Precipitable water vapor (PWV) is a crucial factor in regulating the Earth’s climate. Moreover, it demonstrates a robust correlation with precipitation. Situated in a region known for the generation and development of tropical cyclones, Guangxi in China is highly susceptible to floods triggered via intense rainfall. The atmospheric water vapor in this area displays prominent spatiotemporal features, thus posing challenges for precipitation forecasting. The water vapor products within the MERRA-2 and ERA5 reanalysis datasets present an opportunity to overcome constraints associated with low spatiotemporal resolution. In this study, the PWV data derived from GNSS and meteorological measurements in Guangxi from 2016 to 2018 were used to evaluate the accuracy of MERRA-2 and ERA5 water vapor products and their relationship with water vapor variations during extreme rainfall. Using GNSS PWV as a reference, the average bias of MERRA-2 PWV and ERA5 PWV for heavy rainfall was −0.22 mm and 1.84 mm, respectively, with average RMSE values of 3.72 mm and 3.31 mm. For severe rainfall, the average bias of MERRA-2 PWV and ERA5 PWV was −0.14 mm and 2.92 mm, respectively, with average RMSE values of 4.28 mm and 4.01 mm. During heavy rainfall days from Days 178 to 184 in 2017, the average bias of MERRA-2 PWV and ERA5 PWV was 0.92 mm and 2.42 mm, respectively, with average RMSE values of 4.04 mm and 3.40 mm. The accuracy was highest at the Guiping and Hechi stations and lowest at the Hezhou and Rongshui stations. Furthermore, when comparing MERRA-2/ERA5 PWV with GNSS PWV and actual precipitation, the trends in the variations of MERRA-2/ERA5 PWV were generally consistent with GNSS PWV and aligned with the increasing or decreasing trends of actual precipitation. In addition, ERA5 PWV exhibited high accuracy. Before the onset of heavy rainfall, PWV has a sharp surge. During heavy rainfall, PWV reaches its peak value. Subsequently, after the cessation of heavy rainfall, PWV tends to stabilize. Therefore, the reanalysis data of PWV can effectively reveal significant changes in water vapor and actual precipitation during periods of heavy rainfall in the Guangxi region.

Extracting rampart crater ejecta blankets is crucial for understanding impact crater formation and material transport processes, offering key insights into the distribution of subsurface water and ice on Mars. However, traditional methods often fail to extract rampart crater ejecta blankets due to complex terrain, noise interference, and blurred boundaries. To overcome these challenges, we propose an edge-aware segment anything model (SAM) sputter analysis (EASSA) framework for the rapid and accurate extraction of rampart crater ejecta contours. EASSA comprises three key components. First, Wiener filtering and multiscale Retinex preprocessing are applied to suppress terrain noise and enhance the visual distinction between rampart features and the background. Second, an SAM-based unsupervised segmentation module is employed to automatically identify rampart crater boundaries. Finally, we refined the extracted edges by designing a contour optimization pipeline that applies classical image operators, such as Sobel to enhance gradients, Suzuki–Abe to correct contours, and Douglas–Peucker to simplify and smooth shapes. To validate EASSA, we construct a multiscale rampart crater dataset using context camera (targeting craters 1 km) in the Chryse Planitia and Arabia Terra regions. Experimental results demonstrate that our method achieves a detection accuracy of 97.36%, a recall of 93.36%, and an intersection over union of 0.93, significantly outperforming the baseline SAM segmentation. Morphological analysis further reveals that rampart ejecta in both regions exhibit mobilities greater than 2, an average lobateness coefficient of 1.06, and relatively shallow excavation depths. Additionally, we observe that elevated terrains exhibit lower ejecta flow mobility under similar latitudinal conditions, while geomorphic evidence of past fluvial activity remains evident. These findings provide new insights into Martian subsurface water dynamics and the mobility characteristics of ejecta under varied geologic settings.

Layered Ejecta Craters (LECs) are distinctive Martian craters with fluidized ejecta whose morphological characteristics correlate with Martian liquid water and the distribution of subsurface water ice. Existing crater extraction methods struggle to effectively extract LECs with complex edges or significantly degraded boundaries and fail to sufficiently characterize weak-texture targets under complex Martian backgrounds. To address these issues, an Edge-Guided Dual-Branch Transformer Network (EDTNet) fusing convolutional neural network (CNN) and Transformer is proposed for automated LEC extraction. The encoder adopts a dual-branch design integrating CNN and Transformer, retaining CNN’s local feature extraction capability while leveraging Transformer’s long-range global information capture. An Edge-Guided Semantic Focusing (EGSF) module is integrated into the CNN branch to improve the extraction accuracy of degraded LEC boundaries. A Fast Fourier Convolution Attention (FFCA) mechanism is introduced in the Transformer branch to mitigate complex Martian background interference and strengthen weak-texture target representation. A Feature Fusion Module (FFM) is incorporated between the two branches to fuse local and global features. Extensive experiments on the self-constructed Mars Layered Ejecta Crater (MLEC Dataset) demonstrate that EDTNet achieves an mIoU of 90.35%, outperforming SOTA methods such as MobileNet, DeepLabv3, ConvNeXt, Swinformer, SegFormer and EASSA.

River ice semantic segmentation is a crucial task that provides essential information for hydrological monitoring and infrastructure protection in cold regions. Previous works mainly focus on global long-range dependency modeling or local feature extraction, while the balance between computational efficiency and fine irregular-boundary preservation is often neglected. In this paper, we propose IceRWKV, an efficient semantic segmentation network for river ice based on the Receptance Weighted Key Value (RWKV). First, the RWKV sequence model is introduced into this task to break the quadratic complexity bottleneck, achieving high-precision global–local feature aggregation with low computation cost. Then, a novel Geometry-Direction Co-sensing Module (GDCM) is adopted to fit irregular ice contours and suppress background noise through an adaptive geometric correction and polarization feature-refinement strategy. Furthermore, Haar wavelet downsampling (HWD) is utilized to replace traditional downsampling operations, effectively mitigating feature aliasing and preserving high-frequency details. We conduct extensive experiments on the NWPU_YRCC_EX, NWPU_YRCC2, and Alberta River Ice Segmentation datasets. Comprehensive experimental results demonstrate that IceRWKV achieves state-of-the-art (SOTA) performance against 10 competing methods. Specifically, on the NWPU_YRCC_EX dataset, our method achieves a Mean Intersection over Union (mIoU) of 93.41% and an inference speed of 37.19 Frames Per Second (FPS) on NWPU_YRCC_EX, demonstrating a favorable trade-off between segmentation accuracy and computational efficiency.