Landslides are the main secondary effects of earthquakes in mountainous areas. The spatial distribution of these landslides is controlled by the seismic ground motion and the local slope stability. While gravitational instabilities in arid and semi-arid environments are understudied, we document the landslides triggered by the Sarpol-Zahab earthquake (November 12, 2017, Mw7.3, Iran/Iraq border), the largest event ever recorded in the semi-arid Zagros Mountains. An original earthquake-induced landslide inventory was derived, encompassing landslides of various sizes and velocities (from rapid disrupted rockfalls to slow-moving coherent landslides). This inventory confirms the low level of triggered landslides in semi-arid environments. It also displays clear differences in the spatial and volumetric distributions of earthquake-induced landslides, having a 386 rockfalls of limited size triggered around the epicenter, and 9 giant (areas of ca. 10 6 m 2 ) active and ancient rockslides coseismically accelerated at locations up to 180 km from the epicenter. This unusual distant triggering is discussed and interpreted as an interaction between the earthquake source properties and the local geological conditions, emphasizing the key role of seismic ground motion variability at short spatial scales in triggering landslides. Finally, the study documents the kinematics of slow-moving ancient landslides accelerated by earthquakes, and opens up new perspectives for studying landslide triggering over different time-scales.shaking, generating fewer shallow disrupted slides in soils and colluvial deposits, which usually predominate during earthquakes in wet conditions (Harp and Jibson, 1995;Keefer, 2002). Arid and semi-arid regions are, however, of great interest for the study of factors controlling earthquake-triggered landslides, because the interfering effect of rainfall occurs little or not at all and site effects should be limited.This study analyzes the distribution of landslides triggered by the Sarpol-Zahab earthquake (Mw7.3, Iran) that occurred on November 12th 2017 in the semi-arid (mean precipitation of 230 mm/yr) northwest region of the Zagros Mountains (Fig. 1). This major earthquake, associated with the rupture of a low angle blind thrust fault at depths of 15-20 km (Gombert et al., 2019), occurred at the end of the dry season in an area that encompasses many giant paleo-landslides, of volumes up to 30 km 3 (Ghazipour and Simpson, 2016). Following this earthquake, a few coseismic landslides of various types (debris fall, boulder/rock fall) were reported near the epicenter (Miyajima et al., 2018;Vajedian et al., 2018). The earthquake also reactivated the giant Mela-Kabod landslide (4-km-long, 1-km-wide) located ~40 km south of the epicenter (Vajedian et al., 2018;Goorabi, 2020). To our knowledge, no reactivation was reported at other giant landslides mapped by Ghazipour and Simpson (2016) in the region.Accordingly, this study, which is based on an exhaustive inventory of the induced landslides, aims at understanding ...

<p>                Zagros Mountains form a seismically active fold and thrust belt in western Iran. In addition to the high levels of seismicity, slope failures are common throughout the region, where historical records of very large landslides (> 30 km<sup>3</sup>) are documented. On the November 12<sup>th</sup> 2017, the largest earthquake (Mw 7.3) ever recorded in the Zagros occurred near the town of Sarpol-Zahab (NW Zagros/Iraq border). Following the earthquake, only one large co-seismic rockslide and some small rockfalls were documented near the epicenter. This rather small landslide activity for such a large earthquake raises the question of both the observation completeness and the controlling factors of the landslide triggering in this arid mountainous environment.</p><p>            We conducted an original inventory mapping of the landslides induced by this event along 200 km of the Iran-Iraq border. The landslides were detected by different methods: the scars of rapid co-seismic landslides were mapped using a comparison of pre- and post-seismic Planetlab images (3 m resolution), whereas slow-moving landslides (cm/yr-m/yr) were detected by deriving time-series of ground deformation from radar and optical satellite images. Interferometric measurements were constructed for 3 ascending and descending Sentinel-1 SAR tracks, over a time period of 15 months (spanning 6 months before and 9 months after the main shock), allowing the detection and monitoring of very-slow-moving landslides (cm/yr), while slow-moving landslides of higher velocities (m/yr) were detected from correlation of pre and post-earthquake optical satellite images (Planet and SPOT67 imagery; 3 m and 1.5 m resolution, respectively), orthorectified over precise DEMs.</p><p>            We detected 8 giant rotational rockslides (3.10<sup>6 </sup>to 3.10<sup>7</sup> m<sup>2</sup>) and 360 small rockfalls (2.10<sup>2</sup> to 2.10<sup>4</sup> m<sup>2</sup>) in our study area. The small slope-failures were concentrated in the steepest areas around the epicenter (within a radius of 45 km) while the giant ones were situated in far fields (150 km far from the epicenter). Geomorphological analysis of the giant landslides revealed the reactivation of huge masses with several hundreds meters scarps at their top and runout distance of several hundreds meters, advancing over a river at their toe. The geodetical analysis of these giant landslides, show their co-seismic acceleration by few cm.  Furthermore, the analysis of the displacement time-series of these giant rockslides shows that four of them are destabilized over the longer term. This observation raises question both of the risk posed by these rockslides and the controlling factors of their initiation. A geological and seismological analysis suggests that the triggering of these giant rockslides can be controlled by the geological structure (stratigraphy and folding) and the resulting topography, as well as by the fault mechanism of major earthquakes. Finally, the landslide reactivation mechanism during the Sarpol-Zahab earthquake is discussed.</p>