SPAC analysis hanging

[kschramm]

Hello!

I am trying to do a SPAC analysis on a 24-element array
with 1 hour data @ 250 samp/sec. I am on a mac running
SnowLeopard (10.6.4). When I try to run the SPAC toolbox from the GUI
- pressing the "Start" button under the "Status" tab, it says "Loading
Samples" in the bottom left corner of the main GUI window. The SPAC
GUI always says "Not Running" although it shows a status bar that says
Finished and has a blue/white striped bar that moves. Nothing else
ever happens - it just hangs. I have attached the .log file.

Any advice on how to fix this would be appreciated.

Kimberly

[admin]

I checked your log and found that no time windows are selected for all frequencies (end of log). From the information available in the log, I found no irregular settings. Can you send me a sample of these files by email or through a ftp site, to check things?

[admin]

After checking your dataset, I get the same results as you had.

If you browse the log messages after trying to run the SPAC process, you will notice this warning:

Code: Select all

Station 'C02': incompatible sampling from 79205 to 81000 s, skipping range

It means that station C02 has not the same sampling (distinct sampling frequency or t0) as the last station checked so far (C01). If you display all signals in a table, effectively nearly all T0 are different and not compatible with each other (difference is not a multiple of sampling period).

I tried to load the SEGY files (that resulted to be little endian Passcal SEGY). The T0 are also not compatible. Components are identified as Vertical, North or East which does not look so correct. For PASSCAL SEGY, component are stored at offset 13 to 16 from the beginning of header, by a simple number. This field does not seem to be set according to general standards (IRIS). How did you obtain those SEGY files?

[kschramm]

The data was recorded on four separate Reftek 72 recorders and converted from reftek .dat files to Passcal Seg-Y using the Passcal software suite program 'ref2segy' (Version Number 2009.232). This was done on an intel OS X machine. The start times should have been made consistent when I used the Passcal program 'segymerge' as follows:

segymerge -c 02.271.*8144.1 -s 02:271:22:00:00 -l 3600 > C01.sgy
segymerge -c 02.271.*7999.1 -s 02:271:22:00:00 -l 3600 > C02.sgy
etc...

Looking at the headers in the passcal program PQL, the sampling rate is a consistent 250 SPS, while the start time does vary by 0.003 seconds (but that can be fixed by starting the analysis at t0 + 1 or so seconds, correct?).

Thanks for any help.

[admin]

Effectively sampling frequency is a consistent 250 Hz. The problem is these small shifts in T0 (0.003, 0.006,...) that are not a multiple of the sampling period. You can ignore them and round all T0 to the closest multiple of 0.004. T0 are not necessarily equal, they just need to be compatible (differences must be a multiple of sampling period). Another more rigorous way would be to shift the phase by a fraction of the sampling period (menu "Waveform/Phase shift"). It does not the change T0, only the signal is translated. Then, manually adjust T0.

Code: Select all

If T0=0.003
Phase shift by 0.001
Set T0 to 0.004

Positive or negative shift is always an issue. I never remember what direction to choose. The best is to check it graphically. You can check the result between the original signal and the shifted one by plotting both signals in the same graphical window with the mode "Overlayed" for "Y axis" (layer properties).