Introspective Systems earthquake location software was presented at the American Geophysical Union (AGU) convention in San Francisco between December 14-18 and made quite a buzz with participants. Sandia National labs, Berkeley University and Air Force Tactical Applications Command (AFTAC) have spent 7 years and millions of dollars in attempts to creat a real-time bayesian approach to locating earthquakes. They each spent more than 2 hours each studying the research and the real-time earthquake display being presented.
About the Software
Legacy processing approaches to seismic networks are based upon algorithms developed in the last century. Introspective Systems has developed new approaches that bring seismic Networks into the 21st century using all of the capabilities of modern computers for unprecedented analysis and monitoring of seismic events. Introspective Systems network capabilities are based upon xGraph a highly dynamic executable graph framework that distributes the analytics across millions of processes or processors.
The job of an earthquake data associator is to gather an ensemble of data types such as phase picks, beams, or other data from social media into another ensemble of collections identified as discrete events, each representing the occurrence of an earthquake, quarry blast, or nuclear detonation. Legacy approaches to the automation of this process have traditionally been procedural in nature, and more recently combinatorial as in some of the applications of Markov chain Monte Carlo (MCMC) in this problem domain. A similar approach, as a dense, global network of association nodes, has been in use at the International Data Center (IDC) of the Comprehensive Nuclear-Test-Ban-Treaty Organization since its inception in the late 1990s.
Introspective Systems Seismic Networks implementation is radically different from those in use by other seismic observatories. The xGraph framework provides a real-time, in tempo approach that streamlines acquisition and analysis without batching of picks or other data. The algorithm is neither procedural nor is it combinatorial as with MCMC based approaches, for example. Instead it is draws from complexity theory where a catalog is created as a guided, self-organized criticality, designed for a cloud-based environment. Figuratively speaking with respect to the common “sand pile” metaphor, the grains of sand are the swarm of arriving picks and other data, and the emergent sand pile is an earthquake catalog where the binding energy is represented as the Bayesian affinity between picks and quakes.