Date of Graduation


Document Type


Degree Name

Master of Science in Geography (MS)

Degree Level





Jackson Cothren

Committee Member

William F. Limp

Second Committee Member

Mathew Covington


Cave Survey, Photogrammetry, Structure from Motion


Nearly all aspects of karst science and management begin with a map. Yet despite this fact, cave survey is largely conducted in the same archaic way that is has been for years - with a compass, tape measure, and a sketchpad. Traditional cave survey can establish accurate survey lines quickly. However, passage walls, ledges, profiles, and cross-sections are time intensive and ultimately rely on the sketcher’s experience at interpretively hand drawing these features between survey stations.

This project endeavors to experiment with photogrammetry as a method of improving on traditional cave survey, while also avoiding some of the major pitfalls of terrestrial laser scanning. The proposed method allows for the creation of 3D models which capture cave wall geometry, important cave formations, as well as providing the ability to create cross sections anywhere desired. The interactive 3D cave models are produced cheaply, with equipment that can be operated in extremely confined, harsh conditions, by unpaid volunteers with little to no technical training.

While the rapid advancement of photogrammetric software has led to its use in many 3D modeling applications, there is only a sparse body of research examining the use of photogrammetry as a standalone method for surveying caves. The proposed methodology uses a GoPro camera and a 1000 lumen portable floodlight to capture still images down the length of cave passages. The procedure goes against several traditional rules of thumb, both operating in the dark with a moving light source, as well as utilizing a wide angle, fish eye lens, to capture scene information that is not perpendicular to the camera's field of view. Images are later processed into 3D models using Agisoft’s PhotoScan.

Four caves were modeled using the method, with varying levels of success. The best results occurred in dry confined passages, while passages greater than 9 meters (30ft) in width, or those with a great deal of standing water in the floor, produced large holes. An additional experiment occurred in the University of Arkansas utility tunnel.