Whether subducted oceanic reliefs such as seamounts promote seismic rupture or aseismic slip remains controversial. Here we use swath bathymetry, prestack depth‐migrated multichannel seismic reflection lines, and wide‐angle seismic data collected across the central Ecuador subduction segment to reveal a broad ~55 km × 50 km, ~1.5–2.0 km high, low height‐to‐width ratio, multipeaked, sediment‐bare, shallow subducted oceanic relief. Owing to La Plata Island and the coastline being located, respectively, ~35 km and ~50–60 km from the trench, GPS measurements allow us to demonstrate that the subducted oceanic relief spatially correlates to a shallow, ~80 km × 55 km locked interplate asperity within a dominantly creeping subduction segment. The oceanic relief geometrical anomaly together with its highly jagged topography, the absence of a subduction channel, and a stiff erosive oceanic margin are found to be long‐term geological characteristics associated with the shallow locking of the megathrust. Although the size and level of locking observed at the subducted relief scale could produce an Mw >7+ event, no large earthquakes are known to have happened for several centuries. On the contrary, frequent slow slip events have been recorded since 2010 within the locked patch, and regular seismic swarms have occurred in this area during the last 40 years. These transient processes, together with the rough subducted oceanic topography, suggest that interplate friction might actually be heterogeneous within the locked patch. Additionally, we find that the subducted relief undergoes internal shearing and produces a permanent flexural bulge of the margin, which uplifted La Plata Island.

International audienceShelf promontories exhibit very specific bathymetric features with regards to tsunamis. Because of their submerged cape morphology, a potential tsunami generated seawards of the promontory will exhibit a specific mode of propagation and coastal impact. To identify this peculiar tsunami signature, the Atacames Promontory, Ecuador, was chosen as a case study (another example is the shelf of the Nile delta, Egypt). The area is tectonically very active, hosts earthquakes among the most powerful recorded, as well as areas of slope instabilities that have triggered significant submarine landslides in the past (several cubic kilometres of volume). Both types of events are likely to be tsunamigenic. To examine the tsunami behaviour at the coastal area of the promontory and at its vicinity, we have considered two examples of tsunamigenic landslides of which scars were identified near the base of the continental slope. We also took into consideration two earthquake scenarios that are likely to represent most classes of earthquakes possibly occurring in this area depending on their locations and subsequent tsunami directivity, that is, a sensitivity test investigation. We took two distinct earthquake scenarios which are based on the 1942 and 1958 events that stroke the area. Then we computed their derived tsunamis and analysed their coastal impact. We found that significant tsunamis can be generated by either landslides or earthquakes. However, the maxima of wave amplitude occur offshore (but still above the underwater promontory): the concave-type shape of the bathymetric field often yields a refraction/focusing area that is located on the shelf promontory and not at the coast area of the promontory: the wave propagates first through the focusing area before striking the considered coast. This area may be considered as a sheltered zone. Besides, in the vicinity of the promontory (not exactly concerned by the study), the city of Esmeraldas, is relatively sheltered due to the presence of the underwater canyon at its termination and due to diverging waves