- templates_max_similarity(st, time, streams_templates)¶
Compares all event templates in the streams_templates list of streams against the given stream around the time of the suspected event. The stream that is being checked has to include all trace ids that are included in template events. One component streams can be checked as well as multiple components simultaneously. In case of multiple components it is made sure, that all three components are shifted together. The traces in any stream need to have a reasonable common starting time. The stream to check should have some additional data to left/right of suspected event, the event template streams should be cut to the portion of the event that should be compared. Also see obspy.signal.trigger.coincidence_trigger() and the corresponding example in the Trigger/Picker Tutorial.
- computes cross correlation on each component (one stream serves as template, one as a longer search stream)
- stack all three and determine best shift in stack
- normalization is a bit problematic so compute the correlation coefficient afterwards for the best shift to make sure the result is between 0 and 1.
>>> from obspy import read, UTCDateTime >>> import numpy as np >>> np.random.seed(123) # make test reproducible >>> st = read() >>> t = UTCDateTime(2009, 8, 24, 0, 20, 7, 700000) >>> templ = st.copy().slice(t, t+5) >>> for tr in templ: ... tr.data += np.random.random(len(tr)) * tr.data.max() * 0.5 >>> print(templates_max_similarity(st, t, [templ])) 0.922536411468
- time (UTCDateTime) Time around which is checked for a similarity. Cross correlation shifts of around template event length are checked.
- st (Stream) One or multi-component Stream to check against event templates. Should have additional data left/right of suspected event (around half the length of template events).
- streams_templates (list of Stream) List of streams with template events to check for waveform similarity. Each template has to include data for all channels present in stream to check.
Best correlation coefficient obtained by the comparison against all template events (0 to 1).