Digital outcrop modelling using “structure-from- motion” photogrammetry: Acquisition strategies, validation and interpretations to different sedimentary environments

Abstract

Structure from Motion-Multi-View Stereo (SfM-MVS) is a relatively new technique that is being adopted in the analysis of sedimentary systems. The technique is especially applicable to studies which focus on the distribution, geometry and quantification of geological bodies including the analysis of sedimentary forms and tectonic structures. This photogrammetric method allows for generating virtual outcrops (VO) and high-resolution digital terrain models (DTMs). In comparison with more sophisticated and expensive techniques such as airborne LiDAR scanning or terrestrial laser scanning, SfM-MVS is a fast low-cost method which with careful use of ground control points, can produce models which are comparable to other digital survey methods for spatial accuracy. In this study, the effectiveness of SfM-MVS in a series of quantitative case studies in sedimentary and basin analyses is discussed. Focus is placed on how the research questions define the workflow strategy to know the limitations, sources of error and potential utility that the SFM-MVS technique has when a virtual sedimentary system is interpreted. Three case studies of different sedimentary environments, previously analyzed with traditional fieldwork techniques, were selected and re-analyzed. For each case study the objective was to resolve quantitative questions that were not possible to determine without the use of VO or DTMs. Questions include: (i) geometrical restoration of a growth strata succession associated with thrust systems; (ii) stacking patterns of littoral ridges during a relative sea-level fall after the Holocene climatic optimum; and (iii) quantification of lake level fall during the past 44 years of a closed lake basin. The results of the work show that the technique has significant application in the analysis of sedimentary systems. However, since geological interpretations depend on the quality of the models that are generated, planning and construction of the VO or DEMs along with their validation are essential to ensure reliable results.

Tracking Organomineralization Processes from Living Microbial Mats to Fossil Microbialites

“…A field campaign was carried out in 2011, and living microbial mats were identified and sampled for the first time in several ponds of the Maquinchao Riverbed [35]. Successive field campaigns (i.e., in 2015-2018 revealed that the ponds containing microbial mats had dried out completely ( Figure 1A,B), and the water level of Carri Laufquen Grande had undergone fluctuations [36]. However, several mineralized microbial mats were preserved on hard substrates mostly of basaltic composition.…”

“…It is composed nowadays of two main lakes, connected by the ephemeral Maquinchao River, partially sustained by groundwater. Fossil microbialites attest to a former higher water level in the basin [34,36]. Living and mineralizing mats were observed in 2011 in several ponds of the Maquinchao River [35].…”

Unravelling the widening of the earliest Andean northern orogen: Maastrichtian to early Eocene intra‐basinal deformation in the northern Eastern Cordillera of Colombia

“…Digital outcrop models, based on Structure from Motion techniques, were generated from photos collected systematically in terrain (see procedures in Fabuel‐Perez, Hodgetts, & Redfern, 2010 or in Bilmes et al., 2019). These digital models were used to define the geometry of strata and palaeocurrents (Duarte, Bayona, Ramírez, Baquero, & Tabares, 2019; Gómez, 2019), to add mesoscale information to the stratigraphic analysis.…”

Sources of Uncertainty in Remote Stratigraphic Observations

“…Structure-from-Motion (SfM) with Multi-View Stereo (MVS) was originally developed for the computer vision and visual perception collective and became more popular alongside the development of automatic feature matching algorithms (Spetsakis and Aloimonos, 1991;Boufama et al, 1994;Szeliski and Kang, 1994;Bilmes et al, 2019).…”

“…Current SfM workflows consist of automatic feature matching via the scale-invariant feature transform (SIFT) for a set of images that can be taken with different cameras from several perspectives as long as there is substantial overlap between the images (Lowe 1999;Lowe 2004;Bistacchi et al, 2015;Chesley et al, 2017;Nesbit et al, 2018;Bilmes et al, 2019). Successful triangulation of feature coordinates results in a sparse point cloud which also contains information about the camera specifications like geometry, optical information, and spatial orientation (Chesley et al, 2017;Bilmes et al, 2019). In the final stages of the workflow, a colorized dense point cloud is obtained once the tangential and spherical distortion are removed from the mathematically oriented images (Bilmes et al, 2010;Bistacchi et al, 2015;Chesley et al, 2017;Bubniak et al, 2019).…”