Digital Linework (1979-2008) Associated With Eastern Denali Fault Surface Trace Map, Eastern Alaska and Adjacent Canada

PHOTOGRAMMETRY -- We used Agisoft PhotoScan photogrammetry software to generate dense point clouds from legacy air photos of the fault trace between the Alaska-Yukon border and Haines Junction, Yukon, and constructed DEMs from the point clouds using Quick Terrain Modeler software. Computer hardware system requirements to run ArcMap, PhotoScan, and Quick Terrain Modeler software are similar. We used a Dell Precision Tower optimized to run PhotoScan with a Dual Intel Xeon processor, AMD FirePro 8GB video card, 500 GB Serial-ATA hard drive, and 16 GB memory. To generate the air photo-based DEMs, we generally followed the PhotoScan tutorial workflow (available by searching www.agisoft.com), however we modified the workflow with the following specific steps. We acquired TIF format scans (300 dpi minimum) of legacy air photos from the US Geological Survey Earth Explorer web portal (earthexplorer.usgs.gov) and from Geomatics Yukon (geomaticsyukon.ca). We tended to process no more than two flight lines (up to 12 photos) at a time in PhotoScan. Prior to aligning the photos in PhotoScan, we used the “smart scissors” tool to mask out fiduciary stamps and borders on the photos, and constrained the alignment by the masked photos. Because PhotoScan relies on pixel similarity to align photos, masking prevents misalignment of the photos based on the recurring frame location and, unlike cropping the imported photos outside of PhotoScan, masking does not remove useful data from the source photo set. Before optimizing the calibration of the reconstructed camera positions, we georeferenced each set of aligned photos. To georeference each photoset, we identified 10-12 matching features visible in both Google Earth and in at least two photos (e.g., very small closed lakes, intersecting roads, prominent bedrock features). We used PhotoScan to create reference points on the features in the air photos, and applied the Google Earth coordinates and altitude of each feature to the corresponding reference point. We also assigned a nominal 4 m uncertainty to reflect the reference point location accuracy that we estimated visually. We exported the georeferenced dense point clouds from PhotoScan as LAS files, and used the adaptive triangulation gridding algorithm in Quick Terrain Modeler to construct 1 m resolution DEMs from the point clouds. Although PhotoScan is independently capable of generating DEMs from the point clouds, we used Quick Terrain Modeler because it provides a range of options for gridding, and allows the user to progressively repair holes between points, which we did over distances up to 12 m. We also used Agisoft PhotoScan to generate registered orthomosaics of the air photos.

MAPPING -- We used ESRI ArcMap (version 10.3.1) software to map the surface trace of the Eastern Denali Fault over a combination of the four moderate (5 m) to high (0.5 m) resolution DEMs. Hillshade and slope layers derived from the DEMs enabled visualization of fault trace features where they occur. We used 0.5 m vertical resolution DEMs and derivatives based on a 2009 EarthScope airborne lidar dataset that covers the central Denali Fault in Alaska to map the ~100 km fault length between the Totschunda-Denali Fault junction and the Chisana River. To map the fault trace along the ~45 km fault length between the Chisana River and the Alaska-Yukon border we used a 5 m vertical resolution DEM and derivatives based on 2010 US Geological Survey interferometric synthetic aperture radar (IfSAR) data. We are aware that TransCanada Corporation acquired a lidar dataset along the fault trace in Yukon for the purpose of understanding the related earthquake hazards, but we were unable to access or utilize this dataset. In place of the Yukon lidar, we used Agisoft PhotoScan photogrammetry software to generate dense point clouds from legacy air photos of the fault trace between the Alaska-Yukon border and Haines Junction, Yukon, and constructed DEMs from the point clouds using Quick Terrain Modeler software (process step described above in "PHOTGRAMMETRY"). We digitized fault trace-related features including scarps, landslides, and sackungen in categories determined by our confidence in the location of each feature similar to the schema used by Haeussler (2009) (Table 1). Scarp features mark the surface trace of the Eastern Denali Fault, which we identified as a range of geomorphic features including linear troughs, hillside benches, sub-vertical (east- to northeast-facing) scarps, and linear sediment mounds prominent on the Yukon section of the fault (Clague, 1979). We map the scarp in close agreement with the air photo and field-based mapping of Clague (1979), and base the location of the scarp beneath Kluane Lake on seismic reflection data shown in Figure 30.5 of Clague (1979). Landslide features outline landslides along the Eastern Denali Fault that may or may not have occurred during past earthquakes. Similarly, sackung features delineate sackungen on ridges along the Eastern Denali Fault that developed either as a result of shaking during past earthquakes or from gravitational failure of the steep bedrock topography.