Farmland segmentation is crucial for precision agriculture and land resource management, and very high resolution synthetic aperture radar (VHR SAR) offers valuable data support. Yet, effectively leveraging this data remains challenging. Convolutional neural networks (CNNs) struggle to capture long-range dependencies, while transformers, though effective in long-range modeling, suffer from high computational costs. Additionally, noise in VHR SAR images often results in ambiguous edge semantics. To address these issues, a farmland segmentation network for VHR SAR, named the CNN-VMamba network (CVMNet), is proposed in this letter. Specifically, a gradient-curvature joint algorithm (GCJA) is applied to VHR SAR images to generate edge strength map, providing enhanced contour information. In parallel, a CNN-based local feature extraction (LEF) block, sensitive to tubular structures, is employed to capture fine-grained spatial details. For efficient long-range modeling with linear computability, a 2-D selective scan (SS2D) mechanism from VMamba is integrated. Furthermore, an SS2D-based shallow-deep feature fusion module (SDFFM) is proposed to facilitate cross-layer feature fusion. Experiments using TerraSAR-X images showed that the proposed method outperforms compared methods, yielding an F1-score of 96.95% and an intersection over union (IoU) of 94.08%. The code is available at https://github.com/onesar/CVMNet.

The complex and diverse geological structures on the lunar surface serve as a direct record of its long evolutionary history. Comprehensively identifying and classifying these geological structures can not only deepen our understanding of lunar evolution, but also support the planning of future lunar exploration missions and guide the detection of lunar energy resources. However, the diverse and intricate morphologies of lunar geological structures, coupled with certain similarities between different formations, make their automatic identification and classification a significant challenge. To address this issue, we propose a multi-type lunar geological structure recognition network based on cross-view constraints, which mines differentiated feature information to enhance target identification and discrimination capabilities. This network effectively extracts heterogeneous and complementary features through cross-view feature extraction constraints. By employing a discrepancy-weighted loss function, the network focuses on regions where discrepancies arise in the recognition results across multiple views, thereby enhancing attention to divergent areas and learning feature representations in complex scenarios. Additionally, multi-scale contextual information aggregation combines contextual features from different receptive fields, leveraging surrounding terrain to enhance discriminative information for lunar geological structures. Experimental results demonstrate that the proposed method exhibits significant superiority in the task of identifying and classifying multi-type lunar geological structures, with a 1.9% improvement in mIoU.

Soil moisture is an important parameter affecting environmental processes such as hydrology, ecology and climate. Microwave remote sensing is an effective means of surface soil moisture measurement. Aiming at the influence of vegetation cover in the process of surface soil moisture inversion of winter wheat farmland by microwave remote sensing, a cooperative inversion method using multi-source remote sensing data is proposed in this paper. Thirty-three feature parameters are extracted from Radarsat-2 full polarization SAR data and Sentinel-2 optical data, and ten parameters with high correlation with soil moisture are selected to participate in soil moisture inversion by Pearson correlation analysis. Combined with the ground sampling data, four machine learning models, including Random Forest, Generalized Regression Neural Network, Radial Basis Function and Extreme Learning Machine, are used for quantitative inversion of soil moisture to reduce the impact of vegetation and improve the inversion accuracy. The experimental results show that the Random Forest model is the optimal. The average of determination coefficient is 0.63959, and the average of root mean square error is 0.0317 cm3/ cm3, which provides a reference for the inversion of soil moisture in farmland using multi-source remote sensing data.

The lower reaches of the ancient Yellow River (AYR) migrated very frequently, like a loong swinging its tail on the land of China. As the origin of Chinese civilization, AYR provided natural conditions for the people to thrive. However, the sediment carried by AYR still has a negative impact on local agricultural production. This letter, based on Gaofen-3, extracted scattering characteristics of archeological anomalies, detecting part of AYR during Song and Jin Dynasties, which was confirmed with field investigations. Then, an adaptive irregular convolution kernel U-Net (AICK-U-Net) was proposed to reconstruct the channel of AYR, based on the images obtained by the different polarization decomposition methods in October, and the precision and recall reached 96.21% and 94.45%, respectively. Finally, two decision-level methods were proposed to optimize the reconstruction results, improving the precision and recall to 96.39% and 97.36%, respectively. In summary, Spaceborne synthetic aperture radar (SAR), with the application of polarization decomposition and neural network, provides new insights for detecting archeological anomalies and reconstructing archaeolandscapes.

Synthetic aperture radar (SAR) often shares spectrum with other systems, such as radio, TV, cellular network, and so on, which are likely to produce radio frequency interference (RFI). Pulse RFI, which hinders SAR signal processing and image interpretation severely, is a common form of RFI and cannot be neglected. The simple and easy-to-implement frequency-domain notch filtering (FNF) method has been widely used to mitigate narrowband pulse RFIs. However, a well-known problem with abnormal sidelobe effect, which is caused by missing spectrum gaps due to the notch operation, is aroused when using FNF. In this letter, a novel time-domain notch filtering (TNF) is proposed. In the proposed method, pulse RFI occurrences are detected and notched by a simple log-ratio operator in a pulse by pulse manner. Then, missing-data iterative adaptive approach (MIAA) is performed to recover the notched signal to avoid ghosts. Experimental results via simulated and real L-band airborne SAR raw data validate the performance of the proposed method.

The equivalent rotation vector synthesized by 3-D nonuniform rotation of ship target leads to the time-varying characteristic of Doppler frequency of ship echo data, which will cause the image to be defocused in azimuth. In order to obtain high-resolution inverse synthetic aperture radar (ISAR) images of ship targets, we developed an optimal time selection algorithm based on time-frequency analysis of multiple scatterers. In this letter, we addressed the challenges of the serious cross-term interference in time-frequency analysis and measuring the stability of Doppler frequency. The time interval with a minimum variation of Doppler frequency was determined by using time-frequency analysis of multiple isolated scatterers and root mean squared error (RMSE). The effectiveness and robustness of the proposed algorithm were evaluated by both simulated and real ISAR data.