Automatic Fault Mapping in Remote Optical Images and Topographic Data With Deep Learning

Abstract

Faults form dense, complex multi‐scale networks generally featuring a master fault and myriads of smaller‐scale faults and fractures off its trace, often referred to as damage. Quantification of the architecture of these complex networks is critical to understanding fault and earthquake mechanics. Commonly, faults are mapped manually in the field or from optical images and topographic data through the recognition of the specific curvilinear traces they form at the ground surface. However, manual mapping is time‐consuming, which limits our capacity to produce complete representations and measurements of the fault networks. To overcome this problem, we have adopted a machine learning approach, namely a U‐Net Convolutional Neural Network (CNN), to automate the identification and mapping of fractures and faults in optical images and topographic data. Intentionally, we trained the CNN with a moderate amount of manually created fracture and fault maps of low resolution and basic quality, extracted from one type of optical images (standard camera photographs of the ground surface). Based on a number of performance tests, we select the best performing model, MRef, and demonstrate its capacity to predict fractures and faults accurately in image data of various types and resolutions (ground photographs, drone and satellite images and topographic data). MRef exhibits good generalization capacities, making it a viable tool for fast and accurate mapping of fracture and fault networks in image and topographic data. The MRef model can thus be used to analyze fault organization, geometry, and statistics at various scales, key information to understand fault and earthquake mechanics.

Fault Trace Corrugation and Segmentation as a Measure of Fault Structural Maturity

“…Although the structural maturity is not informed in the figures, data falling in the upper left and lower right parts of the graphs represent immature and mature faults, respectively. Nowadays, deep learning can efficiently assist geologists to map surface fault traces automatically at high resolution and accuracy in optical images of the ground (Mattéo et al, 2021). This provides the opportunity to rapidly map the surface traces of many faults worldwide, and to analyze them as described here to recover the structural maturity of these faults.…”

“…Nowadays, deep learning can efficiently assist geologists to map surface fault traces automatically at high resolution and accuracy in optical images of the ground (Mattéo et al., 2021). This provides the opportunity to rapidly map the surface traces of many faults worldwide, and to analyze them as described here to recover the structural maturity of these faults.…”

Semiautomatic Algorithm to Map Tectonic Faults and Measure Scarp Height from Topography Applied to the Volcanic Tablelands and the Hurricane Fault, Western US

“…The stereosatellite imagery was acquired on August 17, 2017, by the Pléiades satellites. To produce digital surface models (DSMs), we follow the same method as Mattéo et al [22]. Each dataset has a one panchromatic image at a 50 cm resolution and a multispectral image with four bands at a 2 m resolution.…”

“…Given the recent increase in topographic data availability and quality, there have been recent efforts to automatize portions of dip-slip fault mapping and scarp height calculations on the assumption that scarp height indicates the cumulative vertical fault slip [10]. Automatic approaches for mapping strike-and dip-slip faults are often applied to optical imagery and include edge-detection methods (e.g., [18]), "ant-tracking" methods [19], and an increasing variety of deep-learning approaches (e.g., [20][21][22]). Methods for estimating fault height or surface slip often use topography data.…”

Neural Network‐Based CO2 Interpretation From 4D Sleipner Seismic Images

“…Machine learning (ML) provides a rapid automatic option, which comprises a plethora of numerical methods that can learn from available data and make predictions in unseen data (Mattéo et al., 2021). Hence, different from conventional interpretation, ML‐based seismic interpretation, trained by labels from experienced interpreter or realistic model building, can deliver satisfying results with much higher efficiency, e.g., in seismic facies analysis (Wrona et al., 2018), faults identification (Wu et al., 2019), salt bodies delineation (Guillen et al., 2015), and horizon detection (Geng et al., 2020).…”