Comparison of TLS and sUAS point clouds for monitoring embankment dams

Abstract

Monitoring of dams is an essential surveying task to guarantee the safety of operation and understand the physical processes concerning their movement. Point cloud generating technologies are increasingly being utilized for monitoring of engineered structures. This paper compares point clouds acquired from terrestrial laser scanning (TLS) and small unmanned aerial systems (sUAS)-based photogrammetry for monitoring of the Francis E. Walter dam in northeast Pennsylvania. Authorized for construction by the Flood Control Act of 1946, and with renewed interest due to extensive flooding in 1955 caused by the back-to-back hurricanes Connie and Diane, this earth-filled embankment dam was completed in June of 1961 by the U.S. Army Corps of Engineers. It is currently operated in conjunction with Beltzville Lake for stage reductions on the Lehigh River. The dam is being monitored through conventional surveying methods (total station) every five years. In spring of 2021 a TLS and sUAS data acquisition took place to assess the feasibility and utility of using modern point cloud technologies for monitoring. This paper presents a comprehensive comparison and accuracy assessment of the two point cloud collection methods, considering several parameters for the generation of the sUAS photogrammetric point cloud. Results show the advantages and disadvantages of the two methods. For instance, TLS offers high accuracy (cm-level), but suffers from data gaps due to line of sight blockage/occlusion. On the other hand, sUAS photogrammetry offers more complete point clouds, but presents more challenges in georeferencing and in the generation of accurate point clouds. Similar insights and lessons learned are useful for future surveying tasks and monitoring of similar embankment dam structures.