Fixed recurrence and slip models better predict earthquake behavior than the time‐ and slip‐predictable models: 2. Laboratory earthquakes

Abstract

The behavior of individual stick‐slip events observed in three different laboratory experimental configurations is better explained by a “memoryless” earthquake model with fixed inter‐event time or fixed slip than it is by the time‐ and slip‐predictable models for earthquake occurrence. We make similar findings in the companion manuscript for the behavior of natural repeating earthquakes. Taken together, these results allow us to conclude that the predictions of a characteristic earthquake model that assumes either fixed slip or fixed recurrence interval should be preferred to the predictions of the time‐ and slip‐predictable models for all earthquakes. Given that the fixed slip and recurrence models are the preferred models for all of the experiments we examine, we infer that in an event‐to‐event sense the elastic rebound model underlying the time‐ and slip‐predictable models does not explain earthquake behavior. This does not indicate that the elastic rebound model should be rejected in a long‐term‐sense, but it should be rejected for short‐term predictions. The time‐ and slip‐predictable models likely offer worse predictions of earthquake behavior because they rely on assumptions that are too simple to explain the behavior of earthquakes. Specifically, the time‐predictable model assumes a constant failure threshold and the slip‐predictable model assumes that there is a constant minimum stress. There is experimental and field evidence that these assumptions are not valid for all earthquakes.