The dynamic response of historical masonry structures involves multiple sources of nonlinearity, arising from the materials used, the ageing, the complex geometries and boundary conditions involved. As a result, modelling the seismic response of these buildings requires detailed instrumentation beforehand. Crossed by active faults and frequently shaken by moderate earthquakes (Mw3-4), the Cusco region (Peru) has many stone and earth masonry buildings that turn out to be particularly vulnerable to the seismic hazard. We therefore conducted an ambient vibration-based survey in the 17th-century church of San Cristobal in Cusco, seriously damaged by the 1950 earthquake. By combining an Operational Modal Analysis, single-sensor monitoring for over a year and free-field microtremor measurements, our work highlights the existence of strong soil-structure interaction and topographic effects resulting in the excitation of a rigid-body-like mode. Continuous instrumentation also made it possible to study the structure's response to earthquakes, revealing an unexpected frequency drop during a Mw4.2 earthquake, followed by a slow recovery process that lasted more than two months. These results shed new light on the seismic vulnerability of the church, and call for further investigation into the processes behind the site effects and nonlinear dynamics that characterise the response of Andean built heritage.

Ambient seismic noise and microseismicity analyses are increasingly applied for the monitoring of landslides and natural hazards. These methodologies can offer a valuable monitoring tool also for glacial and periglacial bodies, to understand the internal processes driven by external modifications in air temperature and rainfall/snowfall regimes and to forecast possible melting-related hazards in the light of climate change adaptation. We applied the methods to an almost continuous year of data recorded by a network of four passive seismic stations deployed in the frontal portion of the Gran Sometta rock glacier (Aosta Valley, NW Italian Alps). The spectral analysis of ambient seismic noise revealed frequency peaks related to stratigraphic resonances inside the rock glacier. Although the resonance frequency related to the bedrock interface was constant over time, a second higher resonance frequency was identified as the effect of variations in the active layer thickness driven by external air temperature modifications at the daily and seasonal scales. Ambient seismic noise cross-correlation highlighted coherent shear wave velocity modifications inside the periglacial body. The microseismicity dataset extracted from the continuous ambient noise recordings was analyzed and clustered to further investigate the ongoing internal processes and gain insight into their source mechanism and location. The first cluster of events was found to be likely related to the basal movements of the rock glacier and to falls and slides of the debris material. The second cluster was possibly related to shallow ice and rock fracturing processes. The validation of the seismic results through simple models of the rock glacier physical and mechanical layering, the internal thermal regime and the surface displacements allowed for a comprehensive understanding of the rock glacier's reaction to the external conditions.

Understanding the slope hydrology and failure processes of rainfall-induced landslides is key to landslide early warning; the heterogeneity of soil (e.g., grain-size distribution in different layers) can markedly affect rainfall infiltration and slope failure patterns. However, the hydrological and failure processes of heterogeneous slopes layered by different soil groups have received little attention. In this study, we use a typical landslide soil composition of rainfall-induced landslide in fault zones as a prototype and via flume experiments to simulate the hydrological evolution, failure processes, and patterns under rainfall conditions on material heterogeneity slopes with a combination of colluvial deposit and fault gouge. Our results showed that rainfall-induced slope settlement and rapid saturation of shallow layers of colluvial deposits led to the occurrence of layer-by-layer shallow flow-slides. The spatial variability of infiltration led to the generation of a relatively dry-wet interface in deeper layers, causing differential changes in the mechanical properties of the fault gouge; this was conducive to the formation of a steep landslide back wall, perched water table in the shallow layer of the fault gouge, and a rapid increase in porewater pressure, which triggered deep sliding, with a change in the failure pattern to a retrogressive mode. There was a strong linear correlation between the displacement rate before slope instability and the Arias intensity (IA) of the seismic signal; an abrupt change and rapid increase in IA may indicate that the slope entered an accelerating creep stage before failure. The results of this study provide a physical basis for related numerical simulation research and a reference for landslide early warning based on seismic signals.

A deep-seated landslide could release numerous microseismic signals from creep-slip movement, which includes a rock-soil slip from the slope surface and a rock-soil shear rupture in the subsurface. Machine learning can effectively enhance the classification of microseismic signals in landslide seismic monitoring and interpret the mechanical processes of landslide motion. In this paper, eight sets of triaxial seismic sensors were deployed inside the deep-seated landslide, Jiuxianping, China, and a large number of microseismic signals related to the slope movement were obtained through 1-year-long continuous monitoring. All the data were passed through the seismic event identification mode, the ratio of the long-time average and short-time average. We selected 11 days of data, manually classified 4131 data into eight categories, and created a microseismic event database. Classical machine learning algorithms and ensemble learning algorithms were tested in this paper. In order to evaluate the seismic event classification performance of each algorithmic model, we evaluated the proposed algorithms through the dimensions of the accuracy, precision, and recall of each model. The validation results demonstrated that the best performing decision tree algorithm among the classical machine learning algorithms had an accuracy of 88.75%, while the ensemble algorithms, including random forest, Gradient Boosting Trees, Extreme Gradient Boosting, and Light Gradient Boosting Machine, had an accuracy range from 93.5% to 94.2% and also achieved better results in the combined evaluation of the precision, recall, and F1 score. The specific classification tests for each microseismic event category showed the same results. The results suggested that the ensemble learning algorithms show better results compared to the classical machine learning algorithms.

Permafrost degradation is rapidly increasing in response to a warming Arctic climate, altering landscapes and damaging critical infrastructure. Solutions for monitoring permafrost thaw dynamics are essential to understand biogeochemical feedbacks as well as to issue warnings for hazardous geotechnical conditions. We investigate the feasibility of permafrost monitoring using permanently installed fiber-optic seismic networks. We conducted a 2-month seismic monitoring campaign during a controlled thaw experiment using a permanent surface orbital vibrator (SOV) and a 2D-array of distributed acoustic sensing (DAS) cables, and observed significant (15%) shear-wave velocity (V-s) reductions and approximately 2 m depression of the permafrost table beneath the heating zone. These observations were validated by time-lapse horizontal-to-vertical spectral ratio (HVSR) analysis from three co-located broadband seismometers. The combination of SOV and DAS provided unique seismic observations for permafrost monitoring at the field scale, as well as a basis for design and development of early warning systems for permafrost thaw.