Synthetic Aperture Radar Imagery Research Articles

Rainband‐Occurrence Probability in Northern Hemisphere Tropical Cyclones by Synthetic Aperture Radar Imagery

AbstractRainbands are essential to tropical cyclones (TCs), significantly affecting TC structure and intensity change. High‐resolution synthetic aperture radar (SAR) imagery can capture the footprints of rainbands caused by rain‐induced sea surface roughness modification. Using 464 SAR TC images, we investigated the rainband‐occurrence probability of TCs under different hemispheres, local times (LTs), intensities, and ocean basins. Results show that the rainband‐occurrence probability is highest in the downshear‐left quadrant for Northern Hemisphere TCs (downshear‐right quadrant for Southern Hemisphere TCs). For Northern Hemisphere TCs, the rainband‐occurrence probability is overall higher in the early morning (LT), and the peak region of rainband‐occurrence probability appears farther from the TC center in the evening (LT). Compared with weak TCs, the rainband‐occurrence probability becomes higher for strong TCs in the Northern Hemisphere. Furthermore, TCs have a higher rainband‐occurrence probability in the Northwest Pacific than in the North Atlantic and Northeast Pacific.

Deep learning techniques for enhanced sea-ice types classification in the Beaufort Sea via SAR imagery

This study proposes a dual-branch encoder U-Net (DBU-Net) deep learning model to classify sea ice types based on synthetic aperture radar (SAR) images in the Beaufort Sea. The DBU-Net can segment multi-year ice (MYI), first-year ice (FYI), open water (OW), and leads on SAR images. We design a dual-branch encoder to fuse the polarization and the grey-level co-occurrence matrix (GLCM) information of SAR images to improve the model's classification capability. The model is subsequently fine-tuned using lead samples to identify leads. 24 Sentinel-1 SAR images acquired in the Beaufort Sea are utilized for model training and testing. The accuracy (Acc), mean intersection over union (mIoU), and kappa coefficient (Kappa) are employed as evaluation metrics. Experiments show that DBU-Net achieves 91.83%/0.841/0.849 in Acc/mIoU/Kappa in classifying MYI, FYI, and OW, significantly outperforming three traditional models based on support vector machine, random forest, or convolutional neural network. Compared with the original U-Net, the dual-branch encoder and the GLCMs improve 1.45%/4.4%/2.8% in Acc/mIoU/Kappa in MYI, FYI, and OW. Acc/mIoU/Kappa metrics of leads detection is 99.49%/0.801/0.754. Besides, 454 Sentinel-1 SAR images are fed into the optimal DBU-Net to generate 80 m sea ice products in the Beaufort Sea for winters 2018–2022. As the MYI draws wide attention and the FYI and MYI are complementary in the area during the Winter, we discuss the variation of MYI based on the generated sea ice products and explore the relationship between MYI's variation and the Beaufort High. We found that the MYI export in the 2018/19 winter was due to large summer sea ice remains and the abnormal sea ice motion caused by the southeast shifting Beaufort Atmospheric Pressure High (Beaufort High). The MYI import in the 2020/21 winter was due to a strong northward MYI import caused by the powerful Beaufort High.

Trend Research on Maritime Autonomous Surface Ships (MASSs) Based on Shipboard Electronics: Focusing on Text Mining and Network Analysis

The growing adoption of electric propulsion systems in Maritime Autonomous Surface Ships (MASSs) necessitates advancements in shipboard electronics for safe, efficient, and reliable operation. These advancements are crucial for tasks such as real-time sensor data processing, control algorithms for autonomous navigation, and robust decision-making capabilities. This study investigates research trends in MASSs, using bibliographic analysis to identify policy and future research directions in this evolving field. We analyze 3363 MASS-related articles from the Web of Science database, employing co-occurrence word analysis and latent Dirichlet allocation (LDA) topic modeling. The findings reveal a rapidly growing field dominated by image recognition research. Keywords such as “datum”, “image”, and “detection” suggest a focus on collecting and analyzing marine data, particularly with deep learning for synthetic aperture radar imagery. LDA confirms this, with “image analysis and classification research” as the leading topic. The study also identifies national and organizational leaders in MASS research. However, research on Arctic routes lags behind that on other areas. This work provides valuable insights for policymakers and researchers, promoting a deeper understanding of MASSs and informing future policy and research agendas regarding the integration of electric propulsion systems within the maritime industry.

On the Initial Phase of the Ongoing Unrest at Campi Flegrei and Its Relation with Subsidence at Vesuvio (Italy)

The densely inhabited area of Naples (Italy), between the Campi Flegrei and Vesuvio volcanoes, is one of the most hazardous regions in the world. After two decades of sustained subsidence, Campi Flegrei has been experiencing an accelerating uplift since 2005. The uplift is currently associated with unusual seismicity and increased degassing. To try to identify the cause of the shift from subsidence to uplift and explore any connection between Campi Flegrei and Vesuvio, we analysed the ground displacement time series of the two volcanoes from 1993 to 2010, obtained from ERS/ENVISAT Synthetic Aperture Radar imagery. To distinguish between the various sources of deformation, we used simple scatter plots and a blind source separation technique called variational Bayesian independent component analysis (vbICA). We obtained consistent results using both approaches. Specifically, with vbICA, we identified two significant independent components (ICs). IC1 describes the subsidence that occurred at Campi Flegrei prior to 2000, including the mini-uplifts of 2000 and 2005, and part of the post-2005 uplift. The expansion and contraction of two volumes beneath Campi Flegrei satisfy IC1: a sill-shaped volume at a depth of approximately 3 km and a small volume at a depth of 1–2 km, respectively. The two sources of deformation reproduce the large-scale deformation in the Campi Flegrei area and the local deformation in the Solfatara area, respectively. In the Campi Flegrei area, IC2 exhibits primarily uplift, which is concentrated in the eastern part of the caldera. The deformation pattern is complex and difficult to interpret. If we model it using simple spheroidal deformation sources, the pattern suggests that two volumes at depths of approximately 9 and 8 km are experiencing opposite activity, namely contraction (beneath the southwestern part of the caldera) and expansion (beneath the central part of the caldera). In the Vesuvio area, IC2 is consistent with the deformation induced by the contraction of a volume at a depth of around 9 km. The contraction beneath Vesuvio is smaller in magnitude than the expansion/contraction beneath Campi Flegrei. The correlation observed after 2002 between uplift at Campi Flegrei and subsidence at Vesuvio suggests the transfer of magma and/or magmatic fluids between the two plumbing systems at 8–9 km depth. This implies that part of the ongoing unrest at Campi Flegrei may have been promoted by mass transfer from below Vesuvio.

Analysis of Multiple Scattering Characteristics of Cable-Stayed Bridges with Multi-band SAR

Abstract. Accurate localization of multi-scattering features of cable-stayed bridges in multi-band Synthetic Aperture Radar (SAR) imagery is crucial for intelligent recognition of bridge targets within images, as well as for precise water level extraction. This study focuses on the Badong Yangtze River Bridge, utilizing Unmanned Aerial Vehicle (UAV) LiDAR data of the bridge, and analyzes the multi-scattering characteristics of different bridge structural targets based on Geometric Optics (GO) methods and the Range-Doppler principle. Furthermore, the study integrates LiDAR data of the bridge's cable-stays to examine their multi-scattering phenomena, finding that the undulations of the Yangtze River's surface waves significantly contribute to the pronounced double scattering features of the bridge's cable-stays. Additionally, statistical analysis of multi-source SAR data indicates that this phenomenon is not directly correlated with radar wavelength, implying no direct connection to surface roughness. Utilizing LiDAR point cloud data from the bridge's street lamps, this paper proposes a novel method for estimating water level elevation by identifying the center position of spots formed by double scattering from lamp posts. The results show that using TerraSAR ascending and descending orbit images, this method achieves a water level elevation accuracy of approximately 0.2 meters.

An Advanced Quality Assessment and Monitoring of ESA Sentinel-1 SAR Products via the CyCLOPS Infrastructure in the Southeastern Mediterranean Region

The Cyprus Continuously Operating Natural Hazards Monitoring and Prevention System, abbreviated CyCLOPS, is a national strategic research infrastructure devoted to systematically studying geohazards in Cyprus and the Eastern Mediterranean, Middle East, and North Africa (EMMENA) region. Amongst others, CyCLOPS comprises six permanent sites, each housing a Tier-1 GNSS reference station co-located with two calibration-grade corner reflectors (CRs). The latter are strategically positioned to account for both the ascending and descending tracks of SAR satellite missions, including the ESA’s Sentinel-1. As of June 2021, CyCLOPS has reached full operational capacity and plays a crucial role in monitoring the geodynamic regime within the southeastern Mediterranean area. Additionally, it actively tracks landslides occurring in the western part of Cyprus. Although CyCLOPS primarily concentrates on geohazard monitoring, its infrastructure is also configured to facilitate the radiometric calibration and geometric validation of Synthetic Aperture Radar (SAR) imagery. Consequently, this study evaluates the performance of Sentinel-1A SAR by exploiting the CyCLOPS network to determine key parameters including spatial resolution, sidelobe levels, Radar Cross-Section (RCS), Signal-to-Clutter Ratio (SCR), phase stability, and localization accuracy, through Point Target Analysis (PTA). The findings reveal the effectiveness of the CyCLOPS infrastructure to maintain high-quality radiometric parameters in SAR imagery, with consistent spatial resolution, controlled sidelobe levels, and reliable RCS and SCR values that closely adhere to theoretical expectations. With over two years of operational data, these findings enhance the understanding of Sentinel-1 SAR product quality and affirm CyCLOPS infrastructure’s reliability.

Liquefaction ground deformations and cascading coastal flood hazard in the 2023 Kahramanmaraş earthquake sequence

The 2023 Kahramanmaraş earthquake sequence produced extensive liquefaction-induced ground deformations and ongoing flooding along the shoreline of the Mediterranean port city of İskenderun, Türkiye. This study compiles field observations and analyses from cross-disciplinary perspectives to investigate whether earthquake-induced liquefaction was a significant factor for increasing the flood hazard in İskenderun. Geotechnical reconnaissance observations following the earthquakes included seaward lateral spreading, settlement beneath buildings, and failures of coastal infrastructure. Three presented lateral spreading case histories indicate consistent ground deformation patterns with areas of reclaimed land. Persistent scatterer interferometry (PSI) measurements from synthetic aperture radar (SAR) imagery identify a noticeably greater rate of pre- and post-earthquake subsidence within the İskenderun coastal and urban areas relative to the surrounding regions. The PSI measurements also indicate subsidence rates accelerated following the earthquakes and were typically highest near the observed liquefaction manifestations. These evaluations suggest that while the liquefaction of coastal reclaimed fill caused significant ground deformations in the shoreline area, ongoing subsidence of İskenderun and other factors likely also exacerbated the flood hazard. Insights from this work suggest the importance of evaluating multi-hazard liquefaction and flood consequences for enhancing the resilience of coastal cities.

A comparison of global flood models using Sentinel-1 and a change detection approach

AbstractAdvances in numerical algorithms, improvement of computational power and progress in remote sensing have led to the development of global flood models (GFMs), which promise to be a useful tool for large-scale flood risk management. However, performance and reliability of GFMs, especially in data-scarce regions, is still uncertain, as they are difficult to validate. Here we aim at contributing to develop alternative, more flexible, and consistent methods for GFM validation by applying a change detection analysis on synthetic aperture radar (CD-SAR) imagery obtained from the Sentinel-1 imagery, on a cloud-based geospatial analysis platform. The study addresses two main objectives. First, to validate four widely adopted GFMs with flood maps generated through the proposed CD-SAR approach. This exercise was conducted for eight different large river basins on four continents, to account for a diverse range of hydro-climatic environments. Second, to compare CD-SAR-derived flood maps with those obtained from alternative remote sensing sources. These comparative results offer valuable insights into the reliability of CD-SAR data as a validation tool, more specifically how it stacks up against flood maps generated by other remote sensing techniques.

SwinYOLOv7: Robust ship detection in complex synthetic aperture radar images

Using satellite-based SAR (Synthetic Aperture Radar) imagery to detect and track ships is a formidable challenge. However, accurate analysis is hampered by inherent difficulties such as obscured edges, multiple targets, varying dimensions and complex backgrounds. These factors contribute to a poor signal-to-noise ratio. Various artificial intelligence models have been developed to improve these problems, especially with YOLO-based models. To overcome the above challenges and achieve robust performance, this study presents a model called SwinYOLOv7, a novel fusion of the YOLOv7 framework with the Features Pyramid Network and the Swin Transformer using ground-breaking anchor-free detection algorithms. This innovative method aims to increase the accuracy of vessel detection while reducing the impact of background clutter. The proposed model improved by YOLOv7 examined three different datasets in detail: SRSDD-v1.0, HRSID, and SSDD to optimize the performance of the model. The training process was consistently verified using superior recall, precision, and F1-score values, which can be easily compared with previous studies. The results show that the model using the Swin Transformer attention mechanism and using an image size of 640×640 achieves the highest accuracy of 96.59%. Alternative attention mechanisms, including the Squeeze-and-Excitation Network (SEnet), the Convolutional Block Attention Module (CBAM), Channel Attention (CA) and Efficient Channel Attention (ECA), deliver poorer accuracy rates. The combination of YOLOv7 and Swin Transformer yielded encouraging results that enabled the proposed model to outperform the current benchmarking models. Therefore, the proposed model provides a compelling solution to ensure accurate vessel identification in complex search and rescue scenarios.

Where is my attention? An explainable AI exploration in water detection from SAR imagery

Attention mechanisms have found extensive application in Deep Neural Networks (DNNs), with numerous experiments over time showcasing their efficacy in improving the overall performance of DNNs. Considering the black-box nature of DNNs, it is unclear how attention mechanisms affect model decision-making processes. Therefore, it is of great significance to explore the internal relationship between the focused area of attention mechanisms and the prediction results. For the first time, an explainable framework for the attention mechanism of Synthetic Aperture Radar (SAR) image analytics is proposed, integrating DNN and eXplainable Artificial Intelligence (XAI), which realizes the effective explanation for attention mechanisms across different locations. The framework consists of three parts: the water extraction network which contains the attention mechanism at different locations; the proposed Attention- Gradient Class Activation Mapping (A-GCAM) method for attribution analysis of attention mechanisms; and we invent Category-specific channel Score of Confidence Mapping (CSCM) to perform geo-visualization for the output features of attention mechanisms. Experiments are conducted with the widely used Sentinel-1 system for water detection, in which three attention mechanisms with different characteristics are added to the encoder and decoder in DNNs. The results show that the framework can make the decision-making process of attention mechanisms transparent, thus improving their comprehensiveness and trustworthiness in various tasks and providing a reliable approach to selecting suitable attention mechanisms for the given task.

Estimating Landfill Landslide Probability Using SAR Satellite Products: A Novel Approach

This article presents a methodology for evaluating the susceptibility of landfill areas to develop landslides by analyzing Synthetic Aperture Radar (SAR) satellite products. The deformation velocity of the landfills is computed through the Persistent Scatterer Method on SAR imagery. These data, combined with a deformation model based on the shallow water equations (SWE), form the foundation for a Monte Carlo experiment that extrapolates the current state of the landfill into the future. The results of this simulation are then employed to determine the probability of a landslide occurrence. In order to validate the methodology effectiveness, a case study is conducted on a landfill in Zaldibar, Spain, revealing its effectiveness in estimating the probability of landfill landslides. This innovative approach emerges as an asset in large landfill management, acting as a proactive tool for identifying high-risk sites and preventing potential landslides, ultimately safeguarding human life and the environment. By providing insights into landslide probabilities, this study enhances decision-making processes and facilitates the development of intervention strategies in the domain of landfill risk assessment and management.

HyBNN: Quantifying and Optimizing Hardware Efficiency of Binary Neural Networks

Binary neural network (BNN), where both the weight and the activation values are represented with one bit, provides an attractive alternative to deploy highly efficient deep learning inference on resource-constrained edge devices. However, our investigation reveals that, to achieve satisfactory accuracy gains, state-of-the-art (SOTA) BNNs, such as FracBNN and ReActNet, usually have to incorporate various auxiliary floating-point components and increase the model size, which in turn degrades the hardware performance efficiency. In this article, we aim to quantify such hardware inefficiency in SOTA BNNs and further mitigate it with negligible accuracy loss. First, we observe that the auxiliary floating-point (AFP) components consume an average of 93% DSPs, 46% LUTs, and 62% FFs, among the entire BNN accelerator resource utilization. To mitigate such overhead, we propose a novel algorithm-hardware co-design, called FuseBNN , to fuse those AFP operators without hurting the accuracy. On average, FuseBNN reduces AFP resource utilization to 59% DSPs, 13% LUTs, and 16% FFs. Second, SOTA BNNs often use the compact MobileNetV1 as the backbone network but have to replace the lightweight 3 × 3 depth-wise convolution (DWC) with the 3 × 3 standard convolution (SC, e.g., in ReActNet and our ReActNet-adapted BaseBNN) or even more complex fractional 3 × 3 SC (e.g., in FracBNN) to bridge the accuracy gap. As a result, the model parameter size is significantly increased and becomes 2.25× larger than that of the 4-bit direct quantization with the original DWC (4-Bit-Net); the number of multiply-accumulate operations is also significantly increased so that the overall LUT resource usage of BaseBNN is almost the same as that of 4-Bit-Net. To address this issue, we propose HyBNN , where we binarize depth-wise separation convolution (DSC) blocks for the first time to decrease the model size and incorporate 4-bit DSC blocks to compensate for the accuracy loss. For the ship detection task in synthetic aperture radar imagery on the AMD-Xilinx ZCU102 FPGA, HyBNN achieves a detection accuracy of 94.8% and a detection speed of 615 frames per second (FPS), which is 6.8× faster than FuseBNN+ (94.9% accuracy) and 2.7× faster than 4-Bit-Net (95.9% accuracy). For image classification on the CIFAR-10 dataset on the AMD-Xilinx Ultra96-V2 FPGA, HyBNN achieves 1.5× speedup and 0.7% better accuracy over SOTA FracBNN.

Accuracy Assessment of Geometric-Distortion Identification Methods for Sentinel-1 Synthetic Aperture Radar Imagery in Highland Mountainous Regions.

SAR imagery plays a crucial role in geological and environmental monitoring, particularly in highland mountainous regions. However, inherent geometric distortions in SAR images often undermine the precision of remote sensing analyses. Accurately identifying and classifying these distortions is key to analyzing their origins and enhancing the quality and accuracy of monitoring efforts. While the layover and shadow map (LSM) approach is commonly utilized to identify distortions, it falls short in classifying subtle ones. This study introduces a novel LSM ground-range slope (LG) method, tailored for the refined identification of minor distortions to augment the LSM approach. We implemented the LG method on Sentinel-1 SAR imagery from the tri-junction area where the Xiaojiang, Pudu, and Jinsha rivers converge at the Yunnan-Sichuan border. By comparing effective monitoring-point densities, we evaluated and validated traditional methods-LSM, R-Index, and P-NG-against the LG method. The LG method demonstrates superior performance in discriminating subtle distortions within complex terrains through its secondary classification process, which allows for precise and comprehensive recognition of geometric distortions. Furthermore, our research examines the impact of varying slope parameters during the classification process on the accuracy of distortion identification. This study addresses significant gaps in recognizing geometric distortions and lays a foundation for more precise SAR imagery analysis in complex geographic settings.

INVESTIGATING LOSS FUNCTIONS FOR SEGMENTING AND DETECTING SHIPS ON SAR IMAGERY

Abstract. In accordance with the United Nations (UN) Sustainable Development Goal (SDG) 16: Peace, Justice, and Strong Institutions, this study explores ship monitoring through the use of Synthetic Aperture Radar (SAR) for its potential applications to economic and security purposes. One method to extract ships through SAR-derived imagery is to employ the use of convolutional neural networks (CNN). However, the extraction of small features continues to be a challenging task for CNNs. To improve the performance in such cases, one way is to employ the use of an appropriate loss function, which helps guide the CNN model during training. In this paper, Focal Combo (FC) loss, a recent loss function designed for extreme class imbalance, will be investigated to analyze its effects when applied to ship extraction. In doing so, this paper also presents a thorough comparison of existing loss functions in their capability to segment and detect ships on SAR imagery. Making use of the U-Net model, our results demonstrate that by using FC loss we can observe an increase in segmentation of about 9% in terms of f3-score and a decrease in missed detections by about 17 ships (after post-processing) when compared to cross-entropy loss. Unfortunately, it has also shown a significant drop in precision of about 35% resulting in an additional 270 ships being incorrectly detected in the background. In future work, varying CNN models shall be tested to see if the pattern persists and several trials shall be conducted to assess consistency.

The rectangular tile classification model based on Sentinel integrated images enhances grassland mapping accuracy: A case study in Ordos, China.

Arid zone grassland is a crucial component of terrestrial ecosystems and plays a significant role in ecosystem protection and soil erosion prevention. However, accurately mapping grassland spatial information in arid zones presents a great challenge. The accuracy of remote sensing grassland mapping in arid zones is affected by spectral variability caused by the highly diverse landscapes. In this study, we explored the potential of a rectangular tile classification model, constructed using the random forest algorithm and integrated images from Sentinel-1A (synthetic aperture radar imagery) and Sentinel-2 (optical imagery), to enhance the accuracy of grassland mapping in the semiarid to arid regions of Ordos, China. Monthly Sentinel-1A median value images were synthesised, and four MODIS vegetation index mean value curves (NDVI, MSAVI, NDWI and NDBI) were used to determine the optimal synthesis time window for Sentinel-2 images. Seven experimental groups, including 14 experimental schemes based on the rectangular tile classification model and the traditional global classification model, were designed. By applying the rectangular tile classification model and Sentinel-integrated images, we successfully identified and extracted grasslands. The results showed the integration of vegetation index features and texture features improved the accuracy of grassland mapping. The overall accuracy of the Sentinel-integrated images from EXP7-2 was 88.23%, which was higher than the accuracy of the single sensor Sentinel-1A (53.52%) in EXP2-2 and Sentinel-2 (86.53%) in EXP5-2. In all seven experimental groups, the rectangular tile classification model was found to improve overall accuracy (OA) by 1.20% to 13.99% compared to the traditional global classification model. This paper presents novel perspectives and guidance for improving the accuracy of remote sensing mapping for land cover classification in arid zones with highly diverse landscapes. The study presents a flexible and scalable model within the Google Earth Engine framework, which can be readily customized and implemented in various geographical locations and time periods.

Using SAR-based products to calculate potato carbon uptake in a tropical Andean region

ABSTRACT Gross primary productivity (GPP) is an essential parameter to estimate the efficiency of carbon transfer in terrestrial ecosystems. The daily GPP has been monitored in the past using mainly optical satellite imagery. However, GPP has never been monitored using satellite Synthetic Aperture Radar (SAR) imagery. We evaluate the possibility of using Sentinel-1 SAR data to estimate GPP and compare with field GPP measurements using eddy covariance (EC) systems in three commercial potato fields, located in the Andean region of Colombia, using three different water irrigation levels. Raw and processed radar data from Sentinel-1 were compared with the daily and accumulated GPP from the EC. Results indicate that SAR data has a high correlation with the accumulated GPP measured by the EC in the field. The normalized radar backscatter and radar brilliance coefficients with VH polarization show a good correlation with the EC accumulated GPP in irrigated potato fields (R2: 0.77–0.81), while radar vegetation indices show a good correlation with the EC accumulated GPP in potato fields with no irrigation (R2≈0.82). In particular, the accumulated GPP during the whole potato crop cycle was estimated with good accuracy (~5% error with irrigation and ~10% error with no irrigation).

MMSeaIce: a collection of techniques for improving sea ice mapping with a multi-task model

Abstract. The AutoICE challenge, organized by multiple national and international agencies, seeks to advance the development of near-real-time sea ice products with improved spatial resolution, broader spatial and temporal coverage, and enhanced consistency. In this paper, we present a detailed description of our solutions and experimental results for the challenge. We have implemented an automated sea ice mapping pipeline based on a multi-task U-Net architecture, capable of predicting sea ice concentration (SIC), stage of development (SOD), and floe size (FLOE). The AI4Arctic dataset, which includes synthetic aperture radar (SAR) imagery, ancillary data, and ice-chart-derived label maps, is utilized for model training and evaluation. Among the submissions from over 30 teams worldwide, our team achieved the highest combined score of 86.3 %, as well as the highest scores on SIC (92.0 %) and SOD (88.6 %). Notably, the result analysis and ablation studies demonstrate that instead of model architecture design, a collection of strategies/techniques we employed led to substantial enhancement in accuracy, efficiency, and robustness within the realm of deep-learning-based sea ice mapping. Those techniques include input SAR variable downscaling, input feature selection, spatial–temporal encoding, and the choice of loss functions. By highlighting the various techniques employed and their impacts, we aim to underscore the scientific advancements achieved in our methodology.

‘Nowhere and no one is safe’: spatial analysis of damage to critical civilian infrastructure in the Gaza Strip during the first phase of the Israeli military campaign, 7 October to 22 November 2023

BackgroundSince the Hamas attacks in Israel on 7 October 2023, the Israeli military has launched an assault in the Gaza Strip, which included over 12,000 targets struck and over 25,000 tons of incendiary munitions used by 2 November 2023. The objectives of this study include: (1) the descriptive and inferential spatial analysis of damage to critical civilian infrastructure (health, education, and water facilities) across the Gaza Strip during the first phase of the military campaign, defined as 7 October to 22 November 2023 and (2) the analysis of damage clustering around critical civilian infrastructure to explore broader questions about Israel’s adherence to International Humanitarian Law (IHL).MethodsWe applied multi-temporal coherent change detection on Copernicus Sentinel 1-A Synthetic Aperture Radar (SAR) imagery to detect signals indicative of damage to the built environment through 22 November 2023. Specific locations of health, education, and water facilities were delineated using open-source building footprint and cross-checked with geocoded data from OCHA, OpenStreetMap, and Humanitarian OpenStreetMap Team. We then assessed the retrieval of damage at and with close proximity to sites of health, education, and water infrastructure in addition to designated evacuation corridors and civilian protection zones. The Global Moran’s I autocorrelation inference statistic was used to determine whether health, education, and water facility infrastructure damage was spatially random or clustered.ResultsDuring the period under investigation, in the entire Gaza Strip, 60.8% (n = 59) of health, 68.2% (n = 324) of education, and 42.1% (n = 64) of water facilities sustained infrastructure damage. Furthermore, 35.1% (n = 34) of health, 40.2% (n = 191) of education, and 36.8% (n = 56) of water facilities were functionally destroyed. Applying the Global Moran’s I spatial inference statistic to facilities demonstrated a high degree of damage clustering for all three types of critical civilian infrastructure, with Z-scores indicating < 1% likelihood of cluster damage occurring by random chance.ConclusionSpatial statistical analysis suggests widespread damage to critical civilian infrastructure that should have been provided protection under IHL. These findings raise serious allegations about the violation of IHL, especially in light of Israeli officials’ statements explicitly inciting violence and displacement and multiple widely reported acts of collective punishment.

Detecting rotated ships in SAR images using a streamlined ship detection network and gliding phases

ABSTRACT Deep learning methods have greatly promoted the ship detection performance in synthetic aperture radar (SAR) images. However, due to the different imaging mechanisms and target characteristics, detecting ships in SAR imagery precisely and efficiently is still challenging. To address this issue, a streamlined rotational ship detection network (SRSD-Net) is proposed in this letter. First, we construct a lightweight network using one-level feature and a receptive field block (RFB) to extract ship features with rich context. Second, a simple and effective representation for rotated ship bounding boxes using the minimum external horizontal box and two phase factors is proposed. Finally, to reduce the missed detections of small ships, k-nearest label assignment (KNLA) is designed to make all targets equally contribute to the network training. Extensive experiments are conducted on high-resolution SAR images dataset (HRSID) and the results show that the proposed method can achieve high detection accuracy with low computational cost.

YOLOv8-BYTE: Ship tracking algorithm using short-time sequence SAR images for disaster response leveraging GeoAI

Ship tracking technology is crucial for emergency rescue in the event of a disaster. Quickly identifying the position and status of vessels is vital for rescue teams to be able to deploy efficiently in disaster areas. When responding to emergencies or natural disasters, ship tracking technology plays a critical role in supporting emergency rescue operations and resource allocation, improving the overall resilience of the maritime transportation system. However, the research on multi-object tracking (MOT) algorithms has primarily focused on optical image datasets. In contrast, image data from synthetic aperture radar (SAR) presents unique challenges, such as defocus interference, a high false alarm rate, and a lack of prior samples. To overcome these particular challenges, we propose a robust MOT algorithm developed for SAR images to achieve effective multi-vessel tracking under difficult imaging conditions. In particular, we optimize the YOLOv8 detection network by introducing a diffusion model-based training method for data augmentation. This method improves the robustness of the network to scaling, rotational and translational deformations. Moreover, an enhanced swin transformer is proposed as a feature extraction network, which strengthens the representation capability of the detection network. Furthermore, the state parameters within the KF technique are enhanced by directly capturing the details of the height and width of the tracking rectangle box. This refinement of the ByteTrack algorithm aims to achieve a more precise and accurate fit of the tracking rectangle to the ship, further improving the overall tracking performance. The experimental results from the ship detection and multiple objects tracking datasets show the impressive performance of the proposed model. With a precision of 97.60%, a recall of 96.36%, and an average precision of 96.72%, the model achieves exceptional detection accuracy with an 18% reduction in model parameters. Furthermore, significant improvements can be observed in key tracking metrics such as HOTA, MOTA and IDF1, with improvements of 4.8%, 8.5% and 6.8% respectively compared to the baseline algorithm, alongside a remarkable 37.5% reduction in IDS. It is noteworthy that the tracker works in real time, achieving an average analysis speed of 47 frames per second. The proposed MOT algorithm achieves state-of-art tracking performance on a SAR image dataset with short time sequences. Therefore, the proposed approach is a compelling solution for ship tracking in SAR imagery.