SYSTEMATICS OF DEEP-FOCUS EARTHQUAKES ALONG THE KURIL-KAMCHATKA ARC AND THEIR IMPLICATIONS ON MANTLE DYNAMICS

Subduction along the Kuril-Kamchatka arc has been proposed as a primary example in which subducted lithosphere, plunges steeply into the lower mantle with little deformation. The occurrence of several unusual earthquakes along this arc during the last few years places new constraints on the configuration and the state of strain of the down-going slab. We systematically determined the source parameters of 27 large- to moderate-sized earthquakes (m(b) greater-than-or-equal-to 5.5) that occurred between 1963 and 1990 at depths greater than 200 km along this subduction zone. To our surprise our results show that the configuration and strain due to earthquakes in subducted lithosphere are, highly variable along both the strike and dip of the arc. Approximately 75 km farther went of the deepest earthquakes in the northern portion of the arc, the earthquake of August 14, 1988, occurred at a comparable depth of 636+/-17 km. Thus at least part of the slab has been displaced in a direction perpendicular to the overall strike of the Wadati-Benioff zone. Earthquakes in the central part of the arc delineate shoaling of the subducted slab at a depth of approximately 570 km, reminiscent of the configuration of subducted lithosphere along the Tonga arc. Both observations indicate that the subducted slab is deforming probably in response to resistance encountered near the 670-km discontinuity. For all earthquakes north of 48-degrees-N, focal mechanisms of all events deeper than 200 km show a remarkably consistent pattern of downdip compression (strain). In the Southern Kuril, the Southern Kuril, the Sakhalin Island earthquake (M(o) approximately 8x10(19) N m) of May 12 1990, is one of three large deep-focus earthquakes that have occurred over 150 km farther west of the well-defined Wadati-Benioff zone. The relationship between the source region of these events and the rest of the subducted slab is equivocal due to the lack of background seismicity. Nonetheless, for earthquakes south of 48-degrees-N, net downdip extension dominates between depths of 200-450 km before switching to downdip compression below 540 km. Thus the strain field within the slab is segmented both along strike and dip. If the slab in southern Kuril-Kamchatka can serve as a stress guide down to depths of 615 km, the changeover from extension to compression occurs at an unusually large depth within the transition zone of the upper mantle. Alternatively, the deep-focus events west of the island of Sakhalin could have occurred in a detached piece of lithosphere. As a whole, the configuration and state of strain of the subducted slab along the Kuril-Kamchatka arc are highly variable, close to a microcosm of variations among different deep subduction systems observed on a global scale. In particular, the subducted lithosphere is apparently undergoing considerable deformation as it approaches the base of the upper mantle- Our results do not preclude the possibility of subducted lithosphere ever reaching the lower mantle. Nevertheless, it seems unlikely that the down-going slab along the Kuril-Kamchatka arc penetrates into the lower mantle unhindered.