Non-Volcanic Tremor Analysis
Overview
The goal is to select a subset of seismological stations that are most suitable for detecting a relatively new seismologic phenomena (nonvolcanic tremor, NVT) throughout most of northern California and extending back in time to ~ 1993.
There are ~ 790 stations to select from. Desired properties of the selected subset are:
* Continuous data (not earthquake triggered) only.
* Maximize operational coverage since 1993
* Separated from other stations by at least 3 to 5 km
* Suitable geographic coverage of subset stations
* Sensitive to nonvolcanic tremor signal (S/N) in 3-8 Hz frequency band
There are meta data lists that allow selection of all but the last item in the list. The last item would require analysis of waveform data. So a potential list without waveform analysis could be generated first and the more involve process of final station selection could follow if time permits. In the final selection, maintaining geographic coverage would take precedence over station sensitivity.
Lots of station data are too noisy. Also, detection of the NVT requires station-pair analysis, so the number of computations for comparison pairs goes up pretty fast N*(N-1)/2 (N=number of stations).
The objective is to find a computationally managable subset of stations that is spatially well distributed with at least 'reasonable' sensitivity to NVT signal and that have been in operation (with relatively few data gaps) since ~1993 (beginning of continuous data available). Stations that have had changes in sampling rate or instrumentation over time are acceptable if sensitivity to NVT (basically S/N) in the 3-8 hz band is maintained.
Dataset
The data are available from the Berkeley Seismological Laboratory.
References
Here is a recent overview of NVTs.