Feature addition

[calum-chamberlain]

The following features should be added:

• [ ] SNR frequency plotting, possibly similar to PPSD, e.g. ratio of possible template (signal) over noise in different frequency bands - either do in time domain with filtered data, or divide the ffts?
• [ ] Template spatial resolution test;
• [ ] Synthetic detection threshold testing;
• [ ] autocorrelation template detections;
• [ ] cross-station template detection.
• [ ] Subspace automatic thresholding from user-defined false alarm probability
• [ ] Add PCA relative moment calculation per Shelly et al., 2016
• [x] Break clustering.cluster() into two functions. One calculates distance matrix, the other does clustering.
• [ ] Add more options for clustering
  • [ ] options for method for hierarchical
  • [ ] divisive clustering
  • [ ] k-means
  • [ ] whatever else could be useful!
• [x] Use logging for debug statements (user to supply logger, or create internally using debug= variable to set level); working on in #225
• [ ] Subspace classes to mimic Tribe, Template, Party and Family; see #154
• [ ] Subspace to make use of correlate functions;
• [ ] lag_calc should have the ability to re-cut templates (e.g. you might use a 6s template with both P and S phases in for match-filtering, but you might then want to separate the P and S phases for relocation).
• [ ] client_detect could run a background download thread to download the next set of data while detection happens (and process it)... This would potentially be tricky in terms of memory usage, so having an option to not do this would be good.

Comments from @nackerley:

• [ ] Make it work better on cluster computers - use a database of some form for detections (pandas?), and possibly Dask for scheduling?
• [x] Allow KeyBoardInterupt for correlate functions/match-filter see #241
• [ ] Propagate Timeouts from Client methods into match-filter.
• [ ] Add ability to download from multiple clients in client_detect method.

[calum-chamberlain]

Added PWS stacking Added clustering

[calum-chamberlain]

Added SVD

[calum-chamberlain]

QuakeML and obspy event now separate issue, overseen by @cjhopp #23

[calum-chamberlain]

Catalog_to_dd should have a way to force separation of earthquakes across a fault. Will stop hypodd from clustering earthquake different terranes

[cjhopp]

I'm running into a similar issue currently.

[FinniganIK]

Couple of minor additions that I would find useful:

• [ ] Add option to utils.catalog_to_dd.write_event to output in GrowClust format (just needs spaces in date&time output)
• [ ] Some kind of progress information output for catalog_to_dd.write_correlations. i.e. report for every 100th catalogue event completed