GOCE derived vertical gravity gradient delineates great earthquake rupture zones along the Chilean margin

In the south Central Andes region, the Nazca oceanic plate that subducts beneath the South American plate is characterized by a rough topography derived from different oceanic features that collide against the continental margin. These features determine an important segmentation of both the margin and of the interplate zone. The Chilean subduction margin has been characterized by megathrust earthquakes affecting the plate interface with large rupture areas reaching hundreds of kilometers parallel to the trench. The occurrence of these phenomena has been linked, among other causes, to the subduction of sediments that fill the trench and their spatial relation to the relatively prominent oceanic features. We calculated the topography corrected vertical gravity gradient from GOCE satellite data and from EGM2008 model in order to delineate mass heterogeneities related to density variations along the south-central Chile subduction zone. Obtained results show a spatial relation between the subduction of the Nazca oceanic highs and associated along-strike segmentation of the vertical gravity gradients over the interplate zone. We compared our results with the different rupture areas and found a good correspondence with the ellipses for the main earthquakes such as the Valdivia-1960 and Maule-2010 ones. Then, we compared vertical gravity gradients with slip distribution obtained from different models, finding that they are actually correlated with high slip over negative vertical gradient. The GOCE derived gradient adjusts better with the main slip distribution contour since its signal has a characteristic high wavelength. Instead, the EGM2008 model presents a better performance in defining the high frequency anomalies. However, the last results need to be considered only in those regions where the statistical comparison with GOCE data shows a good performance. This is because EGM2008 model data present varying quality of the original terrestrial data, while the quality of the GOCE data is locally homogeneous.