Lavenda, B. H.

The Gutenberg-Richter magnitude-frequency law takes into account the minimum detectable magnitude, and treats aftershocks as if they were independent and identically distributed random events. A new magnitude-frequency relation is proposed which takes into account the magnitude of the main shock, and the degree to which aftershocks depend on the main shock makes them appear clustered. In certain cases, there can be two branches in the order-statistics of aftershock sequences: for energies below threshold, the Pareto law applies and the asymptotic distribution of magnitude is the double-exponential distribution, while energies above threshold follow a one-parameter beta distribution, whose exponent is the cluster dimension, and the asymptotic Gompertz distribution predicts a maximum magnitude. The 1957 Aleutian Islands aftershock sequence exemplifies such dual behavior. A thermodynamics of aftershocks is constructed on the analogy between the non-conservation of the number of aftershocks and that of the particle number in degenerate gases.

A compound Poisson process is used to derive a new shape parameter which can be used to discriminate between large earthquakes and aftershock sequences. Sample exceedance distributions of large earthquakes are fitted to the Pareto tail and the actual distribution of the maximum to the Fréchet distribution, while the sample distribution of aftershocks are fitted to a Beta distribution and the distribution of the minimum to the Weibull distribution for the smallest value. The transition between initial sample distributions and asymptotic extreme value distributions shows that self-similar power laws are transformed into nonscaling exponential distributions so that neither self-similarity nor the Gutenberg-Richter law can be considered universal. The energy-magnitude transformation converts the Fréchet distribution into the Gumbel distribution, originally proposed by Epstein and Lomnitz, and not the Gompertz distribution as in the Lomnitz-Adler and Lomnitz generalization of the Gutenberg-Richter law. Numerical comparison is made with the Lomnitz-Adler and Lomnitz analysis using the same Catalogue of Chinese Earthquakes. An analogy is drawn between large earthquakes and high energy particle physics. A generalized equation of state is used to transform the Gamma density into the order-statistic Fréchet distribution. Earthquaketemperature and volume are determined as functions of the energy. Large insurance claims based on the Pareto distribution, which does not have a right endpoint, show why there cannot be a maximum earthquake energy.