3D model of a box-type structure under a small cairn near the Bear Trap in Northwest Greenland

• Main Authors: Carlson, Daniel F., Walsh, Matthew J., Tejsner, Pelle, Thomsen, Steffen
• Format: Dataset
• Language: English
• Published: 2020
• Subjects: Greenland; Archaeology; Structure from Motion Photogrammetry; 3D model; Arctic; Cairn; McGee; Nuussuaq; Port Refuge; Skraeling Island; arctic small tool tradition; greenlandic; kalaallisut
• Online Access: https://zenodo.org/record/4011977; https://doi.org/10.5281/zenodo.4011977

This dataset consists of files that can be used to view a high-resolution 3D model of a box-type structure under a small cairn in the vicinity of ‘The Bear Trap’, a Norse ruin at the western end of the Nuussuaq Peninsula in NW Greenland (also called ‘Bjørnefælden’ in Danish, and ‘Putdlagssuaq’ or ‘The Great Trap’ Greenlandic Kalaallisut). The interior of the stone box appears to have been completely empty. Similar small stone box structures have been identified and discussed by Schedermann (e.g. 1990: 159) for Arctic Small Tool tradition (ASTt) sites on Skraeling Island and by McGee (1979) for Port Refuge in the Canadian High Arctic. Similar and equally enigmatic features have also been described by Knuth (1966/67: 203) for far north and northeast Greenland. They have been alternatively attributed to numerous PalaeoEskimo cultural complexes, but without dated materials from the site current attribution of the feature’s function, significance or date are not possible. The 3D model was created from 280 digital photographs that were processed using Structure from Motion Multiview Stereo photogrammetry software (in this case Agisoft PhotoScan Pro v1.4; Linux Ubuntu). A 24 megapixel Sony a6000 APS-C mirrorless camera fitted with a 17 mm lens was used to acquire ground-level imagery of the structure. Pictures were visually assessed prior to processing in PhotoScan to exclude any that were blurry and/or overexposed. The image alignment or bundle adjustment was performed using ‘High’ accuracy, a key point limit of 60000, no tie point limit, and generic preselection. After the sparse point cloud was created, the bounding box was adjusted to enclose the area of interest and the camera parameters were optimized. Gradual selection was used to remove tie points with relatively high uncertainties. The sparse point cloud was scaled using markers with known dimensions that were placed in the area of interest, and which remained stationary throughout the entire photo survey. The dense point cloud was computed using the ‘Medium’ ...