GEOPHYSICAL RESEARCH LETTERS, VOL. 9, NO. 4, PAGES 393-396, APRIL 1982 A MECHANICAL MODEL OF PRECURSORY SOURCE PROCESSES FOR SOME LARGE EARTHQUAKES Renata Dmowska* Center for Earth and Planetary Physics and Division of Applied Sciences, Harvard University, Cambridge, Massachusetts 02138 Victor C. Li Department of Civil Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 Abstract. A mechanical model is presented of precursory source processes for some large earth- quakes along plate boundaries. It is assumed that the pre-seismic period consists of the upward progression of a zone of slip from lower portions of the lithosphere towards the Earth's surface. The slip front is blocked by local asperities of different size and strength; these asperities may be zones of real alteration of inherent strength, or instead may be zones which are currently stronger due to a local slowdown of a basically rate-dependent frictional response. Such blocking by a single, large asperity, or array of asperi- ties, produces quiescence over a segment of plate boundary, until gradual increase of the stress concentration forces the slip zone through the blocked region at one end of the gap, thus nucle- ating a seismic rupture that propagates upwards and towards the other end. This model is proposed to explain certain distinctive seismicity pat- terns that have been observed before large earth- quakes, notably quiescence over the gap zone fol- lowed by clustering at its ends prior to the main event. A discussion of mechanical factors influ- encing the process is presented and some intro- ductory modelling, performed with the use of a generalized El%asser model for lithospheric plates and the "line spring" model for part- through flaws (slip zones) at plate boundaries, is outlined briefly. Introduct ion Spatio-temporal variations of seismicity be- fore major earthquakes have been studied recent- ly by many investigators with the general aim of earthquake prediction. Results of these studies reveal different modes of seismicity, associated with particular tectonic setting, and, although these seismic patterns differ with place and time and do not repeat themselves perfectly within each seismic cycle, some general features of seismicity before large earthquakes have been es- tablished (Kelleher and Savino (1975), McCann et al. (1979), Kanamori (1981) and many others). A pattern occurring repeatedly consists of back- ground seismicity, followed by seismic quiescence along some part of the plate boundary, accompa- nied by an increase of seismicity at the ends of * On leave from the Institute of Geophysics, Polish Academy of Sciences, Warsaw, Poland. Copyright 1982 by the American Geophysical Union. Paper number 2L0203. 0094-8276/82/002L-0203 $3.00 393 the gap zone by comparison to the surroundings (giving a so-called doughnut pattern, see Mogi (1969, 1979), Kanamori (1981)), and by clusters of activity happening relatively close in time to the main event, in the region of the gap and usu- ally close to its ends. Few as they are, investi- gations of cluster seismicity suggest a differ- ence of seismic spectra for earthquakes belonging to a cluster in comparison with background activ- ity (Ishida and Kanamori (1977, 1980)). Also, some evidence of high stress-drop earthquakes generated in a cluster series (House and Boatwright (1980)) suggests that they may occur in regions of elevated stress level. The main event (large earthquake, with rupture length of order 200 to 600 km) usually starts near one end of the gap zone and propagates into the quiet re- gion, the motion usually being stopped close to the other end (see e.g., Kelleher et al. (1973) and Kelleher and Savino (1975)). This general se- quence is what we propose as being explainable in terms of the model presented here. It must be noted, however, that some seismici- ty gaps are not closed by a big earthquake, and also that there are clusters of earthquakes not followed by a large event. Furthermore, identifi- cation and classification of seismicity patterns depend upon the catalog used for the study, mag- nitude threshold and judgement of investigators, so any seismicity pattern should be treated not as a unique feature of particular earthquake or region, but rather as a manifestation of physi- cal processes occurring in the zone investigated. In general, two different approaches to the mechanics of rupturing plate boundaries have been proposed. In the first approach the notion of an asperity, namely, an area representing higher me- chanical strength than the surrounding region (Byerlee (1970), Scholz and Engelder (1976), Kanamori (1981), Lay and Kanamori (1981)) is used with interaction between asperities being includ- ed (Lay and Kanamori (1981)). The plate boundary ruptures by subsequent failures of asperities un- der growing tectonic stress, with each asperity failure adding supplementary stress to adjacent regions. In the second approach, the rupture of a plate boundary is viewed as a process starting at the bottom of the lithosphere and propagating up- wards (Turcotte and Spence (1974), Savage (1975), Stuart (1979), Prescott and Nur (1980), Li and Rice (1980), Nur (1981)), until instability in the form of a large earthquake occurs. In our model we combine these two approaches as sug- gested by Li and Dmowska (1981); also, we present' some introductory modelling of such a process us- ing a formalism for modelling advance of the slip front developed by Li and Rice (1980)(see also Li (1981)).