gamma-phase-associator

You associate seismic phase picks by applying Gaussian Mixture Models to arrival data. This groups P/S phase picks with station metadata to estimate earthquake locations, origin times, and magnitudes across a network.

Phase association can use either Bayesian Gaussian Mixture Models (BGMM) or standard GMM. BGMM incorporates prior distributions to better handle uncertainty in arrival data, while standard GMM fits clustering without explicit prior probability weighting.

Clustering phase picks with GaMMA detects multiple earthquake events by fitting Gaussian mixture models to seismic arrival data. This probabilistic clustering groups discrete P/S phase picks into distinct events based on spatial and temporal proximity.

Earthquake phase association requires input DataFrames for both seismic picks and station metadata, along with a configuration dictionary. These inputs allow the associator to process networks and output detected events and assignments.

You can use Python to automate earthquake detection by processing picks and stations DataFrames through the associator's API. This outputs a list of detected events with estimated locations and times across multiple seismic stations.

Using Gaussian Mixture Models for seismic phase association requires accurate P/S phase picks and station metadata to function effectively. This clustering approach estimates earthquake parameters probabilistically, meaning severely noisy arrival data can degrade event detection accuracy.