Extreme rainfall can trigger numerous small, boundary-fragmented and spatially clustered shallow landslides. This study combined improved instance segmentation with reproducible spatial analysis to map and assess shallow landslides associated with the 2023 ‘23.7’ rainstorm in Zhaitang Town, mountainous Beijing. An improved Mask R-CNN integrated a convolutional block attention module, a path aggregation feature pyramid network and Soft-NMS. Under the random tile-level test design, the model achieved 89.51
Groundwater serves as a critical water source for domestic, industrial, and ecological security in valley-type cities across the Qinghai-Tibet Plateau, China. Under the combined pressures of climate change and urbanization, the evolution of regional groundwater systems has become increasingly complex. Taking Xining, a representative valley-type city on the northeastern margin of the Qinghai-Tibet Plateau, as the study area, this study systematically analyzes the dynamic evolution characteristics and driving mechanisms of groundwater based on groundwater depth data from 126 monitoring wells (1980–2025) and multi-source datasets including precipitation, evaporation, temperature, surface runoff, population, economy, land use, and groundwater extraction. By integrating the Mann–Kendall (M–K) test, Grey Relational Analysis (GRA), and Principal Component Regression (PCR), this study reveals the following findings: From 1980 to 2025, the groundwater depth in Xining increased from 10.7 m to 16.2 m before retreating to 13.5 m, undergoing four distinct phases: continuous decline, rapid rebound, dynamic stability, and slow recovery. Concurrently, precipitation and surface runoff exhibited increasing trends at rates of 1.61 mm/a and 0.06 m³/s, respectively, while evaporation decreased at a rate of 11.52 mm/a and temperature showed a slight decline of 0.007 °C/a. Groundwater extraction rose from 0.32 × 10⁸ m³/a to a peak of 2.30 × 10⁸ m³/a before dropping to 1.18 × 10⁸ m³/a. GRA identified groundwater extraction, precipitation, surface runoff, and population as the factors most closely associated with groundwater-depth variation, with relational degrees of 0.862, 0.816, 0.736, and 0.709, respectively. PCR further showed that anthropogenic and natural factors accounted for 59.1
Rainfall infiltration influences mechanical properties of the near surface and plays a fundamental role in controlling subsurface hydrological processes. Seismic ambient noise-based methods have emerged as valuable, non-invasive tools for monitoring hydrological dynamics. In this study, we investigate seismic velocity variations (δν/ν) derived from the coda portion of same-component autocorrelations and cross-component correlations of ambient seismic noise recorded by a single triaxial station installed at the Água Limpa Farm, in Brasília, Brazil. The objective of this work is to evaluate how δν/ν computed from different component combinations respond to precipitation and to assess whether rainfall-related hydrological processes can be tracked through single-station monitoring. A total of 420 days of continuous seismic data were acquired using a RaspberryShake RS3D sensor and processed through a workflow involving preprocessing, phase cross-correlation, temporal stacking, and velocity variation estimation via the Moving Window Cross-Spectral (MWCS) method. Seismic velocity series were then compared with rainfall records and Change Point Analysis was applied to identify significant velocity drops. Most component pairs showed increasing δν/ν during the climatological dry period and decreasing values during the rainy period, although component-dependent differences were observed. Strong correlations were observed between δν/ν drops and cumulative rainfall, particularly for horizontal component combinations. In contrast, vertical-component autocorrelations showed weaker and less consistent relationships with precipitation. We interpret the seismic velocity variations derived from single-station cross-component ambient-noise analysis as consistent with hydrological modulation. Moreover, the observed component-dependent sensitivity highlights the importance of incorporating multi-component data to enhance the detection and interpretation of subsurface hydromechanical processes.
Groundwater potential mapping in semi-arid karst regions is often compromised by spatially optimistic validation, limited assessment of predictive variability, and insufficient independent hydrogeological verification. This study introduces TriFusion-Karst, a stacked-ensemble model using RF and LightGBM as base learners and XGBoost as the meta-learner, embedded within a spatially robust and hydrogeologically validated workflow. Thirteen geo-environmental predictors were extracted at 7,313 groundwater monitoring wells, and GTD observations collected during 2018–2024 were pooled and used exclusively as the regression target. TriFusion-Karst and six alternative standalone and ensemble configurations were evaluated using five-fold spatially blocked cross-validation with leakage-controlled out-of-fold meta-feature generation, followed by 30-m spatial prediction. TriFusion-Karst achieved the best performance (RMSE = 1.47 m, MAE = 0.99 m, R² = 0.930, NSE = 0.926, and r = 0.964). Fold-wise spatial prediction variability was quantified using the pixel-wise standard deviation of five fold-specific predictions as an indicator of model stability rather than total predictive uncertainty. Independent validation using 250 mapped karst springs excluded from model development showed that 91.3
The geochemical and mineralogical processes underlying the mechanical weakening of rock and soil in subtropical South China remain insufficiently quantified. We evaluated lithology-dependent chemical weathering as an intrinsic background for collapses and landslides in northern and eastern Guangdong, China, coupling 6,841 mapped hazard sites with X-ray fluorescence and whole-rock X-ray diffraction (XRD) analyses of 40 samples from 11 sites. Igneous-rock areas show higher total hazard density (13.58 vs. 9.49 sites/100 km²) and 1.9-fold higher collapse density than sedimentary-rock areas. After correction for silicate CaO (CaO*) using measured P₂O₅, igneous products have site-balanced means (site means averaged equally) of 88.3 for the chemical index of alteration (CIA) and 71.9 for parent-rock-normalized CIA-rate, versus 85.0 and 21.7 for sedimentary products. CIA distributions overlap despite a modest sample-level difference (exact p = 0.043), whereas CIA-rate separates the product groups completely (exact p = 1.0 × 10⁻⁷). Whole-rock assemblages show feldspar depletion, quartz persistence and kaolinite enrichment in igneous products. Sedimentary products, with each subtype represented by one site, remain heterogeneous because of inherited 2:1 clay minerals and framework grains. Mineral-ratio differences within the conservative XRD sensitivity envelopes are not interpreted. These patterns characterize lithology-specific weathered materials associated with regional hazards; the observational design does not establish independent causal contributions. CIA-rate expresses normalized differences from selected lithology-specific parent-material baselines. Mineral ratios support interpretation of relative weathering intensity in the sampled igneous rocks. In sedimentary rocks, they serve only as lithology-sensitive descriptors of mineral assemblages.
