1. Random kinematics of unbounded earthquake rupture propagation simulated using a cellular model.
- Author
-
Gusev, AA
- Subjects
- *
KINEMATICS , *GEOLOGIC faults , *SPECTRUM analysis , *DYNAMICS , *SEISMOLOGY - Abstract
Spectra of high-frequency (HF) waves radiated by earthquake sources have specific features: above the common corner frequency fc1 they manifest second corner frequency, fc2; and beyond, a plateau in acceleration spectrum. To explain these features, convoluted, 'lacy' geometry of earthquake rupture front was recently proposed. In order to realize such geometry, random space–time functions were used; this simple approach permitted to reproduce both the mentioned spectral properties. However, the random structure of the front was introduced in that earlier study in an a priori manner. Presently, stochastic evolution of a rupture is described in a less formal way, employing a continuous-time cellular model. Each cell of a model geological fault is occupied by an automaton, with three possible states: intact, failing, or broken. A failing cell can 'ignite', after certain delay time Δ t, failures in neighbour cells. Local values of Δ t are fixed in advance and represent a realization of self-similar random 2-D field. A local Δ t value is believed to reflect local resistance to failure: the stronger a cell is, the larger is its Δ t. Through a succession of local failures, a convoluted or even multiply connected rupture front is formed. Viewed at low resolution, such front occupies a certain strip of finite width. Numerically, this width can be described through the rms width parameter, w. As was shown in earlier simulations, there exists a close relationship between w and fc2: fc2∝1/ w. Using this relationship one can verify a stochastic fault model through comparison of the predicted scaling behaviour of fc2 against the observed one. For the relationships of the kind log fck = – βk log M 0 + const, observations give values of β 2 in the range 0.16–0.27, against β 1 ≈ 1/3 for the common corner frequency, fc1. Thus, the behavior of fc1 agrees with the similarity assumption, whereas that of fc2 does not. Through comparison of calculated estimates for β 2 and for rupture velocity with their observed values, realistic ranges of parameters of the developed model were obtained. With appropriate parameter ranges selected, the current model successively reproduces both the observed range of β 2 and the observed range of average earthquake rupture velocity, (0.5–0.8) cS. When a simulated rupture front history is combined with a plausible 2-D field of local stress drop, broad-band time histories can be generated, with verisimilar spectra. These have two corner frequencies and a plateau in acceleration source spectrum. It was found that permissible variation of model parameters can significantly modify, at given M 0 and stress drop, the relative levels of HF radiation from a source; this may give better insight into causes of variability of strong motion amplitudes. [ABSTRACT FROM AUTHOR]
- Published
- 2018
- Full Text
- View/download PDF