Artificial islands face severe land subsidence (LS) challenges, yet accurate spatiotemporal prediction remains difficult. This study investigates LS patterns on Hengqin Island, China, by integrating Persistent Scatterer Interferometry (PS-InSAR) with machine learning (ML). We analyzed 96 Sentinel-1 images (2016-2023), validating measurements against historical levelling data (RMSE = 3.5 mm/yr). PS-InSAR results revealed a maximum subsidence rate of 334 mm/yr in 2019, which decelerated to 202 mm/yr by 2022. To forecast future deformation, four ML architectures were evaluated. The Back Propagation (BP) neural network demonstrated superior predictive accuracy (RMSE = 2.59 mm, R2 = 0.99). Using the trained BP model, we forecasted LS trends for 2024-2026, indicating persistent risks primarily in areas with high-saturation soft soils. This study pioneers a hybrid framework synergizing PS-InSAR time-series with explainable ML, providing critical insights into the geotechnical stability of artificial islands to support sustainable coastal planning and disaster mitigation.

Groundwater resources are essential to global freshwater supply, and accurate groundwater level prediction is critical for sustainable water resource management. To overcome the limitations of traditional deep learning models in long-sequence groundwater forecasting, including weak generalization, reduced long-term prediction accuracy, and limited interpretability, this study proposes a dual-path Informer-p model integrated with residual theory. The main path captures nonlinear temporal dependencies and long-term hydrological patterns, while the residual path provides a stable linear prediction baseline to enhance local fluctuation representation and robustness to extreme events. The model was validated using long-term groundwater observations from 34 monitoring stations across five major ecosystems in China. Results from representative stations, including Ailao Mountain, showed that Informer-p achieved excellent predictive performance with RMSE = 0.05 m, MAPE = 1.2%, R2 = 0.95, and KGE = 0.95, reducing RMSE and MAPE by 37.5% and 52%, respectively, compared with the original Informer. Across all stations, Informer-p outperformed the original Informer at 22 stations, with the greatest improvement observed in forest ecosystems. SHAP analysis identified window maximum, original groundwater level, and window minimum as the dominant predictive features. The proposed model provides an effective tool for national-scale groundwater level prediction and sustainable groundwater management.

Highlights What are the main findings? A spatiotemporal synchronous prediction framework is proposed for large-scale complex InSAR ground deformation fields. A combined ICA and K-means approach is proposed to identify dominant evolution patterns of the deformation field and their spatial distributions. What are the implications of the main findings? The proposed framework improves the prediction capability for complex multimodal ground deformation processes. The identified interaction patterns between ground deformation and groundwater provide insights for urban groundwater management and geohazard assessment.Highlights What are the main findings? A spatiotemporal synchronous prediction framework is proposed for large-scale complex InSAR ground deformation fields. A combined ICA and K-means approach is proposed to identify dominant evolution patterns of the deformation field and their spatial distributions. What are the implications of the main findings? The proposed framework improves the prediction capability for complex multimodal ground deformation processes. The identified interaction patterns between ground deformation and groundwater provide insights for urban groundwater management and geohazard assessment.Abstract Ground deformation is a major geohazard in many urban areas, requiring reliable monitoring and forecasting for hazard mitigation. Although Multi-Temporal InSAR enables high-resolution deformation monitoring, most prediction approaches rely on single-point modeling and fail to exploit spatial dependencies within deformation fields. This study proposes a spatiotemporally synchronous prediction framework for large-scale InSAR deformation fields, integrating sequence preprocessing, spatiotemporal modeling, and deformation pattern analysis. First-order differencing reduces sequence non-stationarity, while a patch-based encoder-decoder structure preserves spatial topology during dimensionality reduction. The core prediction model, built on PredRNNv2, captures the long-term spatiotemporal evolution of InSAR deformation sequences. In addition, independent component analysis (ICA) combined with K-means clustering identifies dominant deformation patterns and their geological associations. The framework is evaluated using synthetic datasets simulating multiple deformation mechanisms and Sentinel-1 InSAR time-series data over the Beijing Plain from 2015 to 2025. Results show that the model accurately captures deformation evolution and identifies transitions associated with groundwater regulation. These findings demonstrate the potential of deep spatiotemporal learning for large-scale InSAR deformation prediction and geohazard mechanism interpretation.

Human-landscape interactions in mountainous terrains are complex and multi-faceted. Settlements tend to focus on relatively gently undulating terrains, which are often found in areas where ground conditions are weak and thus substantial ground movements prevail. Interventions in these precarious landscapes, such as progressive expansion of interconnecting transport infrastructure, affect the stress balance and hydrology of already critical slopes, potentially enhancing their sensitivity to changes. The mountainous Bailong River Corridor (BRC) in Northwest China is dominated by large slow-moving landslides, where any transport infrastructure expansion has to transit through. A complex interplay of human, precipitation, and seismic factors determines the triggering dynamics of these large movements. This study integrates displacement time series, precipitation, and road distribution to quantify the impact of road emplacement on the sensitivity of large slow-moving landslides to precipitation regionally using panel regression analysis. It is shown that road disturbance significantly amplifies the sensitivity of landslide displacements to precipitation, and paved roads on the large slow-moving landslides increase their sensitivity to precipitation by 40%. Roads (both paved and unpaved) also reduce the threshold of antecedent cumulative precipitation required to trigger significant displacement, shortening the typical period from 132 days to only 120 days. The enhanced response frequency increases large reactivated landslide risk, and impacts road operation and management. This better understanding of the precipitation signature in the dynamics of large slow-moving landslides transited by roads contributes to improving future road planning, enhancing landslide risk mitigation, and strengthening urban resilience in vulnerable alpine environments.

Reclaimed coastal areas are highly susceptible to uneven ground deformation, posing significant risks to urban infrastructure. This study investigates the long-term subsidence evolution of Hengqin Island by utilizing 105 Sentinel-1A scenes acquired between September 2016 and April 2025 via Persistent Scatterer Interferometric Synthetic Aperture Radar (PS-InSAR). Existing prediction methods often struggle to effectively capture the complex nonlinear spatiotemporal correlations inherent in subsidence data. To address this limitation, we propose a novel Spatial-Temporal Graph Convolutional Network-Transformer (STGCN-Transformer) model. Unlike conventional approaches that treat spatial and temporal features in isolation, our model jointly learns coupled spatiotemporal patterns. Specifically, it utilizes STGCN to extract local spatial dependencies based on geographic proximity topology, while integrating a Transformer with a multi-head self-attention mechanism to capture long-range temporal dependencies within deformation sequences. Comprehensive comparative experiments against CNN, LSTM, and standalone Transformer baselines demonstrate the superiority of the proposed framework, which achieved the highest accuracy across RMSE, R2, MAE, and sMAPE metrics. Based on the validated model, we forecast a continued subsidence trend, with a maximum cumulative settlement projected to reach-64.23 mm by January 2029. Furthermore, a four-level land subsidence risk zonation map was generated based on the predictions, providing actionable insights for urban planning and disaster mitigation in reclaimed areas. This study establishes a robust framework for high-precision InSAR deformation prediction and provides critical scientific support for disaster prevention and sustainable urban planning in complex reclaimed environments.

Highlights What are the main findings? The proposed Spatially Adaptive Ensemble Learning model (SA-GSE) effectively captures the spatial heterogeneity of land subsidence in the Yellow River Delta, achieving superior predictive performance (R2 = 0.7810, RMSE = 40.55 mm/yr) and eliminating spatial autocorrelation in residuals (Moran's I = 0.0334, = 0.206). pDistance to salt pans is the dominant driver of subsidence (importance 0.4456), and a nonlinear "precipitation decoupling" mechanism is revealed in salt pan areas, where high precipitation paradoxically exacerbates subsidence due to anti-seepage treatments and brine density stratification. What are the implications of the main findings? The SA-GSE framework provides a novel methodological pathway for deciphering subsidence mechanisms in heterogeneous regions, overcoming the limitations of traditional global models that assume uniform driving processes. The identification of brine extraction as the primary anthropogenic driver and the mapped high-rate zones offer a scientific basis for targeted land subsidence prevention, groundwater management, and sustainable coastal development.Highlights What are the main findings? The proposed Spatially Adaptive Ensemble Learning model (SA-GSE) effectively captures the spatial heterogeneity of land subsidence in the Yellow River Delta, achieving superior predictive performance (R2 = 0.7810, RMSE = 40.55 mm/yr) and eliminating spatial autocorrelation in residuals (Moran's I = 0.0334, = 0.206). pDistance to salt pans is the dominant driver of subsidence (importance 0.4456), and a nonlinear "precipitation decoupling" mechanism is revealed in salt pan areas, where high precipitation paradoxically exacerbates subsidence due to anti-seepage treatments and brine density stratification. What are the implications of the main findings? The SA-GSE framework provides a novel methodological pathway for deciphering subsidence mechanisms in heterogeneous regions, overcoming the limitations of traditional global models that assume uniform driving processes. The identification of brine extraction as the primary anthropogenic driver and the mapped high-rate zones offer a scientific basis for targeted land subsidence prevention, groundwater management, and sustainable coastal development.Abstract The Yellow River Delta, a young alluvial plain in China, is experiencing severe land subsidence that threatens its ecological security and sustainable development. However, the driving mechanisms of this subsidence exhibit strong spatial heterogeneity, which traditional global models fail to capture. This study integrates high-precision subsidence measurements from Sentinel-1A imagery and SBAS-InSAR technology (2017-2023) with multi-source environmental factors (topography, geology, land use, precipitation) to propose a Spatially Adaptive Ensemble Learning Model with feature selection (SA-GSE). The model concatenates predictions from base learners (CatBoost, XGBoost, Random Forest) with spatial features (e.g., distance to salt pans, local topographic variance) to form meta-features, which are then input into a multilayer perceptron meta-learner. Through 5-fold spatial cross-validation, SA-GSE learns spatially dynamic base-model weights, implicitly adapting to regional variations in subsidence drivers. The model achieves an R2 of 0.7810 and RMSE of 40.55 mm/yr on the test set, outperforming individual base models and ordinary stacking. Residual spatial autocorrelation is substantially reduced, with SA-GSE yielding the lowest Moran's I (0.0334, p = 0. 206) among all evaluated models, confirming effective capture of spatial heterogeneity. Driving force analysis reveals that distance to salt pans is the most important predictor (permutation importance: 0.4456), underscoring the dominant role of brine extraction-induced aquifer compaction. Lagged precipitation importance (0.3191) exceeds that of current precipitation (0.2453), indicating a recharge lag effect. SHAP interaction analysis uncovers a nonlinear "precipitation decoupling" mechanism in salt pan areas, where high precipitation paradoxically exacerbates subsidence. The resultant map of predicted subsidence rates highlights elevated rate zones in the northern salt pans and along the Guangli River. While the map does not represent a full risk assessment-as it does not include exposure or vulnerability-it provides a spatially explicit estimate of hazard likelihood. This ensemble framework yields novel perspectives on subsidence drivers in heterogeneous regions and can support land subsidence prevention and groundwater management planning.

Precipitation regime fluctuations induced by climate change have a profound effect on the stability of earthflows in alpine and gorge regions. Traditional geomorphological mapping of areas at risk of earthflows requires high-resolution optical imagery and digital terrain datasets, which are often unavailable in these remote regions. This study demonstrates that the integration of synthetic aperture radar interferometry (InSAR) observations with topographic analysis enables effective delineation and identification of earthflow runout paths, using a large earthflow in southern Gansu Province, China, as a representative case study. The topographic wetness index (TWI) is shown to be an effective indicator for highlighting characteristic spatial signatures, thereby facilitating the accurate identification of the source zone, main track, depositional area, and runout path of a reactivated earthflow. Monitoring of precursory displacements reveals that continuous, moderate precipitation plays a critical role in the reactivation and eventual failure of a large earthflow. The primary failure was triggered by a daily rainfall event of 37 mm, following an 8-day period of gradual softening along the basal slip surface. Accordingly, the magnitude of this triggering event was governed by both rainfall intensity and the duration of antecedent weakening processes. Deformation patterns indicate that subsurface water accumulation is primarily controlled by rainfall infiltration along preferential flow paths, rather than by diffuse groundwater movement. These observations support the development of a conceptual model for delineating the extent of potential deep-seated earthflows and inferring their runout magnitude. The proposed model integrates InSAR-derived deformation data, spatial TWI distributions, and mechanical-morphological analysis, providing a practical framework for assessing and mitigating the risks associated with earthflow reactivation.

Understanding runoff changes in the Yarlung Zangbo (YZ) basin is crucial for water resource management on the Tibetan Plateau (TP), but has been hindered by sparse data. This study provides a comprehensive hydrometeorological dataset for the YZ basin spanning 1961-2024, including both daily meteorological variables and hydrological components. The dataset integrates observations from 26 meteorological stations, 15 hydrological stations, and hydrological model simulations. Specifically, it includes: (1) daily meteorological variables (precipitation, mean, maximum, and minimum temperature, and wind speed) at both 26 meteorological stations and a 10-km gridded scale; and (2) daily total runoff and its components (snowmelt, glacier runoff, and rainfall runoff) at the 10-km gridded scale, derived from a physically based hydrological model. The meteorological fields were extended using a random forest-based machine learning approach combined with ERA5 reanalysis data, while the hydrological variables were generated from physically based hydrological model simulations. To facilitate direct application of hydrological station data, the gridded daily hydrological data for 1961-2024 were further routed to 15 hydrological stations. The generated dataset was validated against in situ meteorological at different scales and hydrological observations and compared with other meteorological products, demonstrating good applicability in the YZ. This dataset is available at the National Tibetan Plateau Data Center. The dataset enables basin-wide simulations of hydrological processes across multiple scales and provides a valuable resource for assessing water cycle dynamics and their responses to climate and cryospheric changes in the YZ basin.

The Yellow River Delta is a national nature reserve and an important wetland area. Land subsidence in this area poses a significant risk to the regional ecological environment. Therefore, land subsidence prediction is important. However, land subsidence monitoring is conducted more frequently than land subsidence simulation and prediction in the Yellow River Delta. Therefore, on the basis of temporal InSAR data on land subsidence from 2017 to 2023 in this region, we employed four machine learning models (SVM, transformer, LSTM, and PSO-BP) to perform a predictive analysis and comparative experiments. The results reveal that the PSO-BP model performed best, with the lowest root mean square error (RSME) average of 0.023. Compared with those of the SVM, LSTM, and transformer models, the accuracy of the PSO-BP model improved by 23.14%, 34.97%, and 80%, respectively. Predictions for the next eight months revealed a continuous decreasing trend in land subsidence. The experiments revealed that appropriately increasing the training set-to-test set ratio can effectively increase the accuracy of model predictions for small sample data. The precision values for the ratio of 8:2 increased by 38.46%, 43.26%, and 81.95% compared with those for the ratios of 7:3, 6:4, and 5:5, respectively. The selected optimal machine learning model, on the basis of the comparative analysis presented, was employed to determine future land subsidence trends throughout the Yellow River Delta for the next five years. This information will improve strategies for regional management and subsurface brine resource exploitation. These findings support the development of a model selection method for more accurate prediction of land subsidence.

Active potential landslides pose substantial threats to lives and property in alpine‐canyon terrain worldwide. Identifying landslide‐prone areas and assessing the failure likelihood of potential landslides are crucial for risk mitigation. However, uncertainties from incomplete inventories and variable data quality limit the reliability and practical application of landslide hazard assessments. This study proposes a novel metric method to assess potential landslide hazard in alpine‐canyon regions by integrating the advanced observation capability of remote sensing techniques and reliability of geomorphic surveying. A comprehensive inventory of potential landslides was established via multi‐temporal interferometric synthetic aperture radar (InSAR) mapping of the eastern Qinghai–Tibet Plateau, with landslide types classified based on their material compositions and movement characteristics. The observed time‐series displacements and geomorphological deformation features indicate the progressive creep behaviour of landslide movement, reflecting the different hazard levels of potential landslides across their multiple stages of development. The dynamic trends of most potential landslides are characterised by seasonal accelerating creep and geomorphic movement features that range from localised to intense deformation. The hazard assessment demonstrates that 23.7% of potential landslides have reached or exceeded the high hazard level, with most of these having large and deep characteristics, and closely related to active fault zones in the study area. Internal geological conditions and fluctuating precipitation commonly elevate the landslide hazard level in critical regions. This integrated analysis of the dynamic evolution of potential landslides and geomorphic deformation features improves hazard prediction for landslides in mountainous regions undergoing long‐term creep.

Landslide caused catastrophic disasters frequently in the Karakoram Mountains wide range. The Hunza Valley, Pakistan in the Northwest of the Karakoram mountains, which is prone to the clustering development of landslide was taken as a case in this study. The updated complete inventory including 53 SBAS-InSAR detected active landslides and optical image interpreted 65 landslides were constructed, based on Sentinel-1A data in 2019–2020 and several field survey until 2023. Twelve factors related to geomorphology, hydrology, vegetation, geology, tectonics, and environment were incorporated into the model training within twelve machine learning models: Generalized Linear models, Navies Bayes, Nearest Neighbors, Support Vector Machines and so on. The Support Vector Classification was selected for landslide susceptibility mapping (LSM) and its characteristics in geomorphologically meaningful landscape partitions called slope units, with the highest accuracy of 0.96, average AUC of 0.99 for tenfold cross-validation, and high computational efficiency of 6.11 s. The results revealed that the areas with moderate landslide susceptibility account for 62.14

The Multiple Temporal Interferometric Synthetic Aperture Radar (MT-InSAR) technology is capable of effectively generating ground deformation information derived from high-precision and continuous observation by satellites. However, due to the limited operational lifespan of a single SAR satellite, the derived ground deformation result of the study area cannot be ensured long-term (several decades), and merely a few years. With the increasing number of SAR satellite launches, it has become possible to conduct long-term continuous monitoring of ground deformation by combining data from multiple platforms. Nevertheless, several existing methods (e.g., model fitting method, predictive splicing method, etc.) have lower fusion accuracy and are limited to specific deformation patterns. In this study, a Piecewise Exponential Fitting with Weighted Average (PEFWA) method is proposed, which takes into account both the trend and accuracy of the preceding and following deformation time series in the fusion. The experimental results on the simulation data prove that the accuracy and robustness of this method are higher than several other methods. We applied the proposed method to characterize the evolution of ground deformation in the Beijing Plain from 1992 to 2023 using data from four different SAR satellites. The results show that: (1) With the implementation of various policies (e.g., the South-to-North Water Diversion Project, the Ecological Water Replenishment Project, etc.), ground subsidence has generally followed a trend of "worsening initially, then improving". (2) The spatial variability of ground subsidence is primarily influenced by the locations of fault zones. (3) The periodic changes in the ground deformation time series are mainly driven by fluctuations in groundwater levels. The above findings indicate that the method proposed in this study can effectively integrate deformation series with temporal discontinuities, which helps detect the long-term trends and formation mechanisms of ground deformation.

The Zhouqu region is located in the middle reaches of the Bailong River in southern Gansu Province. It is recognised as one of the most active geohazards regions in China. This paper presents more than a decade of observations (2010-2023) of the evolution of landslides along an active fault zone in the Zhouqu region. A varied lithology comprises shales and phyllites in a fault-controlled geomorphology that conditions the slopes and has resulted in large ancient landslide complexes. The activity of these landslides was assessed using InSAR (Interferometric Synthetic Aperture Radar), a technique capable of generating historical and recent ground deformation data. These assessments were validated through field investigations. The region hosts 31 active landslides, including four large ancient landslides with areas greater than 1 km2, each displaying velocities exceeding 320 mm/year between February 2017 and August 2023. These landslides cover an area of approximately 35.7 km2, some 16.4 % of the Zhouqu region (218 km2). The findings suggest that tectonic activity and lithology play critical roles in landslide and landscape development. Gradual changes in climate have the potential substantially alter the precipitation regime, which affects the stability of slopes and the mobility of large, slow-moving landslides. This research highlights the need for long-term monitoring (InSAR plus fieldwork) to achieve a better understanding of the evolution of large landslide in such dynamic regions that are influenced by tectonic, climatic and anthropogenic conditions. This knowledge adds to our understanding of the ways in which humans influence and, in turn, will be influenced by these large slope deformation processes in these dynamic terrains.

Earlier studies have independently examined the hydrodynamic effects of either streamlining angle or vegetation density in model vegetation canopies. However, the coupled influence of these two parameters on the three-dimensional hydrodynamics of infinite arrays of inclined cylinders remains insufficiently understood. This study addresses this gap by employing large eddy simulations to investigate the interplay between streamlining angle and vegetation density in periodic cylinder arrays that mimic aquatic vegetation. The simulations reveal that increasing vegetation density elevates drag, torque, and flow unsteadiness, especially near the bed. The streamlining angle exerts a strong influence on spanwise asymmetry, vortex shedding characteristics, and vertical wake structure. Drag force stability improves at moderate inclinations, while high angles intensify wake three-dimensionality and vertical momentum transport. The Strouhal number and vortex shedding frequency exhibit nonlinear sensitivity to both inclination and spacing, diverging from trends observed in isolated or upright cylinders. Pressure and velocity distributions demonstrate significant vertical heterogeneity, emphasizing the importance of three-dimensional flow modeling. By systematically varying both inclination and spacing in an infinite array context, this study provides the first comprehensive framework to evaluate fluid-vegetation interactions relevant to flexible aquatic canopies.

With climate change, the Qinghai–Tibet Highway (QTH) is facing increasingly severe risks of natural hazards, posing a significant threat to its normal operation. However, the types, distribution, and future risks of hazards along the QTH are still unclear. In this study, we established an inventory of multi-hazards along the QTH by remote sensing interpretation and field validation, including landslides, debris flows, thaw slumps, and thermokarst lakes. The QTH was segmented into three sections based on hazard distribution and environmental factors. Susceptibility modelling was performed for each hazard within each section using machine learning models, followed by further evaluation of hazard susceptibility under future climate change scenarios. The results show that, at present, approximately 15.50% of the area along the QTH exhibits high susceptibility to multi-hazards, with this proportion projected to increase to 20.85% and 23.32% under the representative concentration pathways (RCP) 4.5 and RCP 8.5 distant future scenarios, respectively. Variations in hazard-prone environments dominate the spatial heterogeneity of multi-hazard distribution. Gravity hazards demonstrate limited sensitivity to climate change, whereas thermal hazards exhibit a more pronounced response. Our geomorphology-based segmented assessment framework effectively enhances evaluation accuracy and model interpretability. The results can provide critical insights for the operation, maintenance, and hazard risk management of the QTH.

River blockage induced by tributary debris flow is a common hazardous chain in mountainous areas, which could pose a serious threat to human lives and infrastructures. Especially, the failures of landslide dams in tributaries will amplify the debris flow’s magnitude, which could increase the probability of river blockage. However, the dynamic process and criterion of the river blockage induced by such debris flow have not been well understood. Here, we modeled the entire process of river blockage induced by tributary debris flow through large-scale field experiments. The formation process of the river blockage can be outlined into four stages according to the dynamic characteristics at the junction: erosion and entrainment of the river, expansion of the debris flow deposits, damming of the main river, and the dam overflowing and breaching. The erosion rate of the main river and the momentum of debris flow dominate the first and latter processes of the river blockage respectively, and the shift is accompanied by the increasing momentum of the debris flow. Additionally, the erosion rate of the river and the momentum of the debris flow play differential importance in each type of river blockage. The river’s influence is significant for the formation of major blockage and partial blockage if the momentum of tributary debris flow is relatively small. Once the momentum of debris flow is high enough, dynamic characteristics of the tributary are vital for the type of blockage. Then, we established a criterion based on the process of erosion and deposition in the confluence zone. The criterion was verified by 5 river blockage events in the field. It is found that the coupling of the two indicators will improve the accuracy of identification of river blockage compared with existing criteria. In particular, it is possible to better distinguish the type of blockage. This study would advance the understanding of a debris flow dam’s formation, and it is meaningful for the early identification of river blockage.

Loess landslides, predominantly triggered by human activity and precipitation, dominate the geomorphological evolution of extensive regions of the Chinese Loess Plateau. Tablelands constitute a large proportion of this Loess Plateau. Although the area has a long history of human occupation, the Bailu Tableland is relatively poorly studied. Recent landslide events along the edge of the tableland significantly affected lives and livelihoods. To develop a better understanding of geomorphological evolution and develop appropriate hazard management and mitigation strategies, we characterized recent landslide activity and determined whether there are discernable trends in activity. We used Unmanned Aerial Vehicle (UAV) observations, Interferometric Synthetic Aperture Radar (InSAR) monitoring, and Digital Terrain Analysis (DTA) to observe significant differences in the position, shape, and magnitude between old and new landslides. The geomorphological evolution of the tableland was initially driven by large landslides, while current and future morphological development is more likely affected by small landslide events. We demonstrate the successful application of InSAR to investigate potential landslides in well-vegetated loess landscapes and show that loess landslides tend to develop on slopes with high relative relief and high topographic wetness associated with gully erosion. Unlike landslides induced by groundwater variations in the Heifangtai and South Jingyang tablelands, precipitation and infrastructure construction form the main controls on landslide activity in the Bailu Tableland. We propose two geomorphological evolution models that describe loess tableland recession caused by landslides. The results from this research contribute to a better understanding of the reactivation behavior of landslides along loess tablelands, provide a scientific foundation for landslide hazard and risk management, and supplement understanding of the geomorphological evolution of loess tableland landscapes.

Rainfall can increase the moisture content of a slope, which changes its mechanical properties and thus acts as an important mechanism to trigger landslides. However, it is unclear how moisture contents vary in space and time during rainfall-induced slope movements, and which soil-wetting patterns precede landslide events. Here, we used point sensors and time-lapse 3D electrical resistivity tomography (tl-3D-ERT) technology to monitor the spatiotemporal evolution of the hydrology and movement within a rainfall-induced loess landslide. We observed that movement of the slope involved a semi-continuous process of initiation, acceleration, and deceleration to stabilization. The slope hydrology evolution suggested that initial saturation, dominant flow, and changes in slope recharge and drainage owing to internal seepage and erosion are important factors that affect moisture changes. The movement accelerated when the average saturation value and spatial variation in moisture distribution within the slope increased; however, the movement decelerated when both parameters did not change significantly with time. The accumulation and dissipation of pore water pressure within the slope owing to uneven humidification may be the underlying cause of changes in landslide movement. Our study demonstrates the potential of tl-3D-ERT for monitoring the spatiotemporal variability of moisture evolution within rainfall-induced landslides to determine landslide deformation trends and develop a landslide early warning system.

The continued increase in demand for food is such that flood-irrigation-based agricultural techniques are not expected to change in the future. Consequently, in cultivated areas dominated by river terraces that experience neotectonic movements, catastrophes caused by earthquakes that trigger material liquefaction may continue to occur. Among such catastrophes, low-angle, ultra-long-distance landslides induced by synergistic effects of tectonic and human activities are events that are not fully understood but can potentially be mitigated. A landslide-mudflow that occurred on December 18, 2023, during an Ms 6.2 earthquake in Jishishan, China, on the tertiary terrace of the Yellow River. We reveal the causal mechanisms by which geomorphology, tectonics, and human activities controlled the landslide-mudflow, and outline a new, potentially widespread, dual liquefaction layer-dominated failure mode for earthquake-induced landslides on river terraces. Our findings serve as a warning that urgent landslide-potential assessments should be conducted across river terrace irrigation areas in tectonically active regions.