Historic records of the last 300 years reveal two great interplate earthquakes (1833, MW = 8¾ 1861, MW = 8¼–8½), which ruptured major segments of the Sumatra fore arc in western Indonesia; a significant percentage of interplate slip along this portion of the plate boundary appears to occur seismically. The ends of these rupture zones are coincident with clusters of large (Ms ≥ 7) and moderate (6 ≤ Ms ≤ 7) shocks and with heterogeneities in the plate interface as inferred from geologic and geophysical data. The northern extent of the 1833 rupture zone is coincident with a group of moderate and large earthquakes, a structural arch in the fore arc, and a fracture zone on the subducted plate. The southern portion is coincident with a very prominent cluster and the along‐strike extension of a sediment‐filled trough on the subducted plate. Prior to this study, Sumatra was characterized as relatively aseismic as inferred from the lack of great earthquakes in the instrumental record of this century. Java and the Lesser Sunda Islands had major (Ms ≥ 6) earthquakes in the historic record, but none are of the same magnitude as the great events near Sumatra. There is no clear evidence that shocks in the Java fore arc were interplate events, yet geophysical data indicate recent uplift of the trench slope break and a change in strike of the trench axis where the northernmost flank of the Roo Rise, a prominent bathymetrie high, appears to be interacting with the fore arc. This feature was formed at a ridge crest and is more buoyant, and thus more difficult to subduct, than the surrounding seafloor. There is no increase in seismicity associated with this tectonism, however, that may indicate a lack of contact between the crystalline portions of the plates at this time. We infer that a majority of slip on the plate interface near Java occurs aseismically. The variation in occurrence of great interplate earthquakes along the Sunda Arc may be interpreted in terms of a model that attributes variation in interplate coupling to the age of the subducted lithosphere. Great interplate earthquakes occur near Sumatra, where the age of the youngest crust is 46 Ma. The characteristics of the Sunda Arc, and analogies with Pacific arcs, imply that the entire length of Sumatra has the potential to produce great thrust earthquakes; its seismic potential should be considered high (as there has been no recurrence of the great events of 1833 and 1861). In contrast, no such events have been reported off Java and the Lesser Sunda Islands, where the oldest crust is 152 Ma. The plate interface near Java and the Lesser Sunda Islands should be considered to have low seismic potential.

-The theory of plate tectonics provides a basic framework for evaluating the potential for future great earthquakes to occur along major plate boundaries. Along most of the transform and convergent plate boundaries considered in this paper, the majority of seismic slip occurs during large earthquakes, i.e., those of magnitude 7 or greater. The concepts that rupture zones, as delineated by aftershocks, tend to abut rather than overlap, and large events occur in regions with histories of both long-and short-term seismic quiescence are used in this paper to delineate major seismic gaps.In detail, however, the distribution of large shallow earthquakes along convergent plate margins is not always consistent with a simple model derived from plate tectonics. Certain plate boundaries, for example, appear in the long term to be nearly aseismic with respect to large earthquakes. The identification of specific tectonic regimes, as defined by dip of the inclined seismic zone, the presence or absence of aseismic ridges and seamounts on the downgoing lithospheric plate, the age contrast between the overthrust and underthrust plates, and the presence or absence of back-arc spreading, have led to a refinement in the application of plate tectonic theory to the evaluation of seismic potential.The term seismic gap is taken to refer to any region along an active plate boundary that has not experienced a large thrust or strike-slip earthquake for more than 30 years. A region of high seismic potential is a seismic gap that, for historic or tectonic reasons, is considered likely to produce a large shock during the next few decades. The seismic gap technique provides estimates of the location, size of future events and origin time to within a few tens of years at best.The accompanying map summarizes six categories of seismic potential for major plate boundaries in and around the margins of the Pacific Ocean and the Caribbean, South Sandwich and Sunda (Indonesia) regions for the next few decades. These categories range from what we consider high to low potential for being the site of large earthquakes during that period of time. Categories 1, 2 and 6 define a time-dependent potential based on the amount of time elapsed since the last large earthquake. The remaining categories, 3, 4, and 5, are used for areas that have ambiguous histories for large earthquakes; their seismic potential is inferred from various tectonic criteria. These six categories are meant to be interpreted as forecasts of the location and size of future large shocks and should not be considered to be predictions in which a precise estimate of the time of occurrence is specified.Several of the segments of major plate boundaries that are assigned the highest potential, i.e., category I, are located along continental margins, adjacent to centers of population. Some of them are hundreds of kilometers long. High priority should be given to instrumenting and studying several of these major seismic gaps since many are now poorly instrumented. The categories of potent...

The direction and rate of movement of the Caribbean plate with respect to North America are determined from the slip vectors of shallow earthquakes and from the configuration of downgoing seismic zones in the Greater and Lesser Antilles. A calibration of the relative plate motion for the northeastern Caribbean using data from other subduction zones indicates an average rate of 3.7±0.5 cm/yr for the past 7 million years (Ma). The direction of plate motion inferred from focal mechanisms (ENE) is nearly the same as that deduced from the configuration of downgoing seismic zones going around the major bend in the arc. With respect to North America, the Caribbean plate is moving at an angular velocity of 0.36°/Ma about a center of rotation near 66°N, 132°W. Vector addition using those data and that for the relative motion of North and South America indicates that the Caribbean is moving at an angular velocity of 0.47°/Ma about a center of rotation near 60°N, 88°W with respect to South America. The presence of intermediate‐depth earthquakes beneath Puerto Rico and the Virgin Islands is ascribed to the curvature of the plate boundary and a component of underthrusting that has been going on for at least the past 7 Ma and is likely occurring today. The alternative hypothesis that earthquakes beneath those areas are occurring in materials that were subducted during the Eocene, the last major episode of magmatism, is not tenable from thermal considerations. The lack of recent magmatism in the eastern Greater Antilles is ascribed to the relatively small component of underthrusting. The 2 cm/yr rate of seafloor creation along the mid‐Cayman spreading center for the past 2.4 Ma does not appear to reflect the total Caribbean‐North American plate motion while the 4 cm/yr spreading rate from 6.0 to 2.4 Ma does. Between the mid‐Cayman spreading center and eastern Guatemala, the northern boundary of the Caribbean plate is narrow and follows the southern margin of the Cayman trough. Seismic activity between the spreading center and eastern Hispaniola, however, occurs over a zone about 250 km wide that extends from Cuba to Jamaica and across the entire width of Hispaniola. Individual faults within this broad plate boundary appear to have accommodated differing amounts of motion as a function of geological time while the cumulative plate motion across the zone remained nearly constant. The percentage of total plate motion accommodated near southern Hispaniola and Jamaica is inferred to have increased about 2.4 Ma ago. That change may have been caused by the collision of parts of the Bahama bank and northern Hispaniola. This explanation for the sudden decrease in seafloor creation along the mid‐Cayman spreading center is less catastrophist than the hypothesis that the entire Caribbean plate suddenly changed its velocity with respect to surrounding plates. The Caribbean plate may be regarded as a small buffer plate whose motion is now governed by the movement of the larger North and South American plates which bound it on three sides. Th...

Normal seafloor entering the Puerto Rico and northern Lesser Antillean trenches in the northeastern Caribbean is interrupted by a series of aseismic ridges on the North and South American plates. These topographic features lie close to the expected trend of fracture zones created about 80–110 m.y. ago when this seafloor was formed at the Mid‐Atlantic Ridge. The northernmost of the ridges that interact with the Lesser Antillean subduction zone, the Barracuda Ridge, intersects the arc in a region of high seismic activity. Some of this seismicity including a large shock in 1974, occurs within the overthrust plate and may be related to the deformation of the Caribbean plate as it overrides the ridge. A large bathymetric high, the Main Ridge, is oriented obliquely to the Puerto Rico trench and intersects the subduction zone north of the Virgin Islands in another cluster of seismic activity along the inner wall of the trench. Data from a seismic network in the northeastern Caribbean indicate that this intersection is also characterized by both interpolate and intraplate seismic activity. Magnetic anomalies, bathymetric trends, and the pattern of deformed sediments on the inner wall of the trench strongly suggest that the Main and Barracuda ridges are parts of a formerly continuous aseismic ridge, a segment of which has recently been overridden by the Caribbean plate. Reconstruction of mid‐Miocene to Recent plate motions also suggest that at least two aseismic ridges, and possibly fragments of the Bahama Platform, have interacted with the subduction zone in the northeastern Caribbean. The introduction of these narrow segments of anomalous seafloor into the subduction zone has segmented the arc into elements about 200 km long. These ridges may act as tectonic barriers or asperities during the rupture processes involved in large earthquakes. They also leave a geologic imprint on segments of the arc with which they have interacted. A 50‐km landward jump of the locus of island arc volcanism occurred in Late Miocene time along the northern half of the Lesser Antilles. We postulate that the subduction of a segment of seafloor of anomolously thick crust, being more buoyant than adjacent seafloor, resulted in a marked shoaling in the dip of the descending slab and, therefore, a shift of the locus of volcanism. In the region near western Puerto Rico and eastern Hispanolia, Plio‐Pleistocene interaction with a similar feature, in this case a part of the Bahama Platform, about 3–4 m.y. ago led to a jump in the locus of subduction as evidenced by a gap in the downgoing seismic zone. That segment of the Bahama Platform interferred with the subduction process and was subsequently sutured onto the Caribbean plate when the boundary jumped about 60 km to the northeast. The maximum size of historic shallow earthquakes along the Lesser Antillean arc varies from about 7.0–7.5 in the center of the arc where the dip of the shallow part of the plate boundary is steep to 8.0–8.5 along the northern part of the arc where the dip is shallow. The i...