Agriculture tillage is the process of ripping the soil to prepare it for planting. This is the most common management method in U.S. agriculture. Remote sensing contributes to the monitoring of topographical and land morphological changes. A new paper in the Journal of Surveying Engineering, “Comparison of sUAS Photogrammetry and TLS for Detecting Changes in Soil Surface Elevations Following Deep Tillage,” assesses the effectiveness of small unmanned aerial systems and terrestrial laser scanning to estimate changes in soil surface elevations after deep tillage treatment.
Authors D. Bolkas, B. Naberezny, and M. G. Jacobson used point cloud data from a mine reclamation project in Brisbin, Pennsylvania that employed deep tillage methods to improve soil structure. Practitioners considering using multiepoch and multiplatform point clouds for similar change estimation projects will benefit from this research. Learn more in the abstract below or read the full paper in the ASCE Library.
Abstract
Agricultural and forestry operations significantly affect land morphology. Monitoring topographical and morphological changes is important in agriculture, forestry geomorphology, and soil sciences for improving land management. Tillage is one of the most common land management methods, which aims at loosening the soil and optimizing edaphological conditions. This paper used point cloud data from a mine reclamation project in Brisbin, Pennsylvania that employed deep tillage methods to improve soil structure. In recent years, point cloud technologies such as small unmanned aerial system (sUAS) photogrammetry often has been used to support agricultural and forestry operations, and to provide detail estimation of land morphological changes. This paper assessed the effectiveness of sUAS and terrestrial laser scanning (TLS) to estimate changes in soil surface elevations after deep tillage treatment. A traditional survey with total stations was used as a reference to assess the performance of each method and gain insights. Several parameters that affect the accuracy of such multiepoch surveys were examined, including georeferencing, sUAS camera self-calibration, sUAS software, ground classification methods, the effect of vegetation on point cloud accuracy, the effect of distance from scanner on TLS point cloud accuracy, and merging scenarios to utilize the advantages of both sensors. Results indicated that in areas with low vegetation, both methods can provide reliable land surface estimation; however, in areas with dense and high vegetation, the two methods had considerable vegetation penetration issues. In TLS surveys, the error increased with increasing distance from a scanner setup. TLS achieved higher accuracy than sUAS surveys up to 5 m from a scanner setup, whereas at distances between 5 and 10 m from a scanner setup, accuracy was comparable for the two methods, and beyond 10 m from a scanner setup sUAS achieved better accuracy than TLS. Furthermore, this paper investigated the benefit of using a prior camera self-calibration, estimated on the same site from the first sUAS data set before tillage, versus estimating a new self-calibration for the second sUAS data sets after tillage. Results showed that a prior self-calibration is essential when the number of ground control points (GCPs) is low, i.e., fewer than four GCPs when Global Navigation Satellite System real-time kinematic (GNSS-RTK) is available, and fewer than eight GCPs when GNSS-RTK is not available. Insights gained from this study can assist in improving surveying planning, and they are important to surveyors and practitioners who are employed in agricultural and forestry applications.
Read the full paper in the ASCE Library: https://doi.org/10.1061/(ASCE)SU.1943-5428.0000346
