Chaos and Slow Earthquakes Predictability

Date: 4/25/2019

Time: 11:45 AM

Room: Puget Sound

Slow Slip Events (SSEs) play a significant role in the moment budget along subduction megathrust. We use 352 continuous GPS stations spanning the time range from 2007 to 2017 to model the SSEs in the Cascadia region. In particular, we adopt a variational Bayesian Independent Component Analysis decomposition to extract the tectonic signal and perform a linear inversion to model the SSEs on the megathrust. This approach is very effective in identifying, and filtering out, non-tectonic sources of geodetic deformation (e.g. due to surface loads) and other tectonic sources (e.g., afterslip). The results show a clear segmentation with a few major asperities interacting with one another, a behavior that recalls that of a discrete body system. We then use both classical Embedding Theory and Extreme Value Theory applied to the study of dynamical systems to show that, where the Signal to Noise Ratio is sufficiently high, a low-dimensional (< 5) non-linear chaotic system is more appropriate to describe the dynamics than a stochastic system. We calculate major properties of the strange attractor like its correlation and instantaneous dimension, its instantaneous extremal index and a possible range for the metric entropy of the system. This knowledge is important for the determination of the predictability of the system, since it is related to the rate at which two trajectories in the phase space diverge. We further validate our results with a fuzzy inference system model to check the predictability of the slip and slip rate. In conclusion, SSEs in Cascadia can be described as a deterministic, albeit chaotic, system rather than as a random process. As SSEs might be regarded as earthquakes in slow motion, regular earthquakes might be similarly chaotic and predictable.

Presenting Author: Adriano Gualandi

Authors