Memory overflow and segmentation fault

[paulaasg]

Dear Min Liu,
 
I have been using your repository to analyze seismic data for my research project. However, I am having some issues analyzing my complete data set, which consists of daily waveform files from 2012 to 2023. The program works fine until it starts reading the waveform files; it takes a couple of minutes for the RAM memory to fill up to about half its capacity, which is 82 GB total. Then, the program crashes and I get errors like memory overflow or segmentation fault. I have tried analyzing smaller sections of data and it worked, but only when reading 2 to 6 days maximum.  I’ve also adjusted some parameters like number of event pairs, number of stations, etc., but nothing has worked yet. I’ve looked through some C manuals to see if it could be possible to modify some of the memory locating functions to make it work but haven’t been successful in doing so. I am aware that it is a lot of information to process, but I was hoping there could be a way to solve it.  I’d appreciate any help or advice on this issue.
Thank you in advance for your time, best regards!
 
Paula Salazar

[vanagno]

Hello. I'm having a similar problem and i've noticed that specific waveform files appear to crash the program, leading to a segmentation fault (core dumped) error. Removing those files make the program run successfully, but the waveforms themselves have no problems, NaN values or anything weird. I would also like to know if there is any way to solve this issue, as it is rather tedious to identify which of the otherwise good quality waveforms crash FDTCC.

[dylanmaximos]

hello @paulaasg @vanagno ,
I'm encountering the same issue where SAC files can't be read, which prevents me from successfully running the FDTCC step. Do you happen to have a script you could share to convert MiniSEED (mseed) files to SAC format? That would be really helpful!

[dylanmaximos]

Hello. I'm having a similar problem and i've noticed that specific waveform files appear to crash the program, leading to a segmentation fault (core dumped) error. Removing those files make the program run successfully, but the waveforms themselves have no problems, NaN values or anything weird. I would also like to know if there is any way to solve this issue, as it is rather tedious to identify which of the otherwise good quality waveforms crash FDTCC.

Hello i have the same issue but when the event is <600 there is no error same station number on all SAC files and when the event is > 5000 there is segmentation error.

furthermore, please what are you referring to NaN or weird values"? I have check the error on the SAC files but none, if you can help @vanagno
here is when i cut the events and i works now

Help is really needed @alirezaniki , @MinLiu19 , @Liuminupday @hhhhcsu

[alirezaniki]

@dylanmaximos @vanagno @paulaasg Hi all, my workaround for this issue was to ensure that all my daily data included 8640000 samples (considering a sampling rate of 100 Hz). The program crashes if a single expected time window for P or S phases has no/incomplete data. For the below dt.ct example:

#         1         7
B921      5.345   5.182 1.0000 S
B921      2.845   2.832 1.0000 P
B918      4.665   4.672 1.0000 P
B918      8.405   8.432 1.0000 S
B917      6.785   6.782 1.0000 P
B917     11.935  11.832 1.0000 S

The code calculates CC times for all stations listed below the event pair 1 & 7. Therefore, it looks for (expects available data) the corresponding data from those stations on the days the two events are detected.