Avalanche prediction in Self-organized systems

Abstract

It is a common belief that power-law distributed avalanches are inherently unpredictable. This idea affects phenomena as diverse as evolution, earthquakes, superconducting vortices, stock markets, etc; from atomic to social scales. It mainly comes from the concept of ``Self-organized criticality" (SOC), where criticality is interpreted in the way that at any moment, any small avalanche can eventually cascade into a large event. Nevertheless, this work demonstrates experimentally the possibility of avalanche prediction in the classical paradigm of SOC: a sandpile. By knowing the position of every grain in a two-dimensional pile, avalanches of moving grains follow a distinct power-law distribution. Large avalanches, although uncorrelated, are preceded by continuous, detectable variations in the internal structure of the pile that are monitored in order to achieve prediction.