Conference Agenda

Spaceborne SAR is increasingly expected to support robust characterization of complex environments under limited or uncertain reference information. This challenge is particularly acute in tropical regions, where structurally heterogeneous forests, flooded systems, and forest-wetland mosaics often exhibit overlapping signatures and sparse field validation. In this study, we assess the respective contributions of spaceborne P-band and C-band SAR to the discrimination of tropical ecosystem classes in Brazilian Amazonia, using the first BIOMASS acquisitions together with Sentinel-1 observations.

A central component of the work is a noise-labelling pre-processing workflow designed to derive high-confidence training and evaluation samples from imperfect land-cover references. The approach combines spatial homogeneity constraints, temporal consistency in long-term land-cover histories, and physical plausibility checks based on auxiliary optical and terrain information. This enables the selection of reliable proxy samples despite pixel-level uncertainty in existing thematic maps.

For the filtered samples, we extract BIOMASS DGM and Sentinel-1 GRD backscatter features, derive polarization ratios and simple texture metrics, and evaluate within-class variability and between-class separability across forest, flooded forest, and forest-wetland transition classes. Particular attention is given to class boundaries that are difficult to resolve using conventional frequency bands alone and to the robustness of discrimination under sparse ground-truth conditions.

Preliminary results from the Brazilian test site indicate that P-band reduces within-class variance for forested classes and improves separability between terra firme and flooded forest systems relative to C-band alone, while C-band remains competitive for some open and anthropogenic covers. By extending the comparison to additional Amazonian sites and a richer ecosystem legend, the study provides an early cross-frequency assessment of how advanced feature extraction and confidence-based sample filtering can improve SAR-based classification of heterogeneous tropical environments. The proposed workflow is intended as a transferable strategy for difficult classification settings where label quality is a limiting factor.

The timely and accurate monitoring of water in the world's rivers is of critical importance for people, nature, agriculture and industry. The need for these measurements is increased considerably in times of flood and drought, and in the face of rapidly changing global climates and increasing human populations. Despite this, globally, river measurement networks are in decline, and those that do exist are increasingly overwhelmed by large scale flood events.

In this context, improving capabilities for the observation of river levels and flow rates using earth-observing satellites is an urgent priority.

Although satellite methods have been employed for at least 30 years, only recently have large constellations of very high-resolution satellites emerged, bringing the potential to greatly improve measurement capabilities. Whilst the advantages of very high spatial resolution are easily understood, the benefits to the timeliness of observations are perhaps equally significant. Acquisitions can be tasked with extremely short lead times and sub-daily repeats are possible, allowing the near real-time observation of natural hazard events such as floods. Largely operated by commercial entities however, access to these satellite systems, and hence the development of river-observing solutions, has been limited.

This presentation discusses advances in river observations made with high-resolution optical video products from Planet Labs, and some preliminary findings from an exploration of very high-resolution synthetic aperture radar data products from commercial satellite operator ICEYE for river science.

Abstract:

Landslides represent some of the most destructive natural hazards encountered in unstable mountainous regions, such as the Western Ghats in India. The examination of landslides has garnered significant global attention due to their profound impacts on socio-economic activities. The utilization of remote sensing and geographic information systems has proven valuable for integrating the spatial factors that contribute to landslide occurrences. In this study, satellite imagery from Sentinel 1-C Band has been employed, and further interferometry techniques for detecting the landslide event in 2024 in the Wayanad district, Kerala. Leveraging Artificial Intelligence Techniques in RADAR remote sensing such as machine learning algorithms, particularly the Random Forest (RF) model, were utilized to classify the study area into affected and non-affected regions. The findings also indicate the affected land use and land cover in the given study area. In the end, it can be concluded that significant landslides on 30th July, 2024 in the Wayanad district were primarily precipitated by anthropogenic interventions, compounded by heavy precipitation and unstable topography. Activities such as stone quarrying and infrastructure development emerged as critical factors contributing to these landslides. This research provides valuable insights aimed at mitigating landslide hazards in the Wayanad district, thereby fostering sustainable development.

Keywords: InSAR, artificial intelligence, landslides, SVM, Wayanad district, machine learning

Detecting objects lying beneath a forest cover using SAR measurements represents a major challenge, due to the response of the overlying vegetation volume, wave attenuation caused by propagation through the forest canopy, and high-intensity scattering mechanisms occurring at the ground level. Polarimetric SAR Tomography (PolTomoSAR) represents a promising solution to address these limitations, by leveraging both polarimetric and spatial diversities to discriminte objects from their background. This work investigates two approaches based on PolTomoSAR processing, and adapted to different tomographic acquisition configurations, i.e. different vertical resolution and ambiguity compromises.

The first method relies on PolTomoSAR data that feature high vertical resolution and a wide unambiguous elevation range. Full Rank PolTomoSAR focusing techniques are employed to isolate, with a high resolution, scattering sources located a few meters above the ground, and estimate their full-rank polarimetric responses. The concealed object detection is then conducted, based on polarimetric parameters provided by classical decomposition techniques. A simple detector, combining a few source descriptors, such as the polarimetric entropy and indicators of double-bounce scattering, as well as the elevation information, proves effective in identifying artificial objects embedded in dense, masking vegetation.

The second approach considers a compact configuration consisting solely of a two-image PolInSAR acquisition. No full tomographic inversion is possible, and elevation discrimination relies entirely on the interferometric phase diversity between the two acquisitions. Ground-notched InSAR processing is applied to suppress ground-scattering contributions, while Canopy-notched processing is applied to reduce the contribution of the forest canopy, whose polarimetric and radiometric features may prevent the detection of objects. This provides a filtered image representing a possibly ambiguous sampling of the scene reflectivity in the vertical direction. In the context of concealed object detection, the choice of the interferometric baseline separating the acquisition trajectories is crucial, as it balances the suppression of the forest canopy contribution and the preservation of responses from above-ground objects.

Both methods are evaluated using a 21-image fully polarimetric L-band data set, acquired by the DLR F-SAR sensor over Dornstetten, Germany. The study site consists of a mixed forest area containing several man-made objects, such as vehicles, containers, and corner reflectors, that are deployed both inside and outside the forest. Results demonstrate successful detection of concealed objects for varying baseline configurations.

Maritime surveillance based on Synthetic Aperture Radar (SAR) can be used for applications such as maritime traffic monitoring, environmental protection, and dark vessel detection. However, this is challenging due to the complex nature of different maritime targets, the relative motion between the sensor and the objects, and the presence of strong sea clutter. Furthermore, the motion of the ships in the ocean can introduce smearing and alter the true location of the target under study, and the vibration of their engines adds a micro-Doppler effect, that result in a superimposed phase modulation on the returned signal. Nevertheless, the distinct reflectivity of these objects compared to the sea surface makes SAR imagery appealing for detection and further analysis.

This work presents an end-to-end processing pipeline for automatic location, characterisation, and classification of vibrating maritime targets in SAR data using high-order features. The proposed methodology works on Single Look Complex (SLC) SAR images and aims at detecting potential targets, extracting micro-vibration signatures, and classifying objects based on their dynamic behaviour. The method is able to capture micro-motions signatures that are not detected in standard SAR processing, and uses them to distinguish between vibrating-non vibrating objects.

The pipeline starts from a focused SLC image that is transformed into the Range-Doppler domain in order to expose the micro-Doppler effects that are linked to vibrations. Candidate targets are detected using a Constant False Alarm Rate (CFAR) detector to suppress clutter while preserving strong scatterers, and refinement of the candidates is performed using short-time spectral analysis. From the selected range profiles, high-order features derived from autocorrelation representations of the signals are extracted to characterise vibrational patterns (or lack thereof) and capture the underlying micro-motion signature for classification.

The framework is evaluated using both synthetic and real SAR data, with experimental results showing that the proposed high-order feature characterisation improves the separability between vibrating and non-vibrating targets, leading to a reliable classification even in the presence of strong clutter and noise. Moreover, validation using real SAR data demonstrates the feasibility of detecting vibration signatures and distinguishing active vessels from static objects.

Overall, this works highlights the potential of high-order features extracted from the autocorrelation representation of range profiles for enhance maritime target detection and classification in SAR imagery. The developed pipeline offers an approach for exploiting vibrational signatures in operational SAR data and offers a promising direction for advancing in the field of automated maritime surveillance systems.

Operational monitoring of critical infrastructure using InSAR requires robust, reproducible processing workflows capable of integrating multi-frequency SAR data for cross-validation and uncertainty quantification. However, existing InSAR processing tools lack unified frameworks for comparative multi-sensor analysis, requiring separate installations, configurations, and expertise for each sensor-processor combination (Piter et al., 2024). We present the first fully containerized multi-sensor InSAR framework integrating C-band (Sentinel-1) and X-band (COSMO-SkyMed) processing within a single Docker environment, enabling standardized comparative displacement time series analysis with minimal user intervention.

The framework consolidates five specialized conda environments (ISCE, MintPy and Dolphin for Sentinel-1; ISCE and MintPy for COSMO-SkyMed) into a multi-core processing container (Yunjun et al., 2019) requiring only a configuration file and study area boundary for execution. Sentinel-1 data acquisition is fully automated via ASF Earthdata APIs for any temporal window from 2015 to present, while user-provided CSK imagery undergoes parallel processing using identical MintPy time series inversion workflows, ensuring methodological consistency for direct sensor comparison. The docker generates geocoded vertical velocity maps (GeoTIFF format) and displacement time series (CSV/PNG) for user-specified coordinates, enabling systematic monitoring of infrastructure deformation at any location within the study area.

We demonstrate the framework on the Campolattaro Dam (Benevento, Italy), a €220M infrastructure project under Italy's National Recovery and Resiliency Plan serving 2.5 million people. Processing ~120 Sentinel-1 acquisitions (Nov 2024-Oct 2025) and 14 COSMO-SkyMed images (Nov 2024-Aug 2025) produced vertical velocity maps revealing systematic measurement differences: Sentinel-1 exhibits millimeter-scale stability (±1 mm) while COSMO-SkyMed shows greater variability (-2.0 to +0.7 mm), consistent with X-band sensitivity to surface scattering changes and atmospheric effects documented in comparative studies. Displacement time series extracted at three dam monitoring points demonstrate the framework's capacity for long-term deformation tracking. The containerized approach eliminated environment configuration challenges and enabled complete workflow reproducibility, addressing critical gaps in operational InSAR deployment.

References:

Piter, A., Haghshenas Haghighi, M., & Motagh, M. (2024). Challenges and Opportunities of Sentinel-1 InSAR for Transport Infrastructure Monitoring. PFG – Journal of Photogrammetry, Remote Sensing and Geoinformation Science, 92, 609-627. https://doi.org/10.1007/s41064-024-00314-x

Yunjun, Z., Fattahi, H., & Amelung, F. (2019). Small baseline InSAR time series analysis: Unwrapping error correction and noise reduction. Computers & Geosciences, 133, 104331. https://doi.org/10.1016/j.cageo.2019.104331

This paper introduces the use of distributed SAR to enhance micro-motion frequency resolution, while not suffering of typical strict time synchronization requirements of distributed SAR. By combining SAR data acquired at slightly different times from multiple sensors, the method effectively relaxes the usual trade-off between time and frequency resolution imposed by the uncertainty principle. The focus of this paper is to demonstrate that distributed SAR is well suited for micro-Doppler processing, with far fewer constraints than in imaging or interferometry.

This poster proposes a method of precise micro-Doppler analysis of vibrating targets in synthetic aperture radar (SAR) imagery. The raw SAR data is first processed using a modified backprojection approach to generate a time series of the ground target before a sparse representation approach using the orthogonal matching pursuit algorithm is utilized to de-noise the data. By incorporating the actual pulse timings from the radar metadata, during the dictionary construction, the method effectively handles the variable pulse repetition frequencies of modern SAR sensors. Experimental validation with real spaceborne SAR data of a scene containing an oscillating corner reflector demonstrates millimeter-level displacement accuracy, confirming the efficacy of the proposed approach for precise vibration measurement.

Conventionally, contact sensors such as accelerometers are used to conduct vibration-based structural health monitoring (VBSHM), however the deployment of such sensors is significantly limited by installation and maintenance challenges. Remote sensing alternatives for VBSHM have been proposed for alleviating monitoring costs, including: uncrewed aerial vehicles (UAVs) equipped with cameras for digital image correlation, and both ground-based lidar and millimetre-wave Doppler radar in either real or synthetic aperture modes – however these all require onsite proximity, limiting their coverage due to access and cost limitations.

Spaceborne sensing offers a less occluded vantage point from an orbiting platform, vastly improving sensor coverage. Spaceborne synthetic aperture radar (SAR) has been used in an SHM context, where interferometric SAR (InSAR) has been noted for utility in measuring the long-term displacement of structural elements. As InSAR functions by comparing multiple SAR acquisitions, however, its sampling rate is too low for VBSHM purposes. The alternative presented here is micro-Doppler SAR (MDSAR), which measures motion using a single SAR image by estimating the varying Doppler shifts from a vibrating target observed during the acquisition, which normally cause artefacts such as ghost targets in the image.

In this work time-series measurements and spectra of both calibration tests and a validation experiment conducted on a bridge will be presented. The input data are single-pass SAR images of real-world targets obtained through commercial SAR companies, with synchronous ground truth measured by conventional in-situ sensing. Calibration measurements of radar targets showed good agreement between time histories for velocities down to RMS 0.66 mm/s for 2 Hz oscillation with a 0.4 Hz amplitude modulation, and with frequency peaks consistently identified to within frequency resolution for all measurements.

Validation experiments were carried out on the South Portland Street Suspension Bridge in Glasgow, UK. The results are consistent with calibration tests and demonstrate the feasibility of measuring vibrational velocities as low as 1 mm/s with MDSAR. In the time domain, the average measurement error is approximately 1 mm/s, comparable to the true velocities of the bridge. In the frequency domain, the technique performs well in identifying the dominant vibrational frequency, with a residual less than the frequency resolution of 0.06 Hz determined by the SAR acquisition duration of 16 s. These promising results demonstrate that although MDSAR cannot yet supplant onsite SHM methods, it can provide valuable information integrable into hybrid systems and scope remains for refinement of measurements.

Modern Synthetic Aperture Radar (SAR) systems acquire high-resolution images over wide swaths, generating huge amount of data that need efficient compression for onboard storage and transmission to on-land processing centers. However, SAR images are inherently contaminated by multiplicative speckle noise, which complicates traditional lossy compression techniques. This study investigates the method of lossy compression of multilook SAR images using the Better Portable Graphics (BPG) coder. The primary objective is to evaluate and compare a direct compression approach against a method that applies a logarithmic Variance-Stabilizing Transform (VST) before compression and an inverse transform post-decompression. Efficient lossy compression with high edge preservation is critical for the rapid downlink and timely processing of high-resolution spaceborne SAR data required in time-sensitive security and situational awareness applications. In this context, the research focuses on identifying and analyzing the Optimal Operation Point (OOP)—the specific compression parameter where the decoded image exhibits the highest similarity to the noise-free original. Using test images with simulated speckle properties mirroring Sentinel-1 data, rate/distortion curves were constructed and evaluated using traditional peak signal-to-noise ratio (PSNR) and visual quality metrics (PSNR-HVS-M).

The findings demonstrate that an Optimal Operation Point exists for both the direct and VST-based compression methods, occurring at approximately the same compression control parameter irrespective of image complexity. The logarithmic VST effectively converts the signal-dependent multiplicative speckle into signal-independent additive noise with a nearly Gaussian distribution. While the direct compression approach yields slightly better metric values at the OOP and achieves larger compression ratios, the VST-based method offers superior preservation of edges, details, and textures. It significantly reduces the information smearing typically introduced by compression, particularly in low-intensity regions of the image. Both approaches enable efficient lossy compression in the OOP neighborhood, providing high image quality alongside compression ratios on the order of tens to hundreds. Ultimately, while direct compression maximizes the compression ratio, the VST-based approach favorably alters the statistics of introduced errors and is highly recommended when structural detail preservation is paramount.

Recent advances in Synthetic Aperture Radar (SAR) micro motion extraction have shown that additional insights about targets in a SAR scene can be obtained even from spaceborne sensors. Of particular interest are the micro motions induced by the internal combustion engines of vehicles and ships, which provide distinctive signatures with information that can support a variety of applications. In the maritime domain, engine characteristics and operational status offer valuable information for target discrimination and identification and can serve as proxy measurements for estimating Underwater Radiated Noise (URN).

The capability to monitor vessel engine behaviour from space enables several downstream applications, including critical infrastructure protection (e.g., harbours and subsea communication or pipeline assets), detection of Illegal, Unreported and Unregulated (IUU) fishing, and maritime regulation enforcement aimed at protecting marine wildlife and reducing emissions.

Currently, URN monitoring relies on sparse hydrophone networks and costly surveys, resulting in fragmentary datasets that limit regulatory, mitigation, and ecological utility. A global, high revisit, all weather observation capability from space, covering shipping lanes, marine protected areas, and offshore activity clusters, would close key measurement gaps in space and time. Such a capability would enhance accountability (e.g., detecting prohibited engine RPM ranges in sensitive habitats) and generate actionable insights for regulators, ecologists, and industry stakeholders.

In this work, we investigate how spaceborne SAR measurements can be combined with underwater observations to enable new multi modal sensing solutions for advanced maritime situation awareness and URN prediction. Results from two experimental campaigns, using cooperative and non cooperative vessels and combining SAR, underwater, and in situ vibrational measurements, will be presented. These results demonstrate how SAR can act as a complementary sensor, or even a potential substitute for underwater sensors, for both situational awareness and URN modelling and forecasting.

Detecting concealed ground targets such as wheeled and tracked vehicles, as well as infrastructure, from full-polarimetric SAR data remains a central challenge in foliage-penetrating (FOPEN) radar, particularly due to the near-total absence of annotated training data in operational settings. This work addresses the data bottleneck through a physics-based simulation pipeline that generates realistic L-band full-polarimetric SAR scenes of ground targets beneath forest canopy, explicitly incorporating sensor fidelity.

Target backscatter is modeled using a combination of geometric and polarimetric simulation workflows, with sigma-naught recalibration against empirical backscatter tables to ensure radiometric consistency with real acquisitions. Canopy attenuation is applied via a water cloud model with parameters calibrated to match simulated tree density. Importantly, the simulator is parameterized to reflect the physical characteristics of the target sensor, including range and azimuth resolution, pixel spacing, bandwidth, wavelength, apodization function, and incidence angle.

Simulated scenes are used to train a deep learning based detection method with two polarimetric input channels derived from the coherency structure: the double-bounce channel (HH-VV) and the surface/volume channel (HH+VV). This decomposition exploits the known scattering contrast between man-made targets and natural clutter at L-band, providing the network with physically interpretable features rather than raw SLC amplitudes. On held-out simulated scenes, the model achieves a mean average precision (mAP) exceeding 0.90 across all target classes. Confusion matrices indicate limited inter-class errors, with the primary failure mode being missed detections under dense canopy, consistent with the physical attenuation limits of L-band.

Overall, this work suggests that physically calibrated simulation offers a practical route for scalable dataset construction for FOPEN target detection, enabling approaches such as few-shot or semi-supervised learning, where simulation provides the majority of training diversity and real labeled data are reserved for fine-tuning or domain adaptation.

Harmful Algal Blooms (HABs) represent a growing threat to aquatic ecosystem and human health. Traditional monitoring methods based on optical remote sensing are fundamental but inherently limited by cloud cover and solar illumination. Synthetic Aperture Radar (SAR) technology, being effective regardless of weather and time of the day, offers a complementary approach. However, SAR backscatter over water is primarily sensitive to surface roughness, making direct bloom detection challenging and thus representing an underexplored frontier in environmental monitoring science.

This study explores the potential of detecting HABs by analyzing the error magnitude between observed Sentinel-1 SAR backscatter and a physically-based forward model. This model simulates expected backscatter values for lake water under normal condition, employing ancillary information such as wind speed and direction.

Three lakes with a history of recurrent blooms were analysed: Avernus Lake in Italy, Albufera Lake in Spain and Clear Lake in California. To assess the HABs detection potential, C-band SAR data from Sentinel-1 (both VV and VH polarizations) and multispectral data from Sentinel-2/MSI (for water-masking and spectral validation) were employed. The data analysis was entirely conducted on Google Earth Engine platform. The core method consisted of deriving an observation minus simulation difference, considering the most significant deviations as potential indicators of blooms, which were then validated through multispectral observations.

The study demonstrates that, under appropriate parametrization, the model-observation error from a physically-based SAR simulator can serve as an effective proxy for detecting HABs. It successfully identified increases in chlorophyll from the magnitude of the error between observed and simulated VV and VH values.

By bridging the gap between physical radar modelling and biological oceanography, this approach reinforces the potential of integrating SAR technology with multi/hyperspectral observations for robust HABs monitoring. This integration is particularly valuable in frequently cloud covered regions. Furthermore, it lays the groundwork for the development of hybrid physical-machine learning models that combine domain knowledge with data-driven learning, paving the way for a new generation of environmental impact assessment tools.

The increasing availability of very high-resolution (VHR) spaceborne Synthetic Aperture Radar (SAR) imagery is creating new opportunities for advanced maritime surveillance, particularly by enabling a more detailed characterization of vessel scattering mechanisms. Beyond primary radar returns, multipath effects—such as double- and triple-bounce reflections—carry valuable information about the vertical structure of ships. This work addresses the challenge of extracting such structural information by proposing a methodology for estimating ship vertical profiles through multipath analysis.

A two-dimensional geometric model is introduced to describe the relationship between multipath scattering signatures and the physical height of ship components. Specifically, the model links the slant-range separation of multipath returns to vertical features such as freeboard and mast height. Building on this theoretical framework, a dedicated processing chain is developed to retrieve and enhance these signatures from SAR imagery. The pipeline operates on ship image clips extracted from level-1 SAR products and incorporates inverse SAR (ISAR) autofocus to compensate for motion-induced defocusing when necessary. To ensure accurate detection of multipath contributions, adaptive sidelobe suppression techniques are employed alongside tailored detection logic, allowing weak secondary signals to emerge from the dominant target response.

The proposed methodology is validated using real VHR spotlight SAR data, including imagery acquired from Capella Space. Experimental results on vessels of the same dimensional class, specifically crude oil tankers, demonstrate the capability of the approach to estimate key vertical features and distinguish between different loading conditions, such as ballast versus full load. Overall, the framework provides a novel and effective tool for enhancing maritime situational awareness by leveraging previously underutilized multipath information in SAR data.

Plastic account for nearly 80% of marine debris worldwide. Due to its significant environmental impact, tracking plastic pollution on a large scale is an urgent need and global challenge.

In this context, experimental field campaigns are a fundamental tool for understanding the sensitivity of remotely sensed measurements to plastic litter. The use of a multi-sensor approach, which integrates data from multiple platforms, allows for the overcoming of the limitations of individual technologies.

This study investigates the sensitivity of remotely sensed measurements to floating plastic items, collected using measurements from different collocated remote sensing layers, including the satellite-based one, which consists of using X-band Cosmo-SkyMed 2nd Generation (CSG) Synthetic Aperture Radar (SAR), and the drone-based one, which consists of UAVs equipped with multispectral, LIDAR and thermal instruments.

The insights gained from experimental campaign are discussed in this study.

The campaign was conducted in July 2024, when a plastic target (2 m × 2 m) composed of heterogeneous synthetic materials was released on the Calore lake in the South of Italy.

The experiment confirms the ability of all platforms to observe aggregated plastic targets.

Under calm water conditions, the plastic items result in a SAR backscatter that is larger than the surrounding water and well above the system NESZ. Green and red multispectral bands are particularly effective at detecting plastics in freshwater environments. Thermal measurements show that plastics and the surrounding water have measurable thermal contrasts. LIDAR is able to distinguish between different types of plastic.

In SAR imagery moving targets appear displaced with regards to their actual position within the scene as a consequence of their radial motion relative to the platform. In cases that information on their true position is known (e.g. for naval targets the presence of a wake or AIS information), the radial velocity can be calculated from the displacement observed in the image, providing invaluable insights into target dynamics. However, reference ground truth information is not always available, making techniques to extract velocity information from solely SAR data of relevance for scene characterization. This presentation focuses on a novel technique to extract a target’s radial velocity directly from the image by leveraging two intentionally and differently mismatched filters in the azimuth compression step. With small mismatching values, the filters produce two slightly defocused images of the same target, in which it appears at different azimuth locations. From the relative azimuth displacement between the resulting images, the target radial velocity can be extracted and used to quantify the motion-induced displacement. With this information at hand, the target can also be relocated to its true position. The technique was tested on both simulated and real naval targets extracted from SAR imagery collected by Cosmo-SkyMed in stripmap mode. The validation on real data was possible thanks to the presence of wakes in the image indicating the actual position of the ships in the scene.

ONERA, the french aerospace lab, has been developping a drone-embedded SAR imaging tool, SAR-Light. With full polarimetry capabilities in Ku-, X-, C- and L/UHF- bands, it reaches performances similar to those of higher carriers, while the significantly lower payload puts a high constraint on the material. However this makes an easy-to-deploy, cheaper tool and enables more versatile trajectories. It has been extensively exploited on several use cases over the past few years and has reached a certain maturity. Here we display several real-life measurements in lower band, which is particularly adapted for its penetration capabilities. The first use case focuses on multiple pass change detection in a complex environment, which is particularly demanding on trajectory measurement accuracy and repeatability. The second case study focuses on landmine and buried object detection. We demonstrate the benefit of multipass detection in an operational context. Last but not least, despite the fact that the resolution and quality image parameters displayed were measured without autofocus during the characterisation campaigns, this latter technique can still bring significant improvement on image quality. We will put a special emphasis on this technical aspect at the end of the presentation, as a mean to improve data quality in operational conditions.

With a length of 4,200 km, the river plays a key role in the region. In the river basin, the installation and maintenance of physical stations has several limitations, including high construction and maintenance costs, difficult accessibility in remote areas or during periods of flooding, and risks to personnel in dangerous. The objective of this study is to evaluate the performance of the altimetry-based hydrological monitoring approach in the Niger River basin. The methodological approach consists of using the R*R correlation matrix, Pearson's coefficient, and RMSE to evaluate the performance of altimetry data and altimetry data acquired from the hydrowebnext platform for monitoring water levels at the R_NIGER_KM1635 virtual station. Sentinel-2 data from DEA Wofs products were used to calculate the lateral variation in surface area between 1990 and 2025. Slope data were obtained from Surface Water and Ocean Topography.

The results of the performance evaluation of water levels from observed and satellite data on the Niger River show an R*R of 0.899, an RMSE of 5.66, a bias of +5.62, and a Pearson matrix of 0.94. These indices reveal the performance of altimetry data in hydrological monitoring compared to in situ data, with an average vertical reference error of 5.62. Analysis of the river's lateral extent data shows that in 1991 and 1998, the river dried up, posing a risk to the water supply of the city of Niamey, for which the river is the main source. The slopes obtained from SWOT show that the slope is estimated at 6.36 10^-5 on the reach of the river section closest to the city of Niamey. At the end of this work, it was noted that spatial hydrology through altimetry and SWOT are strategic tools in the monitoring and management of water resources for the Niger River.

Mots-clés : Hydrologie, Altimetrie, SWOT, Fleuve, Niger