Ramdomdec question
Posted: Fri Aug 05, 2022 3:33 pm
Hi Marc,
I submit this case history to you to understand the Ramdomdec parameter setting mode (Damping tool in Geopsy). The reference is the attached image.
The peak at 0.8Hz is indicated in the power spectrum, which I intend to test with this tool to define its nature on the basis of the value of z.
In my opinion, this is clearly an anthropogenic peak (as the Z component is much wider than the horizontal ones and it is impossible to deal with Rayleigh waves at frequencies lower than the minimum at 2.2Hz).
I considered three different sizes of the bp filter, with minimum and maximum frequencies equidistant from 0.8Hz, and two bases (4 and 8 ) of the Butterworth filter. Filter and window lengths equal to 8s (theory would say 10 / 0.8s, but testing with longer lengths gave worse results).
The best result, in my opinion, is the one related to the frequency band 0.6-1.0Hz: the frequency of the vertical peak is calculated at 0.71Hz and the damping z = 3.9%.
The other two solutions with larger bandpass width lead to smaller peak frequency values and z values> 5%.
The conclusion I arrived, however, is that the best setting corresponds to the one for which there is the best fit between the mean of the Random Decrement and the exponentially decreasing function.
I also tested the HV peak with the same approach.
It's correct?
Thank you
Luigi
p.s. If you feel, for any reason, that you do not have / can reply, do not publish this post. Participation in the forum is to expose one's doubts, to learn from the experience of others. But if you do not answer, nobody answers and therefore the level of knowledge does not increase and then the publication alone is of no use. I've noticed that most of the unanswered posts are mine ... I'd like to know why (although I suspect what that might be).
I submit this case history to you to understand the Ramdomdec parameter setting mode (Damping tool in Geopsy). The reference is the attached image.
The peak at 0.8Hz is indicated in the power spectrum, which I intend to test with this tool to define its nature on the basis of the value of z.
In my opinion, this is clearly an anthropogenic peak (as the Z component is much wider than the horizontal ones and it is impossible to deal with Rayleigh waves at frequencies lower than the minimum at 2.2Hz).
I considered three different sizes of the bp filter, with minimum and maximum frequencies equidistant from 0.8Hz, and two bases (4 and 8 ) of the Butterworth filter. Filter and window lengths equal to 8s (theory would say 10 / 0.8s, but testing with longer lengths gave worse results).
The best result, in my opinion, is the one related to the frequency band 0.6-1.0Hz: the frequency of the vertical peak is calculated at 0.71Hz and the damping z = 3.9%.
The other two solutions with larger bandpass width lead to smaller peak frequency values and z values> 5%.
The conclusion I arrived, however, is that the best setting corresponds to the one for which there is the best fit between the mean of the Random Decrement and the exponentially decreasing function.
I also tested the HV peak with the same approach.
It's correct?
Thank you
Luigi
p.s. If you feel, for any reason, that you do not have / can reply, do not publish this post. Participation in the forum is to expose one's doubts, to learn from the experience of others. But if you do not answer, nobody answers and therefore the level of knowledge does not increase and then the publication alone is of no use. I've noticed that most of the unanswered posts are mine ... I'd like to know why (although I suspect what that might be).