[riccardo]

good morning
I have some problem with HVSR tool.
I try to define the SESAME criteria for a reliable H/V curve, using the *.HV and *.log file in output from the H/V tool.

*.HV file is made with some #line that define the f0 from window, f0 from average, f0 peak amplitude and , I think, the standard deviation from "f0 from window, i.e.

# GEOPSY output version 1.1
# Number of windows = 71
# f0 from average 1.18027
# Number of windows for f0 = 71
# f0 from windows 1.17592 1.05407 1.29777
# Peak amplitude 2.98443# Frequency Average Min Max
0.5 2.45394 1.45698 4.13309
0.517476 2.38562 1.41309 4.02747
0.535563 2.32519 1.4446 3.74257
0.554283 2.21936 1.41466 3.48179
0.573656 2.16854 1.45623 3.22929


1) which is the difference between "f0 from average" and "f0 from window"?

2) there is a little error in the HV file, the #Frequency line isn't separate from the peak amplitude line by "end line" (^p), so the import in excel is a little bit more complicate (but it is a very little problem).

3) the number after "f0 from window" are average pick, av. - stdev, av. + stdev?

4) column min and column Max are the av.- stdev and av.+ stdev?

Someone can you help me?

[admin]

1) which is the difference between "f0 from average" and "f0 from window"?

• f0 from average is computed by scanning the average curve and identifying the frequency at which the maximum amplitude occurs.
• f0 from window is computed differently. The scan is run on each individual curve (corresponding to one window). Several peak values are then extracted. From this population an average and a standard deviation is assessed.

2) there is a little error in the HV file, the #Frequency line isn't separate from the peak amplitude line by "end line" (^p), so the import in excel is a little bit more complicate (but it is a very little problem).

Thanks for report, now fixed for future releases

3) the number after "f0 from window" are average pick, av. - stdev, av. + stdev?

Yes, it is

4) column min and column Max are the av.- stdev and av.+ stdev?

No, it is av/stddev and av*stddev. Amplitude statistics are computed on a log scale or in other words it is a geometrical average.

[riccardo]

thank you,

Now I understand what does it mean from SESAME 3rd poit of reliability
(σ A (f) = "standard deviation" of A H/V (f), σ A (f) is the factor by which the mean A H/V (f) curve should be multiplied or divided )

I try to use this information for excel spreadsheet

Riccardo

[riccardo]

Good morning Mark

for the SESAME criteria which f0 I have to use?
f0 from window or f0 from average?

I ask this because I try to create a spreadsheat with calc for the SESAME criteria ... but
in the *.hv txt file there are the data from average h/v curve
and if i look for f0/2 - 2f0 or f0/4 4f0 ranges I study the average curve and I can't find the f0 from window that is the f0 of the site.

differences betwen two f0 is minimal ... but !

what is correct to do?

[bcosenza]

I have the same doubt, about which is the right frequency to use, not only to check the SESAME criteria, but to report as the fundamental frequency of the soil one's studying.

I also would like to know if there is an upper limit to the width of the windows. I read the SESAME guidelines and the minimal width is quite well defined. We suspect our target frequency is between 0.8 and 1 Hz, and we are using 60 second windows over 30 min long registers, could this be a problem?

It is a recommendation to measure as far away from buildings as possible. We have sets of measures on fields but we also have one set of measures that taken inside a building. This measures exhibit a peak displaced to the right by more or less 1.5 Hz. Is the frequency obtained inside the building reliable or should we asume we're not measuring the soil response, but the response of the structure?

Thank you very much for your help!

[Pedro Freire]

Hello,

I have the same question, in some HVSR I have done with lower values of H/V peak( around 2), the f0 from the average curve is a bit different from the f0 from windows, sometimes with differences of 0.2Hz.
When checking the Sesame criteria for a well-defined peak, these curves will most likely not meet the criteria due to their flatter shape and small A0 peak, however wich f0 should be used for the verification, the fo from windows or f0 from average?
The program Geopsy does not seem to calculate standard deviations for the f0 from average, is it correct to use the f0 from windows when checking the Sesame criteria?

Any help would be appreciated!
Thank you.

[admin]

Hi all,

I will try to answer all these previous posts.

As mentioned by Riccardo, the "difference between the two is minimal... but"
The HVSR curve should be taken as it is and it should not be over-interpreted. In my view this is a qualitative estimation of the site resonance frequency, especially if the peak is not so sharp.

I would not care too much about a 0.1 or 0.2 Hz difference except if it is deep site with, for instance, a peak frequency around 0.4 or 0.5 Hz. What is important is the relative error. [To Pedro] What is the obtained f0 value to get an idea of the relative difference?

There is no upper limit for the time windows. The longer is always the best but above 20 or 30 periods (time window length of 30 s for 1 Hz target), provided that the noise is stationary, it makes little difference. The best is to check your self the results. Increase the length and see if it has an influence over the shape of the curve. Below 10 periods, for sure, you can distort the HVSR curve at low frequency.

A single HVSR curve with a peak that do not fit with the SESAME criteria is difficult to interpret and it might reflect a complex geological structure. It is always interesting to compare several curves around and see it the same pattern is observed elsewhere. If so this is certainly the signature of the soil structure even though the SESAME criteria are not met.

Inside the building, computing a HVSR has little meaning, I'm not an expert in this field but plain spectra are better to estimate the building behaviour. If you take your HVSR on the soil too close to the building, the resonance frequency of the building itself will be visible on the soil spectra and hence may bias the HVSR curve. If you can measure several HVSR points at various distances from the building, you will probably see the increasing influence. There are certainly many references in the literature about that, e.g. figure 7:

Cécile Cornou, Philippe Gueguen, Pierre-Yves Bard, E. Haghshenas. Ambient noise energy bursts observation and modelling: Trapping of harmonic structure-soil induced-waves in a topmost sedimentary layer. Journal of Seismology, 2004, 8 (4), pp.507-524

"F0 from average" is a single value picked from the unique average curve. it is not possible to compute a standard deviation out of a single value. sigma_f is the standard deviation computed from the time windows. sigma_A is not computed by geopsy but only sigma_logHV, if I refer to the terminology used in H/V guidelines.

Best regards,

Marc

[Pedro Freire]

Thank you very, mutch Mark!
It was really helpful.
The difference was around 0.15 Hz, between 1.35 and 1.5 Hz, so the relative error is quite small.