The slip behaviour of subduction faults and the viscoelastic rheology of Earth’s mantle govern crustal deformation throughout the subduction earthquake cycle. This Ph.D. dissertation presents research results on two topics: (1) coseismic and postseismic slip of the shallowest segment of subduction faults and (2) postseismic deformation following great subduction earthquakes controlled by mantle viscoelasticity. Topic 1: Slip behaviour of the shallowest subduction faults. By modelling high-resolution cross-trench bathymetry surveys before and after the 2011 Mw 9.0 Tohoku-oki earthquake, we determine the magnitude and distribution of coseismic slip over the most near-trench 40 km of the Japan Trench megathrust. The inferred > 60 m average slip and a gentle increase by 5 m towards the trench over this distance indicate moderate degree of net coseismic weakening of the shallow fault. Using near-trench seafloor and sub-seafloor fluid pressure variations as strain indicators in conjunction with land-based geodetic measurements, we determine coseismic-slip and afterslip distributions of the 2012 Mw 7.6 Costa Rica earthquake. Here, trench-breaching slip similar to the Tohoku-oki rupture did not occur during the earthquake, but afterslip extended to the trench axis and reached ~0.7 m over 1.3 years after the earthquake, exhibiting a velocity-strengthening behaviour. These two contrasting examples bracket a possibly wide range of slip behaviour of the shallow megathrust. They help us understand why large tsunamis are generated by some but not all subduction earthquakes. Topic 2: Postseismic deformation following great subduction earthquakes. Due to the asymmetry of megathrust rupture, with the upper plate undergoing greater coseismic tension than the incoming plate, viscoelastic stress relaxation causes the trench and land areas to move in opposite, opposing directions immediately after the earthquake. Seafloor geodetic measurements following the 2011 Tohoku-oki earthquake, modelled in this work, provided the first direct observational evidence for this effect. Systematic modelling studies in this work suggest that such viscoelastic opposing motion should be common to all Mw ≥ 8 subduction earthquakes. As the effect of viscoelastic relaxation decays with time and the effect of fault relocking becomes increasingly dominant, the dividing boundary of the opposing motion continues to migrate away from the rupture area. Comparative studies of ten 8 ≤ Mw ≤ 9.5 subduction earthquakes in this dissertation quantifies the primary role of earthquake size in controlling the “speed” of the evolution of this deformation. Larger earthquakes are followed by longer-lived opposing motion that affects a broader region of the upper plate. / Graduate

Fault planes of large earthquakes incorporate inhomogeneous structures. This can be observed in teleseismic studies through the spatial distribution of slip and seismic moment release caused by the mainshock. Both parameters are often concentrated on patches on the fault plane with much higher values for slip and moment release than their adjacent areas. These patches are called asperities which obviously have a strong influence on the mainshock rupture propagation. Condition and properties of structures in the fault plane area, which are responsible for the evolution of such asperities or their significance on damage distributions of future earthquakes, are still not well understood and subject to recent geo-scientific studies. <br><br> In the presented thesis asperity structures are identified on the fault plane of the M_w=8.0 Antofagasta earthquake in northern Chile which occurred on 30th of July, 1995. It was a thrust-type event in the seismogenic zone between the subducting pacific Nazca plate and the overriding South American plate. In cooperation of the German Task Force for Earthquakes and the CINCA'95 project a network of up to 44 seismic stations was set up to record the aftershock sequence. The seaward extension of the network with 9 OBH stations increased significantly the precision of hypocenter determinations. They were distributed mainly on the fault plane itself around the city of Antofagasta and Mejillones Peninsula. <br><br> The asperity structures were recognized here by the spatial variations of local seismological parameters; at first by the spatial distribution of the seismic b-value on the fault plane, derived from the magnitude-frequency relation of Gutenberg-Richter. The correlation of this b-value map with other parameters like the mainshock source time function, the gravity isostatic residual anomalies, the aftershock radiated seismic energy distribution and the vp/vs ratios from a local earthquake tomograhpy study revealed some ideas about the composition and asperity generating processes. The investigation of 295 aftershock focal mechanism solutions supported the resulting fault plane structure and proposed a 3D similar stress state in the area of the Antofagasta fault plane. / Die Bruchflaeche grosser Erdbeben umfasst inhomogene Strukturen, die bisher hauptsaechlich in teleseismischen Untersuchungen nachgewiesen werden konnten. Haeufig werden begrenzte Bereiche auf einer Bruchflaeche beobachtet, die durch eine starke Konzentration des freigesetzten seismischen Moments und durch grosse Dislokationen gekennzeichnet sind. Diese Bereiche werden als 'asperities' bezeichnet, die offensichtlich starken Einfluss auf den Bruchverlauf des Hauptbebens ausueben. Beschaffenheit und Eigenschaften der Strukturen in einem Herdgebiet, die verantwortlich sind fuer die Bildung solcher 'asperities' und deren eventueller Bedeutung fuer Schadensverteilungen in zukuenftigen Erdbeben, sind Gegenstand aktueller geowissenschaftlischer Untersuchungen. <br><br> In der vorliegenden Arbeit werden 'asperity'-Strukturen auf der Bruchflaeche des M_w=8.0 Antofagasta Erdbebens vom 30. Juli 1995 im Norden Chiles identifiziert. Es handelt sich hierbei um ein typisches Subduktionsbeben mit Aufschiebungscharakter, das in der seismogenen Zone zwischen der abtauchenden pazifischen Nazca-Platte und der ueberschiebenden suedamerikanischen Platte stattfand. Durch die Zusammenarbeit der Deutschen Task Force fuer Erdbeben und dem sich waehrend des Bebens bereits vor Ort befindlichen CINCA '95 Projektgruppe, konnte ein bis zu 44 Stationen umfassendes seismologisches Netzwerk zur Registrierung der Nachbeben errichtet werden. Vor allem die seeseitige Erweiterung des Netzes durch 9 OBH Stationen trug zur hohen Praezision der Hypozentrenbestimmung der Nachbeben bei, die sich hauptsaechlich auf der Bruchflaeche und damit im Kuestenbereich um die Stadt Antofagasta und der noerdlich gelegenen Halbinsel Mejillones verteilten. <br><br> Die 'asperity'-Strukturen konnten mittels raeumlicher Variationen von lokalen seismologischen Parametern erkannt werden; zunaechst durch die Verteilung des seimologischen b-Wertes auf der Bruchflaeche aus der Magnituden-Haeufigkeitsbeziehung von Gutenberg-Richter. Durch die Korrelation dieser Verteilung mit Parametern wie der Momentenrate aus dem Hauptbeben, der isostatischen Restanomalien des Gravitationsfeldes, der Verteilung der abgestrahlten seismischen Energie durch die Nachbeben und der vp/vs-Verhaeltnisse aus einer lokalen Erdbebentomographie konnten Rueckschluesse auf die Beschaffenheit und damit den Bildungsprozess der asperities gezogen werden. Die Untersuchung der Herflaechenloesungen die fuer 295 Nachbeben bestimmt wurden, ergab eine indirekte Bestaetigung der gefundenen Strukturen und wies auf die Existenz eines 3D Spannungszustands im Bereich der Bruchflaeche des Antofagasta Bebens hin.