The Hebgen Lake earthquake occurred on the 18th August (GMT) 1959 in SW Montana, U.S. This region lies within a zone of slow intracontinental extension, at the intersection between the Yellowstone volcanic system and the Intermountain Seismic Belt (Chang et al., 2013;Smith & Sbar, 1974). The Hebgen Lake earthquake was preceded by at least two events with similar magnitude that were dated at 1-3 and 10-14.5 ka by cosmogenic nuclide geochronology (Schwartz et al., 2009;Zreda & Noller, 1998). The 1959 event consisted of two sub-events with magnitude 7.0 and 6.3, separated by a 5-s time interval, and was followed by large aftershocks on the 18th and 19th of August (Doser, 1985). The subsidence associated with the Hebgen Lake earthquake was recorded over a broad area of approximately 1,500 km 2 , and more than 150 km 2 subsided by more than 3.1 m (Myers & Hamilton, 1964).The earthquake produced structurally complex surface ruptures with notable displacements along the Hebgen fault (HF), the Red Canyon fault (RCF), the Kirkwood fault (KF), and the West Fork fault (WFF). These are west-dipping normal faults with a cumulative length of approximately 35.4 km and maximum vertical offsets of 6.1, 5.8, 0.6, and 1.2 m, respectively (Figure 1, Witkind et al., 1962). Despite some geometric complexity, the ruptures generally strike 130° ± 10°, consistent with the fault plane solution from the main shock (Barrientos et al., 1989;Doser, 1985), and dip SW by 50°-85° (Witkind, 1964). Doser (1985) and Ryall (1962) locate the epicenter 15 km northeast of Hebgen Lake, with a depth of approximately 15-25 km. However, uncertainty in the location, and especially the depth of the hypocenter makes it difficult to assess the spatial relationship between the source and the surface ruptures (Doser, 1985).

<p>Earthquakes on normal faults in the continental setting are relatively uncommon. The scarcity of surface-rupturing events underpins an absence of surface displacement measurements. It is a common practice to use surface offset as a proxy to understand the fault structure at depth. Hence, the lack of comprehensive surface data impedes the subsurface reconstruction of seismogenic normal faults and prohibits the thorough assessment of earthquake hazards. To supplement the available surface displacement measurements and to make statistically significant inferences, we apply optical image correlation (OIC) methods to historical images from three large continental normal earthquakes in the western United States (1954 Dixie Valley (M<sub>w</sub> 6.8) - Fairview Peak (M<sub>w</sub> 7.1) earthquake sequence, the 1959 M<sub>w</sub> 7.2 Hebgen Lake earthquake and the 1983 M<sub>w</sub> 6.9 Borah Peak earthquake). The results of this study are displacement maps with three components of deformation from which we extract high-resolution 3-d measurements everywhere along the surface rupture. </p> <p> </p> <p>The high-resolution 3-d data are used to quantify the magnitude and direction of the earthquake-related offset, the percentage of off-fault damage as well as the width of the fault zone. These parameters represent the fault maturity, geometric complexity and subsurface structure of the fault. Our observations confirm behaviours previously observed along strike-slip faults (e.g. magnitude of off-fault deformation is proportional to the rupture complexity). In addition, a comparative assessment of the results from the three study areas demonstrates that features such as excess slip detected close to the fault scarp are not unique and can be found along multiple dip-slip faults. Consequently, this study documents the variation of the quantifiable parameters along the normal faults. It suggests that while some parameters are a universal reflection of the fault characteristics, others vary according to the geology or topography in the area and should not be accepted without further investigation.</p>

<p>Optical image correlation (OIC) is a powerful tool for measuring the 3-D near-field surface displacements produced in large earthquakes. The method compares pre- and post-earthquake orthorectified images; shifts between common pixels in the image pair reflects a 2-D (horizontal) offset. The third dimension (vertical displacement) is calculated by differencing the pre- and post-topography, while accounting for the horizontal displacements. Optical image correlation has a sub-pixel detection capability<!-- Everywhere, not just close to the rupture -->, and can provide information on the displacement field close to fault ruptures (where InSAR typically decorrelates). Small-scale measurements of the distributed damage provide important constraints on the strain distribution within the fault core and the surrounding damage zone, as well as offering insights into the rupture mechanics.</p><p>OIC is frequently applied to the recent earthquakes where the image footprint is large relative to the rupture extents. However, historical ruptures are documented by aerial photographs which cover a relatively small area. This means that many images are needed to cover the rupture area and all pixels in pre-and post-earthquake images which span the rupture are typically affected by the ground displacement. This creates complications for image co-registration, alignment and correlation of the final mosaics.</p><p>To address this problem we developed a workflow that automatically generates a DEM (digital elevation model) and an orthorectified image mosaic. The process uses structure-from-motion (SfM) and stereo-matching approaches, and results in precise and accurate registration between the image pairs.</p><p>We applied this method to the 1959 Hebgen Lake earthquake, SW Montana, U.S. This large (Mw 7.2) intraplate normal event re-activated pre-existing faults north of the Hebgen Lake reservoir and created a complex rupture network. We used 20 pre-earthquake photographs from 1947 and 70 post-earthquake images from 1977 and 1982. The final results show a 3-D displacement localized onto several prominent structures: the Hebgen fault and the Red Canyon fault, consistent with field mapping following the earthquake. A significant vertical offset and a large horizontal NS-component agree well with SW extension on NW-SE-striking normal faults. Additionally, we used fault-perpendicular profiles to explore the along-strike variation in fault displacement and to determine the extent of the off-fault damage.</p><p>This work demonstrates that the application of OIC techniques to historical earthquakes can provide new information relating to the geometry and displacement of fault ruptures, and isolate the last event from the previously accumulated displacements. Additionally, the method we propose offers potential for the characterisation of historical earthquakes in general, and promises to improve our understanding of rupture behaviour through a statistical analysis of many earthquakes.</p>

The Hebgen Lake earthquake occurred on the 18th August (GMT) 1959 in SW Montana, U.S. This region lies within a zone of slow intracontinental extension, at the intersection between the Yellowstone volcanic system and the Intermountain Seismic Belt (Chang et al., 2013;Smith & Sbar, 1974). The Hebgen Lake earthquake was preceded by at least two events with similar magnitude that were dated at 1-3 and 10-14.5 ka by cosmogenic nuclide geochronology (Schwartz et al., 2009;Zreda & Noller, 1998). The 1959 event consisted of two sub-events with magnitude 7.0 and 6.3, separated by a 5-s time interval, and was followed by large aftershocks on the 18th and 19th of August (Doser, 1985). The subsidence associated with the Hebgen Lake earthquake was recorded over a broad area of approximately 1,500 km 2 , and more than 150 km 2 subsided by more than 3.1 m (Myers & Hamilton, 1964).The earthquake produced structurally complex surface ruptures with notable displacements along the Hebgen fault (HF), the Red Canyon fault (RCF), the Kirkwood fault (KF), and the West Fork fault (WFF). These are west-dipping normal faults with a cumulative length of approximately 35.4 km and maximum vertical offsets of 6.1, 5.8, 0.6, and 1.2 m, respectively (Figure 1, Witkind et al., 1962). Despite some geometric complexity, the ruptures generally strike 130° ± 10°, consistent with the fault plane solution from the main shock (Barrientos et al., 1989;Doser, 1985), and dip SW by 50°-85° (Witkind, 1964). Doser (1985) and Ryall (1962) locate the epicenter 15 km northeast of Hebgen Lake, with a depth of approximately 15-25 km. However, uncertainty in the location, and especially the depth of the hypocenter makes it difficult to assess the spatial relationship between the source and the surface ruptures (Doser, 1985).