Creating 3D point clouds, digital elevation models, and orthomosaics from historical aerial imagery through structure from motion aided photogrammetry

Detailed and accurate historical data are crucial to characterizing landscape change, whether it is caused by natural phenomena or of anthropogenic origin. Aerial imagery has been periodically collected by government and nongovernment agencies, for a broad range of purposes, since the early 20th century. The acquisition of aerial images involves using standardized equipment with large-format, carefully calibrated cameras and lens assemblies. Historical archives of aerial images preserve a record of the landscape and are invaluable sources of information that predate satellite and lidar. Structure from motion (SfM) aided photogrammetry provides an efficient means for creating digital three-dimensional data from the vast archives of historical aerial imagery.

This report presents techniques, methods, best practices, and a standard workflow for creating point clouds, digital elevation models, and orthomosaic digital scans of overlapping (stereographic) vertical aerial photographs using photogrammetry software. A list of critical issues relevant to SfM-aided photogrammetry of historical aerial imagery is provided, along with potential contingencies and solutions. The workflow was tested in many geographic study areas and with historical imagery datasets of different scales and sources. Three case studies are included. By integrating known SfM methods, these procedures enable the consistent and optimal extraction of historical quantitative data from archives like the U.S. Geological Survey (USGS) Aerial Photo Single Frame archive in the USGS EarthExplorer data portal.