We present TerraMind, the first any-to-any generative, multimodal foundation model for Earth observation (EO). Unlike other multimodal models, TerraMind is pretrained on dual-scale representations combining both token-level and pixel-level data across modalities. On a token level, TerraMind encodes high-level contextual information to learn cross-modal relationships, while on a pixel level, TerraMind leverages fine-grained representations to capture critical spatial nuances. We pretrained TerraMind on nine geospatial modalities of a global, large-scale dataset. In this paper, we demonstrate that (i) TerraMind's dual-scale early fusion approach unlocks a range of zero-shot and few-shot applications for Earth observation, (ii) TerraMind introduces "Thinking-in-Modalities" (TiM) -- the capability of generating additional artificial data during finetuning and inference to improve the model output -- and (iii) TerraMind achieves beyond state-of-the-art performance in community-standard benchmarks for EO like PANGAEA. The pretraining dataset, the model weights, and our code are open-sourced under a permissive license.

Performing accurate confidence quantification and assessment in pixel-wise regression tasks, which are downstream applications of AI Foundation Models for Earth Observation (EO), is important for deep neural networks to predict their failures, improve their performance and enhance their capabilities in real-world applications, for their practical deployment. For pixel-wise regression tasks, specifically utilizing remote sensing data from satellite imagery in EO Foundation Models, confidence quantification is a critical challenge. The focus of this research is on developing a Foundation Model using EO satellite data that computes and assigns a confidence metric alongside regression outputs to improve the reliability and interpretability of predictions generated by deep neural networks. To this end, we develop, train and evaluate the proposed Confidence-Aware Regression Estimation (CARE) Foundation Model. Our model CARE computes and assigns confidence to regression results as downstream tasks of a Foundation Model for EO data, and performs a confidence-aware self-corrective learning method for the low-confidence regions. We evaluate the model CARE, and experimental results on multi-spectral data from the Copernicus Sentinel-2 constellation to estimate the building density (i.e. monitoring urban growth), show that the proposed method can be successfully applied to important regression problems in EO. We also show that our model CARE outperforms other methods.

Confidence assessments of semantic segmentation algorithms are important. Ideally, models should have the ability to predict in advance whether their output is likely to be incorrect. Assessing the confidence levels of model predictions in Earth observation (EO) classification is essential, as it can enhance semantic segmentation performance and help prevent further exploitation of the results in the case of erroneous prediction. The model we developed, Confidence Assessment for enhanced Semantic segmentation (CAS), evaluates confidence at both the segment and pixel levels, providing both labels and confidence scores as output. Our model, CAS, identifies segments with incorrectly predicted labels using the proposed combined confidence metric, refines the model, and enhances its performance. This work has significant applications, particularly in evaluating EO Foundation Models on semantic segmentation downstream tasks, such as land-cover classification using Sentinel-2 satellite data. The evaluation results show that this strategy is effective and that the proposed model CAS outperforms other baseline models.

Recent advances in remote sensing and artificial intelligence offer exciting opportunities in an array of fields, with precision agriculture being a notable use case. Here, estimating soil parameters from remotely-sensed hyperspectral imagery at a global scale can play a pivotal role in day-to-day operations, as it may help optimize agricultural management processes, hence positively affecting our planet. In this paper, we tackled the problem of estimating soil parameters from hyperspectral images and introduced heterogeneous regression ensembles for this task. They not only benefit from both classic and deep machine learning models but were also thoroughly investigated and fine-tuned in our rigorous experimental study, performed over a well-established HYPERVIEW benchmark dataset. The experiments showed that such heterogeneous ensembles outperform other techniques and offer a high level of model flexibility.

We investigate the runtime of quantum kernel estimation in the view of quantum kernel concentration effect. The study is performed for projected quantum kernel family evaluated on hyperspectral remote sensing data. The effect of exponential value concentration leads to the indistinguishability of the kernel matrix entries as the size of the quantum device grows. In order to prevent that, kernel values have to be estimated with a better precision. Increasing precision inevitably connects to an increasing number of circuit runs, which influences the runtime of quantum algorithm. This, in turn, frequently obstructs a possible advantage for quantum machine learning methods. We find that, against popular opinions, the effect of exponential value concentration does not rule out the utility of quantum kernel methods and the severity of the issue depends highly on the data used.

Artificial Intelligence (AI) applications on Earth Observation (EO) satellite data, such as those for vessel detection, are gaining attention for their potential to meet strict bandwidth and latency requirements. While traditional on-ground computing pipelines often rely on heavy post-processing, implementing these techniques onboard satellites is challenging due to limited computing resources. To support the development of efficient onboard data processing strategies, this study compares the performance of object detection on raw data from Sentinel-2 and VENμS missions. The study demonstrates that the proposed two-stage approach with a focus on efficiency is capable of identifying vessels in raw data with minimal pre-processing. Specifically, our method achieved a remarkable Average Precision (AP) of 0.841 on the VENμS dataset.

Bare soil detection is an important step in soil composition analysis, as it can prune the areas that should be excluded from more expensive processing aimed at extracting selected soil parameters from hyperspectral images acquired in orbit. This is of paramount importance for on-board applications, where hardware constraints of an edge device (a satellite), such as computational and memory requirements or energy consumption need to be considered while processing big data in space. In this paper, we present a simple yet effective bare soil detection algorithm exploiting vegetation indices that is ready for in-orbit deployment. Our experimental study performed over the airborne hyperspectral data shows that this approach can be robustly used for simulated bands, i.e., wide bands aggregating several narrow neighboring bands within the spectrum. Therefore, we can apply our technique to sensors with lower spectral resolution. Finally, it offers high-quality bare soil delineation reaching the Dice Index of 0.85.

An unprecedented amount of data generated in space missions triggers lots of practical challenges and concerns with its transfer, storage, and analysis. As Lunar and deep space missions emerge, we need to also face the challenges of distributed computing and big data analytics. In this paper, we outline these issues and discuss how to design and analyze Lunar data centers, being space data centers designed for distributed computing, and data analysis for (not only) Lunar missions. We investigate the opportunities and chances of such space architectures to lay the foundations for practical space data centers and real-life use cases.

Satellite imaging brings exciting opportunities in an array of fields, with precision agriculture being a notable example. Soil analysis at scale with the use of Earth observation satellites coupled with on-board and on-the-ground artificial intelligence algorithms offers actionable items that may be exploited by practitioners to optimize their operations, including the fertilization process. Here, bare soil detection is a pivotal step in the processing chain to limit the detailed analysis to the areas of interest. In this paper, we tackle this task with quantum-kernel support vector machines and verify the utility of quantum machine learning in practical Earth observation. Our experimental study, performed over a real-world hyperspectral scene, indicates that the proposed quantum-kernel models are competitive with well-established classical support vector machines, as well as with approaches based on thresholding spectral indices that are widely exploited in the field.

Super-resolution is aimed at reconstructing high-resolution images from low-resolution observations. State-of-the-art approaches underpinned with deep learning allow for obtaining outstanding results, generating images of high perceptual quality. However, it often remains unclear whether the reconstructed details are close to the actual ground-truth information and whether they constitute a more valuable source for image analysis algorithms. In the reported work, we address the latter problem, and we present our efforts toward learning super-resolution algorithms in a task-driven way to make them suitable for generating high-resolution images that can be exploited for automated image analysis. In the reported initial research, we propose a methodological approach for assessing the existing models that perform computer vision tasks in terms of whether they can be used for evaluating super-resolution reconstruction algorithms, as well as training them in a task-driven way. We support our analysis with experimental study and we expect it to establish a solid foundation for selecting appropriate computer vision tasks that will advance the capabilities of real-world super-resolution.

Floods are among the most common and devastating natural hazards, imposing immense costs on our society and economy due to their disastrous consequences. Recent progress in weather prediction and spaceborne flood mapping demonstrated the feasibility of anticipating extreme events and reliably detecting their catastrophic effects afterwards. However, these efforts are rarely linked to one another and there is a critical lack of datasets and benchmarks to enable the direct forecasting of flood extent. To resolve this issue, we curate a novel dataset enabling a timely prediction of flood extent. Furthermore, we provide a representative evaluation of state-of-the-art methods, structured into two benchmark tracks for forecasting flood inundation maps i) in general and ii) focused on coastal regions. Altogether, our dataset and benchmark provide a comprehensive platform for evaluating flood forecasts, enabling future solutions for this critical challenge. Data, code & models are shared at https://github.com/Multihuntr/GFF under a CC0 license.

Remote sensing of rainfall events is critical for both operational and scientific needs, including for example weather forecasting, extreme flood mitigation, water cycle monitoring, etc. Ground-based weather radars, such as NOAA’s Next-Generation Radar (NEXRAD), provide reflectivity and precipitation estimates of rainfall events. However, their observation range is limited to a few hundred kilometers, prompting the exploration of other remote sensing methods, particularly over the open ocean, that represents large areas not covered by land-based radars. Here we propose a deep learning approach to deliver a three-class segmentation of SAR observations in terms of rainfall regimes. SAR satellites deliver very high resolution observations with a global coverage. This seems particularly appealing to inform fine-scale rain-related patterns, such as those associated with convective cells with characteristic scales of a few kilometers. We demonstrate that a convolutional neural network trained on a collocated Sentinel-1/NEXRAD dataset clearly outperforms state-of-the-art filtering schemes such as the Koch’s filters. Our results indicate high performance in segmenting precipitation regimes, delineated by thresholds at 24.7, 31.5, and 38.8 dBZ. Compared to current methods that rely on Koch’s filters to draw binary rainfall maps, these multi-threshold learning-based models can provide rainfall estimation. They may be of interest in improving high-resolution SAR-derived wind fields, which are degraded by rainfall, and provide an additional tool for the study of rain cells.

The amount of data captured in the emerging satellite missions has been continuously growing. Thus, developing resource-efficient predictive models is of paramount importance in an array of onboard space applications, where downlinking the data for further analysis is extremely costly or impossible. In such scenarios, we should extract actionable items on board an edge device, using e.g., a machine learning model. Reducing the model’s complexity which may be significant in deep learning algorithms is not only about fitting a full-size neural net into resource-restrictive hardware, but it may result in decreasing the latency and energy consumption. We tackle this issue and exploit knowledge distillation to elaborate a simpler and memory-efficient version of the large-capacity learner, aiming to preserve the large model quality. The experimental study shows that knowledge distillation may not only improve the classification capability of the original model, but can also dramatically (up to 425×) reduce its size for the on-board classification of Mars imagery.

The Copernicus Hyperspectral Imaging Mission for the Environment (CHIME) will provide routine hyperspectral observations over the land and coastal zone through the Copernicus Program in support of EU- and related policies for the management of natural resources, assets and benefits [1] , [2] . CHIME is an operational mission covering land and coastal areas which already uses of well-established ground processing routines in place that is independent of the AI unit we are discussing here. Areas outside the nominal observation scenario call for new methodologies such as on-board processing which will bring advantages including: fast processing of data, reducing the amount of data to be down-linked especially over ocean that CHIME sensor will be on but there is no regular acquiring plan, capability of on demand request handling. However there are some drawbacks i.e. complex validation and setting up update routines. Therefore, there has been an assessment going on to have an artificial intelligence unit on board to process acquired data and select only part of the data that contains desired targets. Based on the importance of detecting floating plastic debris for applications such as fishing farms preservation, navigation, tourism, etc., in marine environment, this application has been selected to serve as one of the test cases to evaluate Artificial Intelligence (AI) on board [3] .

Hyperspectral sensors are increasing in popularity for Earth Observation applications due to their ability to gather data over multiple spectral bands. However, the processing of such amount of information is difficult to handle for the current computing capabilities of small satellites. Several Band Selection methodologies have been developed in the last years; although, some of them demand very low computational resources, they use, at least, Level 1 data products. Therefore, the Level 0 data needs to be processed and the spectral bands coregistered. Artificial Intelligence has shown its potential to reduce the computational burden while achieving high accuracies in EO applications. In this study, a Neural Network-based methodology is proposed to select a spectral band set directly using non coregistered data captured by hyperspectral sensors.

This paper introduces DTACSNet, a Convolutional Neural Network (CNN) model specifically developed for efficient onboard atmospheric correction and cloud detection in optical Earth observation satellites. The model is developed with Sentinel-2 data. Through a comparative analysis with the operational Sen2Cor processor, DTACSNet demonstrates a significantly better performance in cloud scene classification (F2 score of 0.89 for DTACSNet compared to 0.51 for Sen2Cor v2.8) and a surface reflectance estimation with average absolute error below 2% in reflectance units. Moreover, we tested DTACSNet on hardware-constrained systems similar to recent deployed missions and show that DTACSNet is 11 times faster than Sen2Cor with a significantly lower memory consumption footprint. These preliminary results highlight the potential of DTACSNet to provide enhanced efficiency, autonomy, and responsiveness in onboard data processing for Earth observation satellite missions.

Nowadays, the use of Artificial Intelligence on board Earth Observation satellites is under investigation for applications having strict bandwidth and latency requirements, such as vessel detection. However, many of the on-ground current computing pipelines rely on data post-processing techniques whose applications onboard satellites are tricky because of their limited computing power. To enable the analysis and the research of lightweight onboard data processing techniques, we provide VDS2Raw, the first Sentinel-2 Raw dataset for vessel detection applications. Finally, we also compared different object detection Deep Learning techniques in terms of detection performance and inference time to perform a feasibility analysis of performing onboard vessel detection on raw multi-spectral data.

Artificial intelligence onboard satellites has the potential to reduce data transmission requirements, enable real-time decision-making and collaboration within constellations. This study deploys a lightweight foundational model called RaVAEn on D-Orbit’s ION SCV004 satellite. RaVAEn is a variational auto-encoder (VAE) that generates compressed latent vectors from small image tiles, enabling several downstream tasks. In this work we demonstrate the reliable use of RaVAEn onboard a satellite, achieving an encoding time of 0.110s for tiles of a 4.8x4.8 km2area. In addition, we showcase fast few-shot training onboard a satellite using the latent representation of data. We compare the deployment of the model on the on-board CPU and on the available Myriad vision processing unit (VPU) accelerator. To our knowledge, this work shows for the first time the deployment of a multitask model onboard a CubeSat and the onboard training of a machine learning model.

Hyperspectral unmixing is one of the most challenging tasks in the analysis of such data. There have been an array of algorithms proposed for this problem so far, but they are virtually always verified using random sampling, where training and test examples are drawn from the same image. Since such samples are spatially correlated and may be positioned close to each other, random sampling can induce the training-test information leak in the techniques that exploit spatial information during the unmixing process. We want to raise the attention of the community about this validation flaw in the context hyperspectral unmixing. We introduce the algorithm for unbiased validation of the unmixing techniques through splitting hyperspectral images into training and test samples that do not suffer from the training-test information leak. The experiments showed that the widely-used random sampling verification strategy leads to overly optimistic conclusions concerning the algorithm’s performance. This problem was mitigated with the proposed approach which allows us to rigorously validate unmixing techniques.

Detecting methane in satellite hyperspectral images (HSIs) can play a key role in environmental monitoring, as taking timely actions to reduce its emission and handle (unexpected) super emitters is of paramount importance. We tackle this issue and propose a machine learning pipeline for this task, with the ultimate goal of deploying it on board a satellite. Such solutions can offer global scalability, and they can act as a smart data prioritization step, as only those HSIs which contain methane can be downlinked for further analysis. However, the on-board deployment induces additional practical challenges—such algorithms should be resource-frugal, and should effectively operate on the target image data which may not be available during their development, since the satellite is not in orbit yet. Our experimental study revealed that the data-driven approaches can effectively detect methane in original airborne HSIs, as well as in HSIs emulating the target sensor and generated through data-level simulations.

Improving agricultural practices through exploiting the recent imaging and machine learning advancements plays a key role nowadays to ensure sustainable food security, and to help us deal with the climate change. Quantifying soil parameters can lead to optimizing the fertilization process but it is cumbersome, time-consuming and difficult to scale, as it requires performing in-situ soil measurements that are later analyzed in the laboratory settings. In the HYPER-VIEW challenge, we aim at automating the soil analysis thanks to the utilization of hyperspectral images that capture very detailed information about the scanned objects in hundreds of contiguous hyperspectral bands. Such imagery can be effectively analyzed using an array of classical and deep machine learning approaches. Also, the AI techniques can be deployed on-board the imaging satellites— it opens new doors related to the scalability of the solution. The winners of the challenge will be offered a unique opportunity to run their proposed solution in orbit, on-board the Intuition-1 satellite, equipped with a hyperspectral imager and on-board AI capabilities.

Cloud detection is one of the most important image pre-processing steps that can be performed on-board satellites. It may allow us to reduce the amount of data to analyze or downlink by pruning the cloudy areas, or to make the satellites more autonomous through data-driven image acquisition re-scheduling of the areas obscured by clouds. Thus, building the cloud detection algorithms that can be ultimately deployed in orbit became an important research avenue. In this paper, we investigate the robustness of the fully-convolutional neural networks for cloud detection against the atmospheric conditions that resemble real acquisition settings of the Intuition-1 mission. Our experiments, performed over the original and simulated Landsat-8 images, with the latter reflecting target conditions, shed more light on the performance of deep models and showed how can we verify their robustness in Earth observation tasks for which real images do not exist yet.

The recent advances in Earth observation and artificial in-telligence allow us to improve the agricultural management practices through effectively exploiting the spectral, spatial, and temporal characteristics of the area of interest captured by satellite images. In this paper, we tackle the problem of extracting high-resolution (2.5-meter) cultivated land maps from Sentinel-2 multispectral images, and propose a machine learning algorithm for this task. It aggregates the spectral, spatial, and temporal features of the upsampled images, and is independent from the number of observations captured for a given scene. The experimental results, performed within the framework of the Enhanced Sentinel-2 Agriculture chal-lenge show that our technique manifests high generalization abilities over the unseen data and elaborates high-quality cul-tivated land maps. Finally, utilizing this algorithm led us to taking the $6^{\text{th}}$ place in the aforementioned challenge.

Earth observation is fundamental for a range of human activities including flood response as it offers vital information to decision makers. Semantic segmentation plays a key role in mapping the raw hyper-spectral data coming from the satellites into a human understandable form assigning class labels to each pixel. Traditionally, water index based methods have been used for detecting water pixels. More recently, deep learning techniques such as U-Net started to gain attention offering significantly higher accuracy. However, the latter are hard to interpret by humans and use dozens of millions of abstract parameters that are not directly related to the physical nature of the problem being modelled. They are also labelled data and computational power hungry. At the same time, data transmission capability on small nanosatellites is limited in terms of power and bandwidth yet constellations of such small, nanosatellites are preferable, because they reduce the revisit time in disaster areas from days to hours. Therefore, being able to achieve as highly accurate models as deep learning (e.g. U-Net) or even more, to surpass them in terms of accuracy, but without the need to rely on huge amounts of labelled training data, computational power, abstract coefficients offers potentially game-changing capabilities for EO (Earth observation) and flood detection, in particular. In this paper, we introduce a prototype-based interpretable deep semantic segmentation (IDSS) method, which is highly accurate as well as interpretable. Its parameters are in orders of magnitude less than the number of parameters used by deep networks such as U-Net and are clearly interpretable by humans. The proposed here IDSS offers a transparent structure that allows users to inspect and audit the algorithm’s decision. Results have demonstrated that IDSS could surpass other algorithms, including U-Net, in terms of IoU (Intersection over Union) total water and Recall total water. We used WorldFloods data set for our experiments and plan to use the semantic segmentation results combined with masks for permanent water to detect flood events.

Maintaining farm sustainability through optimizing the agricultural management practices helps build more planet-friendly environment. The emerging satellite missions can acquire multi- and hyperspectral imagery which captures more detailed spectral information concerning the scanned area, hence allows us to benefit from subtle spectral features during the analysis process in agricultural applications. We introduce an approach for extracting 2.5m cultivated land maps from 10m Sentinel-2 multispectral image series which benefits from a compact graph convolutional neural network. The experiments indicate that our models not only outperform classical and deep machine learning techniques through delivering higher-quality segmentation maps, but also dramatically reduce the memory footprint when compared to U-Nets (almost 8k trainable parameters of our models, with up to 31M parameters of U-Nets). Such memory frugality is pivotal in the missions which allow us to uplink a model to the AI-powered satellite once it is in orbit, as sending large nets is impossible due to the time constraints.

Hyperspectral unmixing remains one of the most challenging tasks in the analysis of such data. Deep learning has been blooming in the field and proved to outperform other classic unmixing techniques, and can be effectively deployed onboard Earth observation satellites equipped with hyperspectral imagers. In this letter, we follow this research pathway and propose a multi-branch convolutional neural network that benefits from fusing spectral, spatial, and spectral-spatial features in the unmixing process. The results of our experiments, backed up with the ablation study, revealed that our techniques outperform others from the literature and lead to higher-quality fractional abundance estimation. Also, we investigated the influence of reducing the training sets on the capabilities of all algorithms and their robustness against noise, as capturing large and representative ground-truth sets is time-consuming and costly in practice, especially in emerging Earth observation scenarios.

Spaceborne synthetic aperture radar (SAR) has been used for years to estimate high-resolution surface wind field from the ocean surface backscattered signal. Current SAR platforms have one single fixed antenna, and traditional inversion/retrieval schemes rely on one copolarized channel, leading to an unconstrained optimization problem for providing independent estimates of wind speed and direction. For routine application, this is generally solved witha prioriinformation from the numerical weather prediction (NWP) model, inducing severe limitations for rapidly evolving meteorological systems where discrepancies can be significant between model and measurements. In this study, we investigate the benefit of having two simultaneous acquisitions with phase-preserving information in copolarization and cross polarization provided by Sentinel-1 (S-1). A comprehensive analysis of the co-cross-polarization coherence (CCPC) is performed to adequately estimate and calibrate CCPC values from S-1 interferometric wide (IW) mode images acquired over the ocean. A new polarimetric calibration (PolCAL) methodology based on least-squares (LS) criterion and direct matrix inversion is proposed yielding crosstalk estimates. We document CCPC odd symmetry with respect to relative wind direction for light to medium wind speeds (up to 14 m/s) and incidence angle from 30° to 45°. The azimuthal modulation is found to increase with both wind speed and incidence angle. An analytical model C-band polarimetric geophysical model function (CPGMF) is provided. The synergy of the CCPC with other radar parameters, such as backscattering coefficients or Doppler, to further constrain the inversion scheme is assessed, opening new perspectives for SAR-based wind field retrieval independent of any NWP model information.

Since the first method proposed by Dvorak in the 70's, Tropical Cyclone (TC) monitoring strategy has been routinely improved for both operational analysis and forecasting of tropical cyclone intensity. However, one of the most widely used techniques for operational TC intensity analysis remains largely subjective, dependent on analyst training and lacking in incorporation of imagery beyond visible and infrared window frequencies. Using high-resolution data from the Sentinel-1 Synthetic Aperture Radar (SAR) mission, this study explores a Deep Learning approach to detect the eye-center of tropical cyclones and to estimate their intensity based on topology patterns. Moreover, we apply Gradient-based Class Activation Maps to better understand the characteristics of this learning based method. Experimental results demonstrate that the suggested approach not only recognizes TCs effectively, but also reveals potential to locate their centers accurately and compete in performance with existing subjective and automated intensity estimation techniques.

Segmentations of ocean SAR images (Sentinel-1 A and B) into 10 classes of metoceanic phenomena are for the first time presented, with a 400 m resolution. Ocean SAR images segmentation differs from classic deep learning problems with a high variety of shapes and a particular importance of high-frequency patterns. To this end, an assessment of deep learning frameworks is performed, with a focus on the comparison between weakly supervised and supervised methods. Metrics based on the Wassertein distance indicate best performances by the supervised segmentation (U-Net) given operational constraints, thus highlighting the significance of properly annotated data sets. While available training data sets are made of small$20 \times 20 \text{km}$imagettes, the extension of the inference from imagettes to wide swath images, with a wider variety of incidence angles, presents promising results and opens the way to more extensive oceanographic applications in SAR imagery.

Windrows signatures on SAR images are analyzed to estimate their orientation and their estimated accuracy with respect to reference wind direction provided by buoy and model. The accuracy dependency to several parameters of interest such as wind speed, processing internal variables, incidence angle, a priori presence of wind streaks but also SAR product acquisition modes and polarizations is analyzed. Regression models provide a satisfying method to combine these parameters and predict the a priori error on the estimated wind direction, crucial for any downstream application.

CYMS is an ESA-funded project aiming at scaling up an operational service for Tropical Cyclone monitoring, in view of its potential integration as part of a Copernicus Service. In 2020, the demonstration of such a service has been operated and user's requirements refined. More than 90 scenes of TC have been acquired worldwide with Sentinel-1 A, Sentinel-1 B and Radarsat-2 thanks to the late programming acquisitions. These images have been processed into ocean surface wind field and disseminated to the user community. The user feedback on the data confirms that the capabilities of SAR to probe the ocean surface at high resolution is unique and offer potential for science applications related to the analysis of the TC inner core structure, possibly bringing new insight on the processes within the eye. Those observations are also crucial over regions lacking any aircraft observation or ground-based meteorological Radars for monitoring and validating the cyclone forecasts. One of the main requirements is the need for Near Real-Time distributions of CYMS observations to allow their operational use. This is currently stated as a project but one of the objectives is to ensure that one of the Copernicus services hosts a service dedicated to Cyclone Observation to allow acquisitions on Cyclones in a Copernicus framework.

SAR ocean surface wind retrieval is generally based on the co-polarized Normalized Radar Cross Section (NRCS). Yet, since April 2014, Sentinel-1 TOPS-mode acquisitions enable large swath dual-pol measurements while preserving relative phase information. In this study, this new capability is analyzed via the co-cross coherence from a massive S-1 VV/VH IW dataset. A Polarimetric Calibration (POLCAL) methodology is proposed to calibrate this variable over sea surface. The odd symmetry of the co-cross coherence from S-1 is confirmed, and potential Polarimetric Geophysical Model Function (PGMF) can be now investigated. The integration of this PGMF in the wind field inversion scheme will be the next step.

Millions of wave mode synthetic aperture radar (SAR) images are routinely acquired by the ESA's two sentinel-1 satellites every month over the open ocean. These SAR images capture clear imprints of atmospheric boundary layer (ABL) roll vortices. This provides a new and unique opportunity to investigate the characteristics of ABL rolls globally and statistically. In this study, we take advantage of the deep learning classification tool that has been successfully developed to automatically identify ABL rolls. For each SAR image classified with ABL rolls, roll wavelength and orientation are extracted through spectral analysis. Surface meteorological variables are also collocated with each SAR image to address the atmospheric conditions of roll occurrence. Results show that roll vortices are prevalent over the whole ocean and mainly occur in unstable to near-neutral stratification. Roll characteristics follow the theoretical expectation and are in good agreement with previous studies.

The detection of sea ice leads is a prerequisite for the estimation of ice freeboard and thickness from altimeter data. The classification of altimeter waveforms is generally performed using statistical parameters on the echo power or machine learning approaches directly on the waveforms. The validation and optimization of such algorithms can be carried out using a set of reference cases provided by Earth Observation images. In this paper, we first developed a new lead detector based on Sentinel-1 (S-1) synthetic aperture radar (SAR) images. A robust and consistent methodology for the joint assessment of Altimeter and SAR leads detector is then provided. We propose to fully account for the 2-D geometric problem when comparing the 1-D altimeter track and 2-D SAR image. The surface of the lead intersecting the altimeter footprint and its distance to nadir are considered here. Based on collocated Sentinel-3 (S-3) altimeter data and S-1 images, the performance of our S-3 lead detector is fully assessed. A new parameterization is found resulting in a better tradeoff between good detection and false alarm rate. A similar analysis is performed using AltiKa altimeter data, showing enhanced performance for S-3 altimeter data acquired in Delay-Doppler mode with reduced off-nadir returns.

Sentinel-1 ‘s SARs operate since April 2014 (S1-A) and April 2016 (S1-B) and routinely acquire images over coastal and open waters. Compared to its predecessor ENVISAT/ASAR, Sentinel-1 SAR offers several improvements such as a better Wave Mode imagette coverage, more systematic dual-polarizations, a new TOPSAR acquisition mode over coastal regions and improved Doppler estimator allowing higher resolution Doppler grid. Based on these additional capabilities, many algorithmic and use case scenario improvements have been tested and validated to provide a more complete ocean state view and first direct assessment of extreme events. Instrument level issues like the accuracy of the satellite restituted attitude could also be highlighted.

Sentinel-1's SAR instrument offers a number of improvements with respect to its predecessor ENVISAT/ASAR such as a much better Wave Mode imagette coverage, improved Doppler estimator allowing higher resolution Doppler grid, more systematic dual-polarizations and a new TOPSAR acquisition mode. In the context of SEOM program, the Sentinel-1 ocean study offers to take advantage of these new capabilities to improve the retrieval of ocean sea state parameters: surface wind fields, directional wave spectrum, total significant wave height and surface currents. The study also tackles the ability to conduct a synergetic retrieval scheme in which the mutual effects of sea state components are taken into account.

This paper studies the effects of stationary-based processing of moving ship signatures in synthetic aperture radar (SAR) imagery and introduces a methodology to estimate and compensate for them. SAR imaging of moving targets usually results in residual chirps in the azimuthal SLC processed signal. The fractional Fourier transform (FrFT) makes it possible to represent the SAR signal in a rotated joint time-frequency plane and performs optimal processing and analysis of these residual chirp signals. The along-track defocus can thus be compensated for and the target's azimuthal speed estimated. The impact of higher order motion terms (e.g., acceleration) has been also considered. Experiments were conducted on a large number of ship signatures extracted from Radarsat-2 Multi Look Fine and Ultra Fine SAR images. An intercomparison with a standard Doppler Sublook Decomposition Method (SDM) is carried out, as well as a complete performance analysis with AIS data as ground truth.

This study addresses the performances of ship detection with data acquired by the newly launched Sentinel-1 SAR sensor. An automatic validation approach based on coastal AIS data is employed for measuring the detection efficiency. Results are compared with ship detection capabilities conducted on Radarsat-2 and CosmoSkymed datasets. The influence of different key parameters, such as SAR imaging characteristics (polarization, incidence angle) or meteorological conditions, is addressed. Such an analysis is useful for operational services to determine data specifications that assure optimum vessel detection for maritime surveillance applications.

This paper considers the impact of dynamical targets on SAR imagery, when processed with stationary based techniques. We propose to employ the Fractional Fourier Transform as a tool for estimating the residual Doppler rate corresponding to moving vessels. Hence, the defocusing effect can be corrected and the associated azimuthal velocity can be estimated. The capabilities of the proposed methodology are illustrated with moving vessels extracted from Radarsat-2 Multilook Fine images and AIS data flows as ground truth.

In this paper paper we address performances of ship detection algorithms in medium resolution SAR imagery. An automatic validation approach based on coastal AIS data allows to evaluate detectors efficiency. Detection capabilities remain sensitive to dataset features such as SAR imaging characteristics, meteorological conditions or vessel size. The influence of this different key parameters is fully assessed in this study. This analysis is valuable for operational services, allowing to select the most appropriate type of data for different applications in maritime surveillance.

From its launch in 2006, the phased array L-band synthetic aperture radar (PALSAR) onboard the advanced land observing satellite (ALOS) has acquired many dual-polarized (FBD) images with a 70-km swath width, aiming to produce spatially consistent coverage over tropical rainforest. This paper investigates the relevancy of PALSAR orthorectified FBD product at 50-m resolution for regional land cover classification by the support vector machines (SVM). Our test site is the Riau province, Sumatra island, Indonesia, known to hold vast area of natural peatland forest with an extreme biodiversity threatened by industrial deforestation. Since it is demonstrated the radiometric information (HH and HV channels) cannot be solely used to achieve a good classification, the spatial information in these orthorectified data is investigated. A new tool using the recursive feature elimination SVM-based process and the textural Haralick's parameters is introduced. The real contribution of textures within the land cover classification can be understood. A small set of textural parameters is determined at local scale while being optimal for the land cover discrimination. The SVM-based classifier is carried out across the whole Riau province and its results are compared with a Landsat-based estimation. The agreement is over 70% with six classes and 86% for the natural forest map. These results are remarkable since only one PALSAR FBD product is used and this assessment is performed on more than 40 million pixels. The results confirm the high potential of the PALSAR sensor for forest monitoring at regional, if not global scale.

This study investigates a new technique for land cover analysis by means of the Support Vector Machines. Intrinsic spatial variability within SAR images, beyond that caused by speckle, is of high interest for land cover characterization and classification. However, its use is still an ongoing issue due to its complex multi-scale nature. On the other hand, classification algorithms based on statistical learning methods such as the supervised Support Vector Machines (SVM) approach are implemented in a wide range of data mining applications. SVM can also be used as a technique for feature selection. In this paper, a new tool using the Recursive Feature Elimination SVM-based process (SVM-RFE) and the textural Haralick's parameters is introduced. The real contribution of textures within the land cover classification can be understood. A small set of textural parameters is determined at local scale while being optimal for the land cover discrimination. In this study, orthorectified 50m resolution data acquired by the L-band PALSAR/ALOS sensor are used.

PALSAR orthorectified HH and HV produced at 50m resolution is used for analysis. Since only two bands (HH and HV) have been limited in land cover discrimination, textures have been used as additional information for classification. This research derives second-order textures at different spatial resolutions and compares second-order textures at multiple scales to demonstrate their contributions in land cover classification. The discriminating capability of texture features is derived by the transformed divergence on several selected regions of interest. Optimum combination of backscattering and textures are used as input data into a supervised multi-resolution maximum likelihood classification. It is found that by including the texture information, the overall classification accuracy is improved by 10%.

Frozen ground is a sensitive indicator of how our home planet is changing. In the meantime, new spaceborne SAR systems have been launched, such as the polarimetric PALSAR sensor onboard ALOS in January 2006. In this paper, the relevance of L-band polarimetric SAR data for extracting information on frozen ground is presented. Dealing with ground assessments, the necessity for a validated Electromagnetic (EM) model is of importance. The adequation between Oh's po-larimetric EM model and PALSAR data is first studied over agricultural bare fields in Hokkaido, Japan. The assessment of residual liquid water can be realized by means of bare soil EM backscattering model. Over natural wildland area, an approach is proposed in order to tackle the effect of the vegetation or other irrelevant effects. The monitoring of permafrost active layer is performed over the ANWR, Alaska.

This paper presents a method to characterize snow cover in mountainous regions using dual-polarization C-band synthetic aperture radar (SAR) data. It is demonstrated that an accurate modeling of the liquid water distribution inside the snowpack, using a multilayer meteorological snow model, is required to characterize snow with precision. A multilayer-snow electromagnetic (EM) backscattering model is developed based on the vector radiative transfer, the strong fluctuation theory, and physical parameters supplied by the meteorological model. However, the limited resolution of the meteorological snow model is insufficient for predicting a refined EM backscattering at a massif scale. An adequate spatial reorganization of these snow profiles, based on a comparison between simulated and measured dual-polarization SAR data, leads to a better estimation of some snowpack parameters. In particular, the monitoring of snow liquid water content is presented improving the capacity of wet snow mapping as compared to a classical SAR-based method. This methodology shows good capacities both for qualitative and quantitative snow assessments, opening the way for a new operational method.

Snow classification using full-polarimetric PALSAR data is investigated in this paper. It is first demonstrated that dry snowpack over frozen ground slightly affects polarimetric signature at L-band. Given the fact that PALSAR data do not permit the use of a simplistic threshold-based method, a refined method for Snow Covered Area mapping is outlined. A supervised Support Vector Machine approach is used showing fairly good results within the framework of a three-classes classification (dry snow over frozen ground, wet snow and no snow).

This paper presents a method to characterize snow cover at a massif scale using dual-polarization C-band SAR data. It is demonstrated that it is crucial to exactly model the distribution of liquid water inside the snowpack in order to perform accurate snow characterization at C-band. Consequently, the key point of this new method consists in using a multi-layer meteorological snow model. Based on a validated multi-layer EM backscattering model, SAR data and snow profiles estimated by the weather model can be combined. An adequate spatial reorganization of these snow profiles leads to a refined snow characterization. Accurate snow monitoring like Liquid Water Content is presented, opening the way for a new operational method.

This paper presents a method to characterize snow cover using multi-temporal dual polarization ASAR/ENVISAT data. At first, variations of electromagnetic backscattering of snowpack depending on melting are explained and validated by a multi-layer model. It is demonstrated that it is crucial to exactly model the distribution of Liquid Water Content inside snow pack in C-band. Consequently, a new mapping algorithm based on the French weather model CROCUS is proposed in order to estimate the spatial variability of layered snowpack profile.