Dear Marc,
I'm having some problems with the SPAC tool, I don't know the principle of "rings" window, how to set parameters, how to select site data, and whether the X and Y coordinates of the site should be set accordingly.
I have searched some video tutorials on the Internet, but there is nothing about the SPAC module, so I don't quite understand it. I hope you can give me some guidance and it would be better if there are specific operation steps. Thank you very much!
Best regards,
Wang Ziqian
SPAC
Re: SPAC
Hi Wang,
I suggest to read Bettigs et al. (2001). In Fig 3b, you have the representation of the co-array map that helps the definition of rings. In the classical SPAC, the distance between stations must be the same for a set of station pairs. IN MSPAC proposed by Bettig et al., there is a tolerance and the distances may vary in a limited range, what's called a "ring". Rings can overlap. Rings must be kept as thin as possible in a relative way. The azimuth coverage is important, it must be as large as possible (max 180 deg). Large radius rings can be thicker than smaller ones. You can try several definitions for your rings. It does not change the results in a significant way. It is just a different sampling in the space domain.
About the "Time" parameters, the most important is the window length. A value of "50 T", which means 50 periods (corresponding to the current central frequency) per window, is a good choice. It is an adaptative length which produce larger windows at low frequency.
For the "Processing" parameters, the first part is the frequency sampling which is generally chosen with a log distribution. The next two sections concern the way cross-spectra are computed: block averaging and frequency smoothing (bandwidth). Oversampling is useless for SPAC on passive data.
The option about the imaginary part of the autocorr is a filter that can be used to reject some bad autocorr estimates. The details are available in Asten et al. (2006). A value of 1 keeps all values without filtering.
Best regards,
Marc
I suggest to read Bettigs et al. (2001). In Fig 3b, you have the representation of the co-array map that helps the definition of rings. In the classical SPAC, the distance between stations must be the same for a set of station pairs. IN MSPAC proposed by Bettig et al., there is a tolerance and the distances may vary in a limited range, what's called a "ring". Rings can overlap. Rings must be kept as thin as possible in a relative way. The azimuth coverage is important, it must be as large as possible (max 180 deg). Large radius rings can be thicker than smaller ones. You can try several definitions for your rings. It does not change the results in a significant way. It is just a different sampling in the space domain.
About the "Time" parameters, the most important is the window length. A value of "50 T", which means 50 periods (corresponding to the current central frequency) per window, is a good choice. It is an adaptative length which produce larger windows at low frequency.
For the "Processing" parameters, the first part is the frequency sampling which is generally chosen with a log distribution. The next two sections concern the way cross-spectra are computed: block averaging and frequency smoothing (bandwidth). Oversampling is useless for SPAC on passive data.
The option about the imaginary part of the autocorr is a filter that can be used to reject some bad autocorr estimates. The details are available in Asten et al. (2006). A value of 1 keeps all values without filtering.
What do you mean? X and Y coordinates of station must be properly set before any array processing. See "Set receivers" in menu "Edit".whether the X and Y coordinates of the site should be set accordingly
Best regards,
Marc