v0.0.4

This is a comprehensive update to tackle some bugs and challenges, especially:

• reorganizes package in core and util modules
• core.template_creation:
  • makes parallel worker initiation and data reading quicker with better selection of required inventory / streams
• core.array_tools:
  • adds function to mask data steps on a seismic array that can lead to detection artefacts
• utils.relative_magnitude:
  • includes checks to measure amplitudes only on response-corrected traces
  • excludes outliers from the relative amplitudes for better relative magnitude estimate
• utils.growclust:
  • add functions to update an obsplus-event dataframe with growclust solutions
• models:
  • added barents16 velocity model from Pirli & Schweitzer 2017: Journal of Seismology Year: 2017 Volume: 22 Number: 1 Pages: 69-81
• examples:
  • added example scripts to run template creation, detection, & picking from the command line or with a SLURM job script

0.0.3

This is the first release of RobustRAQN, ready to be deployed for running template matching on large and heterogeneous datasets. It is still in Alpha state, so please don't hesitate to ask for help if you want to try it out!

Contains many improvements and speedups to the detection pipeline:

• more comprehensive & robust selection of real event detections
• full support for fmf2-backends:
  • SYCL code that can be compiled on any hardware with hipSYCL compiler (Nvidia, AMD, Intel GPUs, many CPUs)
  • AVX2 and AVX512 vectorization support for time-domain cross correlations on CPU with AVX2 / AVX512 support
• more criteria to robustly create events from detections:
  • minimum number of recording sites (individual arrays / stations)
  • checks for spurious array-detections on short time windows of original template
• more robust computation of magnitudes
  • min / max deviation from mean within multiple of standard deviations
  • min / max deviation from median within multiple of median average deviation
• quicker production a full bulletin of picks and magnitudes in one run.

v0.0.1-alpha

This is the first public release of RobustRAQN for preproduction use.