Statistical significance of the seismic coupling coefficient

Abstract Variations in the seismic coupling coefficient (χ), the ratio of the observed to expected seismic moment release rates, at subduction zones are commonly used to make inferences about the physics of subduction. Taking into account the power-law form of the earthquake frequency-moment relation, it is shown that the observed distribution of χ at the world's subduction zones based on the twentieth century's earthquakes can be matched by a single value of χ ≈ 0.3 if, as is likely the case, the 90-yr observation time is less than or comparable to the repeat time of the largest possible earthquake. It is shown that because our reliable earthquake record is short, global variations in χ, based on seismicity alone, are poorly resolved and cannot be used to distinguish among subduction models.