Abstract

Over the past few decades, precise satellite positioning measurements have revealed variations in the deformation of the Earth's surface along the South American subduction zone. This variable deformation is indicative of the variable coupling on the interface between the two converging plates. In Chile, the 3 major earthquakes of the early 21 st century (Maule 2010, Iquique 2014, Illapel 2015) occurred in regions previously identified as strongly coupled. This coincidence supports the classic theory of seismic gaps, in which deformation accumulates in certain zones over long periods of time before being released abruptly by an earthquake. It is therefore natural to postulate that major historical earthquakes obey the same rule, and to ask whether the coupled zones of today's earthquakes also correspond to earthquakes of the past. This question comes up against the uncertainties and imprecision, sometimes errors, in our knowledge of past ruptures. The earthquake of November 11, 1922 (Mw 8.5) in the Atacama region of Chile is often described as the second biggest Chilean earthquake of the 20 th century, after Valdivia 1960. In scientific literature, its rupture runs over up to 450 km in length, from 26 degrees S to 30 degrees S. As a result, it seems to have broken two highly coupled segments, Atacama and Cha & ntilde;aral, and crossed a zone of weak coupling, Barranquilla, that were revealed by modern space geodesy. The apparent disparity between the 1922 rupture as described in the existing literature and today's coupling raises an important question: Did the 1922 earthquake, unlike the earthquakes of the 21 st century, not respect the coupling, and then why? Or, on the contrary, could the coupling not be constant and change over time? Here, we show how a careful re-reading of the scientific literature of the time has led us to revise various numbers and change our vision of the 1922 rupture. These revisions lead to map a two-times smaller rupture that appears to coincide much better with the current coupling revealed by modern geodetic measurements. The 1922 earthquake, with a rupture reduced to just 200 km in length, corresponds to the Atacama segment positioned between 28 degrees S and 30 degrees S. On the occasion, we also show how another often neglected earthquake, the December 4, 1918, of magnitude similar to 8, also respects the current segmentation by rupturing the second segment of the area. The 1918 earthquake, with a rupture re-evaluated to 100 km in length, corresponds to the Cha & ntilde;aral segment positioned between 27 degrees S and 26 degrees S. The two segments are well separated by the Barranquilla Low Coupling Zone, probably generated by entry of the Copiap & oacute; ridge in the subduction, precisely at this latitude.