Triggering quakes in a geothermal space: Enhanced geothermal systems (EGSs) provide a potentially clean and abundant energy source. However, two magnitude-5 earthquakes recently occurred in South Korea during EGS site development. Grigoli et al. and Kim et al. present seismic and geophysical evidence that may implicate the second of these earthquakes, which occurred in Pohang, as an induced event. The combination of data from a local seismometer network, well logs, satellite observations, teleseismic waveform analysis, and stress modeling leads to the assessment that the earthquake was probably or almost certainly anthropogenically induced. The possibility remains that the earthquake occurred coincidentally at the EGS site location, but the aftershock distribution and other lines of evidence are concerning for future development of this geothermal resource. Science , this issue p. 1003 , p. 1007

Mapping landslides using automated methods is a challenging task, which is still largely done using human efforts. Today, the availability of high-resolution EO data products is increasing exponentially, and one of the targets is to exploit this data source for the rapid generation of landslide inventory. Conventional methods like pixel-based and object-based machine learning strategies have been studied extensively in the last decade. In addition, recent advances in CNN (convolutional neural network), a type of deep-learning method, has been widely successful in extracting information from images and have outperformed other conventional learning methods. In the last few years, there have been only a few attempts to adapt CNN for landslide mapping. In this study, we introduce a modified U-Net model for semantic segmentation of landslides at a regional scale from EO data using ResNet34 blocks for feature extraction. We also compare this with conventional pixel-based and object-based methods. The experiment was done in Douglas County, a study area selected in the south of Portland in Oregon, USA, and landslide inventory extracted from SLIDO (Statewide Landslide Information Database of Oregon) was considered as the ground truth. Landslide mapping is an imbalanced learning problem with very limited availability of training data. Our network was trained on a combination of focal Tversky loss and cross-entropy loss functions using augmented image tiles sampled from a selected training area. The deep-learning method was observed to have a better performance than the conventional methods with an MCC (Matthews correlation coefficient) score of $0.495$ and a POD (probability of detection) rate of $0.72$ .

We exploit the amplitude information of a sequence of synthetic aperture radar (SAR) images, acquired at different times, in order to generate displacement time-series in areas characterized by large and/or rapid deformation, the size of which is on the order of the image's pixel dimensions. We follow the same rationale of the Small BAseline Subset (SBAS) differential SAR interferometry (DInSAR) approach, by coupling the available SAR images into pairs characterized by a small separation between the acquisition orbits. We exploit the amplitudes of the selected image pairs in order to calculate the relative across-track (range) and along-track (azimuth) pixel-offsets (PO). Finally, we apply the SBAS inversion strategy to retrieve the range and azimuth displacement time-series. This approach, referred to as pixel-offset (PO-) SBAS technique, has been applied to a set of 25 ENVISAT SAR observations of the Sierra Negra caldera, Galápagos Islands, spanning the 2003-2007 time interval. The retrieved deformation time-series show the capability of the technique to detect and measure the large displacements affecting the inner part of the caldera that, in correspondence to the October 2005 eruption, reached several meters. Moreover, by comparing the PO-SBAS results to continuous GPS measurements, we estimate that the accuracy of the PO-SBAS time-series is on the order of 1/30th of a pixel for both range and azimuth directions.

Despite landslides impact the society worldwide every day, landslide information is inhomogeneous and lacking. When landslides occur in remote areas or where the availability of optical images is rare due to cloud persistence, they might remain unknown, or unnoticed for long time, preventing studies and hampering civil protection operations. The unprecedented availability of SAR C-band images provided by the Sentinel-1 constellation offers the opportunity to propose new solutions to detect landslides events. In this work, we perform a systematic assessment of Sentinel-1 SAR C-band images acquired before and after known events. We present the results of a pilot study on 32 worldwide cases of rapid landslides entailing different types, sizes, slope expositions, as well as pre-existing land cover, triggering factors and climatic regimes. Results show that in about eighty-four percent of the cases, changes caused by landslides on SAR amplitudes are unambiguous, whereas only in about thirteen percent of the cases there is no evidence. On the other hand, the signal does not allow for a systematic use to produce inventories because only in 8 cases, a delineation of the landslide borders (i.e., mapping) can be manually attempted. In a few cases, cascade multi-hazard (e.g., floods caused by landslides) and evidences of extreme triggering factors (e.g., strong earthquakes or very rapid snow melting) were detected. The method promises to increase the availability of information on landslides at different spatial and temporal scales with benefits for event magnitude assessment during weather-related emergencies, model tuning, and landslide forecast model validation, in particular when accurate mapping is not required.

Rapid mapping of event landslides is crucial to identify the areas affected by damages as well as for effective disaster response. Traditionally, such maps are generated with visual interpretation of remote sensing imagery (manned/unmanned airborne systems or spaceborne sensors) and/or using pixel-based and object-based methods exploiting data-intensive machine learning algorithms. Recent works have explored the use of convolutional neural networks (CNN), a deep learning algorithm, for mapping landslides from remote sensing data. These methods follow a standard supervised learning workflow that involves training a model using a landslide inventory covering a relatively small area. The trained model is then used to predict landslides in the surrounding regions. Here, we propose a new strategy, i.e., a progressive CNN training relying on combined inventories to build a generalized model that can be applied directly to a new, unexplored area. We first prove the effectiveness of CNNs by training and validating on event landslides inventories in four regions after earthquakes and/or extreme meteorological events. Next, we use the trained CNNs to map landslides triggered by new events spread across different geographic regions. We found that CNNs trained on a combination of inventories have a better generalization performance, with a bias towards high precision and low recall scores. In our tests, the combined training model achieved the highest (Matthews correlation coefficient) MCC score of 0.69 when mapping landslides in new unseen regions. The mapping was done on images from different optical sensors, resampled to a spatial resolution of 6 m, 10 m, and 30 m. Despite a slightly reduced performance, the main advantage of combined training is to overcome the requirement of a local inventory for training a new deep learning model. This implementation can facilitate automated pipelines providing fast response for the generation of landslide maps in the post-disaster phase. In this study, the study areas were selected from seismically active zones with a high hydrological hazard distribution and vegetation coverage. Hence, future works should also include regions from less vegetated geographic locations.