SUMMARY A high‐quality data set of 3790 earthquakes were simultaneously inverted for hypocentre locations and 3‐D P‐wave velocities in Costa Rica. Tests with synthetic data and resolution estimates derived from the resolution matrix indicate that the velocity model is well constrained in central Costa Rica to a depth of 70 km; northwestern and southeastern Costa Rica are less well resolved owing to a lack of seismic stations and seismicity. Maximum H2O content and seismic wave speeds of mid‐ocean ridge basalt and harzburgite were calculated for metamorphic phase transformations relevant to subduction. Both the 3‐D P‐wave velocity structure and petrological modelling indicate the existence of low‐velocity hydrous oceanic crust in the subducting Cocos Plate beneath central Costa Rica. Intermediate‐depth seismicity correlates well with the predicted locations of hydrous metamorphic rocks, suggesting that dehydration plays a key role in generating intermediate‐depth earthquakes beneath Costa Rica. Wadati–Benioff zone seismicity beneath central Costa Rica shows a remarkable decrease in maximum depth toward southeastern Costa Rica. The presence of asthenosphere beneath southeastern Costa Rica, which entered through a proposed slab window, may explain the shallowing of seismicity due to increased temperatures and associated shallowing of dehydration of the slab. Tomographic images further constrain the existence of deeply subducted seamounts beneath central Costa Rica. Large, low P‐wave velocity areas within the lower crust are imaged beneath the southeasternmost volcanoes in central Costa Rica. These low velocities may represent anomalously hot material or even melt associated with active volcanism in central Costa Rica. Tomographic images and petrological modelling indicate the existence of a shallow, possibly hydrated mantle wedge beneath central Costa Rica.

[1] Tomographic images constrain the existence of a subducted seamount beneath the Gulf of Nicoya, Costa Rica. The subducted seamount is found at a depth of 30 km within the rupture area of the March 25, 1990, Mw = 7.0 Gulf of Nicoya earthquake. The Gulf of Nicoya earthquake was a typical thrust-type subduction earthquake and occurred on a shallow dipping thrust fault parallel or along the boundary between the subducting Cocos plate and the overriding plate. Precise relocation of the mainshock and its aftershocks in a 3-D P-wave velocity model shows that the area of the mainshock rupture is coincident with the imaged subducted seamount. Most of the aftershocks are relocated within or close to the inferred subducted seamount above the subducting oceanic plate. We interpret the subducted seamount as an asperity whose rupture caused the 1990 Gulf of Nicoya earthquake.

No abstract

No abstract

Many seismological studies depend on the accuracy of timing of seismological data. In seismic tomography, travel-time residuals defined as differences between the observed and calculated arrival times of seismic phases are minimized to constrain 3D velocity structure. Inconsistencies and large errors in data sets that result from incorrect station coordinates, errors in the timing acquisition system, errors in the merging procedure, inconsistency in the picking and phase misidentification can also generate travel-time residuals, and because of their systematic nature, these errors cannot be treated as random noise even by exploiting a large number of travel times. While the inverse problem is perfectly set up to deal with random noise, systematic errors lead to significant artifacts in the solution, but may not be detected by a posterior error assessment. For this reason, detecting and removing systematic travel-time errors from data sets before inversion is crucial for seismic tomography studies.We present a methodology based on the use of a minimum 1D model to detect and remove systematic errors in travel-time data by detailed analysis of station delays and observation residuals and apply it to a local earthquake data set from Costa Rica. The determination of the exact nature of detected inconsistencies needs further investigations in each individual case. If the cause of detected systematic errors cannot be determined beyond any doubt and the afflicted data may not be corrected, they must be deleted from the data set. To assess the extent of influence of systematic errors on hypocenter locations and their uncertainties, we present two examples showing the effects of station mislocation.