This article presents a novel deep learning method for semi-automated detection of historic mining pits using aerial LiDAR data. The recent emergence of national scale remotely sensed datasets has created the potential to greatly increase the rate of analysis and recording of cultural heritage sites. However, the time and resources required to process these datasets in traditional desktop surveys presents a near insurmountable challenge. The use of artificial intelligence to carry out preliminary processing of vast areas could enable experts to prioritize their prospection focus; however, success so far has been hindered by the lack of large training datasets in this field. This study develops an innovative transfer learning approach, utilizing a deep convolutional neural network initially trained on Lunar LiDAR datasets and reapplied here in an archaeological context. Recall rates of 80% and 83% were obtained on the 0.5 m and 0.25 m resolution datasets respectively, with false positive rates maintained below 20%. These results are state of the art and demonstrate that this model is an efficient, effective tool for semi-automated object detection for this type of archaeological objects. Further tests indicated strong potential for detection of other types of archaeological objects when trained accordingly.

In this research, we present a new approach to define the distribution of block volumes during rockfall simulations. Unmanned aerial vehicles (UAVs) are utilized to generate high-accuracy 3D models of the inaccessible SW flank of the Mount Rava (Italy), to provide improved definition of data gathered from conventional geomechanical surveys and to also denote important changes in the fracture intensity. These changes are likely related to the variation of the bedding thickness and to the presence of fracture corridors in fault damage zones in some areas of the slope. The dataset obtained integrating UAV and conventional surveys is then utilized to create and validate two accurate 3D discrete fracture network models, representative of high and low fracture intensity areas, respectively. From these, the ranges of block volumes characterizing the in situ rock mass are extracted, providing important input for rockfall simulations. Initially, rockfall simulations were performed assuming a uniform block volume variation for each release cell. However, subsequent simulations used a more realistic nonuniform distribution of block volumes, based on the relative block volume frequency extracted from discrete fracture network (DFN) models. The results of the simulations were validated against recent rockfall events and show that it is possible to integrate into rockfall simulations a more realistic relative frequency distribution of block volumes using the results of DFN analyses.

In recent years data acquisition from remote sensing has become readily available to the quarry sector. This study demonstrates how such data may be used to evaluate and back analyse rockfall potential of a legacy slope in a blocky rock mass. Use of data obtained from several aerial LiDAR (Light Detection and Ranging) and photogrammetric campaigns taken over a number of years (2011 to date) provides evidence for potential rockfall evolution from a slope within an active quarry operation in Cornwall, UK. Further investigation, through analysis of point cloud data obtained from terrestrial laser scanning, was undertaken to characterise the orientation of discontinuities present within the rock slope. Aerial and terrestrial LiDAR data were subsequently used for kinematic analysis, production of surface topography models and rockfall trajectory analyses using both 2D and 3D numerical simulations. The results of an Unmanned Aerial Vehicle (UAV)-based 3D photogrammetric analysis enabled the reconstruction of high resolution topography, allowing one to not only determine geometrical properties of the slope surface and geo-mechanical characterisation but provide data for validation of numerical simulations. The analysis undertaken shows the effectiveness of the existing rockfall barrier, while demonstrating how photogrammetric data can be used to inform back analyses of the underlying failure mechanism and investigate potential runout.

This research aims to highlight the importance of adopting a multi-disciplinary approach to understanding the factors controlling large rock avalanches using the Scanno landslide, Italy, as a case study. The study area is the Mount Genzana, Abruzzi Central Apennines, characterized by the regional Difesa-Mount Genzana-Vallone delle Masserie fault zone. The Scanno landslide is famous for its role in the formation of the Scanno Lake. The landslide is characterized by a wide exposed scar, which was interpreted in previous studies as the intersection of high-angle joints and an outcropping bedding plane on which the landslide failed sometime between the Upper Pleistocene and the Holocene. In this study, the Scanno landslide was investigated through the integration of geological, geomechanical and geomorphological surveys. Remote sensing techniques were used to enrich the conventionally gathered datasets, while Geographic Information Systems (GIS) were used to integrate, manage and investigate the data. The results of the authors investigation show that the outcropping landslide scar can be interpreted as a low-angle fault, associated with the Difesa-Mount Genzana-Vallone delle Masserie fault zone, which differs from previous investigations and interpretations of the area. The low-angle fault provides the basal failure surface of the landslide, with two systematic high-angle joint sets acting as lateral release and back scarp surfaces, respectively. In light of these new findings, pre- and post-failure models of the area have been created. The models were generated in GIS by combining LiDAR (Light Detection and Ranging) and geophysics data acquired on the landslide body and through bathymetric survey data of the Scanno Lake. Using the pre- and post-failure models it was possible to estimate the approximate volume of the landslide. Finally, back-analyses using static and dynamic limit equilibrium methods is also used to show the possible influence of medium-to-high magnitude seismic events in triggering the Scanno landslide.

This paper proposes a novel method to incorporate unfavorable orientations of discontinuities into machine learning (ML) landslide prediction by using GIS-based kinematic analysis. Discontinuities, detected from photogrammetric and aerial LiDAR surveys, were included in the assessment of potential rock slope instability through GIS-based kinematic analysis. Results from the kinematic analysis, coupled with several commonly used landslide influencing factors, were adopted as input variables in ML models to predict landslides. In this paper, various ML models, such as random forest (RF), support vector machine (SVM), multilayer perceptron (MLP) and deep learning neural network (DLNN) models were evaluated. Results of two validation methods (confusion matrix and ROC curve) show that the involvement of discontinuity-related variables significantly improved the landslide predictive capability of these four models. Their addition demonstrated a minimum of 6% and 4% increase in the overall prediction accuracy and the area under curve (AUC), respectively. In addition, frequency ratio (FR) analysis showed good consistency between landslide probability that was characterized by FR values and discontinuity-related variables, indicating a high correlation. Both results of model validation and FR analysis highlight that inclusion of discontinuities into ML models can improve landslide prediction accuracy.

The increased accessibility of drone technology and structure from motion 3D scene reconstruction have transformed the approach for mapping inaccessible slopes undergoing active rockfalls and generating virtual outcrop models (VOM). The Poggio Baldi landslide (Central Italy) and its natural laboratory offers the possibility to monitor and characterise the slope to define a workflow for rockfall hazard analysis. In this study, the analysis of multitemporal VOM (2016–2019) informed a rockfall trajectory analysis that was carried out with a physical-characteristic-based GIS model. The rockfall scenarios were reconstructed and then tested based on the remote sensing observations of the rock mass characteristics of both the main scarp and the rockfall fragment inventory deposited on the slope. The highest concentration of trajectory endpoints occurred at the very top of the debris talus, which was constrained by a narrow channel, while longer horizontal travel distances were allowed on the lower portion of the slope. To further improve the understanding of the Poggio Baldi landslide, a time-independent rockfall hazard analysis aiming to define the potential runout associated with several rock block volumetric classes is a critical component to any subsequent risk analysis in similar mountainous settings featuring marly–arenaceous multilayer sedimentary successions and reactivated main landslide scarps.