During an earthquake sequence, there are often multiple recurring landslides. Understanding the spatial distribution of the landslides triggered by the first earthquake can help us predict the landslide susceptibility for subsequent shakes over a short term. This study used two landslide inventories from the Lombok earthquake sequence in Indonesia in 2018 to construct a short-term secondary disaster prediction model and an overall spatial prediction model using four machine learning algorithms. The average accuracy of the positive samples predicted by the prediction model was 7.1% lower than that of the short-term model. The highest accuracy of the overall prediction model was 14.9% higher, on average, and the area under the ROC curve (AUC) score was 8.1% higher, on average, but the corresponding probability thresholds were lower. The reason for this difference is that, in the short-term prediction model, since most of the landslides in the first landslide inventory were prone to fail two or more times due to the effect of multiple earthquakes, the prediction results have a high positive rate. This feature of the short-term prediction model makes it suitable for landslide rescue guidance in a sequence of earthquakes. In contrast, the overall prediction model can better represent the spatial distribution of the earthquake-triggered landslides in the area.

Rapid and accurate characterization of earthquake sources is crucial for mitigating seismic hazards. In this study, based on 18000 scenario ruptures ranging from M_w 6.4 to M_w 8.3 and corresponding synthetic high-rate Global Navigation Satellite System (HR-GNSS) waveforms, we developed a multi-branch neural network framework, the Continental Large Earthquake Agile Respond (CLEAR), to simultaneously determine the magnitude and slip distributions. We apply CLEAR to recent large strike-slip events, including the 2021 M_w 7.4 Maduo earthquake and the 2023 M_w 7.8 and M_w 7.6 Turkey doublet. The model generally estimates the magnitudes successfully at 32 s with errors of less than 0.15, and predicts the slip distributions acceptably at 64 s, requiring only approximately 30 ms on a single CPU. With optimal azimuthal coverage of stations, the system is relatively robust to the number of stations and the time length of the received data.

Operational landslide susceptibility mapping (LSM) remains challenging in regions with pronounced geo-environmental heterogeneity, where single global models often overlook spatially variable landslide-environment relationships. Northern Guangdong, China, is a typical humid mountainous region where steep terrain, diverse lithology, and highly variable rainfall produce non-stationary landslide controls. To address this challenge, we develop a cluster-informed LSM framework that integrates unsupervised consensus K-means sub-zoning with localized Random Forest (RF) models and SHapley Additive exPlanations (SHAP). We use a harmonized inventory of 1510 landslides (2011–2022), together with twelve 30 m conditioning factors, for model training and validation. Compared with logistic regression, Support Vector Machines (SVM), and Light Gradient Boosting Machine (LightGBM), RF consistently achieves higher accuracy across clusters, and the cluster-wise RF ensemble attains pooled ACC = 0.8212, F1 = 0.8176, and AUC = 0.8956. SHAP highlights both regionally consistent predictors (e.g., NDVI, distance to road) and distinct cluster-specific controls linked to geomorphic and hydrologic settings. The proposed framework enhances predictive accuracy, produces finer susceptibility gradients, and yields better-calibrated probability estimates than a single global model. These results demonstrate that explicitly accounting for geo-environmental heterogeneity can generate interpretable, spatially adaptive susceptibility outputs. By identifying high-risk zones for priority monitoring, land-use regulation, infrastructure protection, and mitigation planning, the proposed framework provides a practical decision-support tool for sustainable mountain development and disaster risk reduction in heterogeneous mountainous regions.

Destructive tsunamis are often triggered by shallow coseismic ruptures in subduction zones, making the rapid determination of rupture depth crucial for issuing timely tsunami warnings and mitigating associated hazards. To address this challenge, we propose a deep learning framework for the rapid classification of rupture depth (shallow or deep) based on high-rate GNSS data.Using the Alaska subduction zone as a case study, we generated nearly 10,000 synthetic earthquake scenarios to overcome the scarcity of real-world megathrust earthquake records. From these simulations, we constructed a comprehensive near-field GNSS three-component displacement waveform database. Leveraging this dataset, we designed a deep learning neural network that extracts critical seismic signal features from high-rate GNSS data to accurately classify rupture depth. The model achieved over 90% accuracy, precision, and recall on the test set.We applied the model to the 2021 Mw 8.2 Alaska earthquake and successfully identified it as a deep rupture, with a processing time of approximately 20 ms. Additionally, through transfer learning, we extended the model to the Sumatra subduction zone and successfully identified the 2010 Mw 7.8 Mentawai earthquake as a shallow rupture. This study provides a valuable reference for enhancing the reliability of tsunami early warning systems.

Landslide susceptibility mapping identifies terrain prone to slope failure under comparable environmental and triggering conditions. Grid-based machine-learning outputs can be spatially fragmented and difficult to translate into slope-scale mitigation. We developed a two-stage grid-to-slope-unit framework for rainfall-conditioned susceptibility assessment in Pingyuan County, Guangdong Province, China. Elevation, slope angle, slope aspect, normalized difference vegetation index, and rainfall were derived from an ALOS-PALSAR digital elevation model, Sentinel-2 imagery, and rainfall-station observations. A Random Forest model was trained at grid scale using 28 pre-2024 landslides and 28 pseudo-absence samples. Grid-scale scores and the five conditioning factors were then summarized within 4786 slope units and used in a second Random Forest model. An inventory of 359 landslides mapped after the June 2024 rainfall event was used for event-conditioned validation. High and very-high susceptibility classes occupied 21.10% of the modeled area and contained 299 validation landslides, yielding a capture rate of 83.3%. The very-high class contained 64.62% of validation landslides within 9.10% of the area. The framework produced geomorphically coherent terrain units for regional screening and field investigation, while detailed stability assessment remains necessary for individual sites.

Accurate and rapid estimation of earthquake source parameters is essential for postevent emergency response and cascading hazard assessment. Multisource observations provide complementary constraints on the rupture process, but their heterogeneous data characteristics make effective data fusion a key challenge for rapid source characterization. Here, we propose the multisource prior-constrained magnitude and slip estimator (MPMSE), a multitask deep learning framework that integrates high-rate Global Navigation Satellite System (GNSS) displacement waveforms, strong-motion velocity records, and magnitude-derived prior constraints. The framework combines parallel waveform encoders, station-geometry conditioning, and scaling-law-based priors to jointly estimate earthquake magnitude, rupture area, maximum slip, and spatial slip distribution. Synthetic earthquake tests show that MPMSE achieves stable magnitude estimation using a 90 s input window, with a mean absolute error of approximately 0.03. When applied to Japanese subduction-zone earthquakes and the 2024 Hualien earthquake, MPMSE rapidly recovered the main source characteristics of these real events, with GNSS displacement variance reductions exceeding 66%. Overall, MPMSE improves the reliability and observational consistency of rapid earthquake source characterization by integrating multisource observations and introducing magnitude-derived prior constraints, providing a new technical pathway for post-event rapid response and cascading hazard (e.g., tsunami) assessment.