With increasing research on geological hazards and the development of geographic information technology, slope units play an increasingly important role in landslide susceptibility assessment and prevention work. The scientific and reasonable division of slope units directly impacts the accuracy and practicality of analysis results. Despite the significant progress in slope unit division techniques, most existing methods still have certain limitations, such as a strong dependence on manually set thresholds during the division process, resulting in low levels of automation and efficiency. To address this issue, a new parameter-free slope unit extraction algorithm that integrates terrain factors, called Terrain Factor Parameter-Free Slope Unit Division (TFPF-SU), is introduced. This eliminates the issue of manually setting parameter thresholds during the slope unit division process. This algorithm fully utilizes the terrain information provided by digital elevation models (DEMs) to accurately calculate the curvature, slope, and aspect data for each point. On the basis of the inherent consistency principles among slope, aspect, and curvature, object-oriented image segmentation technology is used to achieve slope unit division. We select Dongchuan District in Yunnan Province, China, as a test area to verify the TFPF-SU algorithm and conduct a detailed comparative analysis and validation of the results with those obtained via traditional hydrological analysis methods from both qualitative and quantitative perspectives. In the quantitative analysis, we utilize the size and shape of the slope units. The results indicate the following: ① the slope units obtained with the TFPF-SU method are more uniform in size, avoiding issues with oversized or irregularly shaped units; ② the slope unit shapes obtained with the TFPF-SU method are more reasonable, with about 70% of the units falling within a reasonable shape index range, compared to only about 32% with the hydrological method; and ③ the slope units produced by the TFPF-SU method align more closely with terrain authenticity, exhibiting a higher degree of topographical conformity.

Semantic segmentation of land use and land cover (LULC) in arid regions remains challenging due to severe class imbalance, fragmented spatial distributions, and high spectral similarity among different land cover types. These characteristics often lead to an information bottleneck in deep segmentation networks and hinder the extraction of discriminative semantic representations. To address these issues, we propose SDS-Former, a lightweight semantic segmentation network specifically designed for remote sensing imagery in arid environments. SDS-Former incorporates an SSM-inspired Lightweight Semantic Enhancement (LSE) module to strengthen contextual modeling and alleviate the loss of discriminative information in deep features. To tackle scale variations, a Dynamic Selective Feature Fusion (DSFF) module is employed in the decoder to adaptively weight and fuse high-level semantics with low-level spatial details. Furthermore, a Feature Refinement Head (FRH) is introduced to enhance boundary localization and improve the recognition of small-scale and sparsely distributed land cover objects. Extensive ablation and comparative experiments demonstrate that SDS-Former consistently outperforms representative semantic segmentation methods across multiple evaluation metrics. On the Tarim Basin dataset, the proposed network achieves a mean Intersection over Union (mIoU) of 82.51% and an F1 score of 86.47%, indicating its superior effectiveness and robustness. Qualitative results further verify that SDS-Former exhibits clear advantages in distinguishing spectrally similar land cover types and preserving the spatial continuity of ground objects in complex arid-region scenes.

The Greiner–Hormann algorithm is a commonly used polygon overlay analysis algorithm. It uses a double-linked list structure to store vertex data, and its intersection calculation step has a significant effect on the overall operating efficiency of the algorithm. To address the time-consuming intersection calculation process in the Greiner–Hormann algorithm, this paper presents two kernel functions that implement a GPU parallel improvement algorithm based on CUDA multi-threading. This method allocates a thread to each edge of the subject polygon, determines in parallel whether it intersects with each edge of the clipping polygon, transfers the number of intersection points back to the CPU for calculation, and opens up corresponding storage space on the GPU side on the basis of the total number of intersection points; then, information such as intersection coordinates is calculated in parallel. In addition, experiments are conducted on the data of eight polygons with different complexities, and the optimal thread mode, running time, and speedup ratio of the parallel algorithm are statistically analyzed. The experimental results show that when a single CUDA thread block contains 64 threads or 128 threads, the parallel transformation step of the Greiner–Hormann algorithm has the highest computational efficiency. When the complexity of the subject polygon exceeds 53,000, the parallel improvement algorithm can obtain a speedup ratio of approximately three times that of the serial algorithm. This shows that the design method in this paper can effectively improve the operating efficiency of the polygon overlay analysis algorithm in the current large-scale data context.

Landslides are a common type of geological hazard, characterized by sudden onset, high destructiveness, and frequent occurrence, and are widely distributed in mountainous areas with complex terrain. In recent years, due to extreme weather and intensified human activities, both the frequency and intensity of landslide disasters in China have increased significantly, posing serious threats to human life, property, and socio-economic development. Although various methods for landslide susceptibility assessment have been proposed, the accuracy of existing models still needs improvement. In this context, this study takes the landslide-prone Dongchuan District of Kunming City, Yunnan Province, as a case study and proposes a coupled model that integrates an autoencoder and a feedforward neural network (AuFNN). The model uses the autoencoder to extract low-dimensional and highly discriminative feature representations, which are then used as input to the feedforward neural network to perform landslide susceptibility assessment. To evaluate the effectiveness of the proposed model, it is compared with four commonly used models, Support Vector Machine (SVM), Random Forest (RF), XGBoost, and Feedforward Neural Network (FNN), based on performance metrics such as the ROC curve, recall, and F1 score. The results indicate that the AuFNN model provides an alternative representation learning framework and achieves performance comparable to that of established machine learning models in landslide susceptibility assessment, as reflected by similar AUC, accuracy, and F1 score values.