Toggle Accessibility Tools

Brian Zhu

Southern California is a region of very high seismic activity. In order to better understand the seismicity of the region, we propose methods to cluster earthquakes using unsupervised machine learning algorithms. We use the k-means algorithm with additional parameters of distance to major fault lines and earthquake density to produce clusters that align more closely to fault zones than those produced with 2D or 3D space as the only parameters. We then analyze clusters where exceptionally large earthquakes have occurred in the past and where groundwater injections have occurred recently for trends in the b-value (ratio of small to large earthquakes).