Landslide displacement prediction is vital for geohazard early warning and infrastructure safety. To address the challenges of modeling nonstationary, nonlinear, and multiscale behaviors inherent in GNSS time series, this study proposes a hybrid predicting framework that integrates Past Decomposable Mixing with an inverted Transformer architecture (PDM-iTransformer). The PDM module decomposes the original sequence into multi-resolution trend and seasonal components, using structured bottom-up and top-down mixing strategies to enhance feature representation. The iTransformer then models each variable’s time series independently, applying cross-variable self-attention to capture latent dependencies and using feed-forward networks to extract local dynamic features. This design enables simultaneous modeling of long-term trends and short-term fluctuations. Experimental results on GNSS monitoring data demonstrate that the proposed method significantly outperforms traditional models, with R² increased by 16.2–48.3% and RMSE and MAE reduced by up to 1.33 mm and 1.08 mm, respectively. These findings validate the framework’s effectiveness and robustness in predicting landslide displacement under complex terrain conditions.

Ecological risk assessment in arid inland river basins is challenging because ecological degradation is jointly controlled by water scarcity, surface exposure, thermal stress, energy exchange, and vegetation buffering. Existing remote sensing indices primarily characterize ecological quality and provide limited representation of risk formation and transmission. This study developed a Dryland Ecological Risk Index (DERI) within the source–pathway–sink framework to assess ecological risks in the Shiyang River Basin from 2000 to 2025. The Bare Soil Index and Temperature Vegetation Dryness Index represented risk sources, land surface temperature and albedo characterized transmission pathways, and the Normalized Difference Vegetation Index and wetness index represented ecological sinks. Principal component analysis, correlation analysis, high-resolution imagery, and comparison with the Remote Sensing Ecological Index were used for model validation, while transition matrices characterized changes among risk classes. The first principal component explained 93.97–97.51% of the total variance, and the directions of indicator loadings were ecologically consistent. DERI was most strongly correlated with drought stress (TVDI, r = 0.58) and effectively identified fine-scale degradation in deserts, bare land, oasis margins, and oasis–desert transition zones. Ecological risk exhibited a pronounced south–north gradient, with lower risks in the Qilian Mountains, oases, and river corridors and higher risks in the northern deserts. Basin-wide risk intensified during 2000–2010, underwent substantial spatial restructuring during 2010–2015, and gradually declined after 2015. The proportion of class V areas decreased from 32.94% in 2000 to 15.35% in 2025; nevertheless, classes IV–V still accounted for 52.40% of the basin in 2025. These findings demonstrate that DERI provides a process-oriented and spatially sensitive tool for identifying ecological risk hotspots and supporting differentiated ecological management in water-limited watersheds.

Inertial integration is an effective method for alignment measurement in confined spaces. Due to factors such as the large volume of data and the complexity of the calculations, there is currently no effective quantitative accuracy prediction method for inertial integrated alignment measurement in confined spaces in advance. This paper proposes an accuracy prediction method for the inertial/odometer/control point integrated measurement method. This method is based on Monte Carlo simulation. It generates true sensor measurements according to the designed alignment and generates error values based on the sensor error model. After introducing the error values into the true measurements, the simulated trajectory is solved. By repeating this process and conducting statistical analysis, the prediction accuracy of the alignment can be obtained. The experimental results indicate that this method achieves a resolution accuracy of better than 5 ppm, with an average relative accuracy of over 80% in real-world scenarios, demonstrating good validity and practicality.