Closed PeterMakus closed 1 year ago
it looks like this is related to the preprocessing of the stream, specifically the tapering prior to response removal. My guess is that this is because the time length to be loaded from mseed is prolonged by the length of the taper.
Solved with latest commit in dev
. Instead of loading in additional data, the traces are now mirrored at the ends before tapering. The mirrored part is later trimmed again.
When
remove_response=True
,Correlator.pxcorr
sometimes acts up, when processing using mpi. This seems to be caused by a process that opens an hdf5 file atCorrelationDataBase._write()
and freezes while doing so. Then, the next process wanting to write to that file crashes as files cannot be opened by several processes.