Search

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 extract rainfall information from SAR imagery. SAR has the advantage of providing global coverage and a very high resolution. 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, which has been implemented here as a neural network in a machine learning framework. 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 great interest for improving the qualification of SAR-derived wind field data., Comment: 9 pages, 9 figures

The wave spectrum is a representation of the state of the ocean surface from which many parameters can be deduced: significant wave height, peak parameters of dominant waves, directional parameters, etc... For more than 30 years, Synthetic Aperture Radars allow their routine montoring far from the coast in all surface conditions (through clouds and despite night), where buoys cannot be deployed. These measurements benefit from scientific efforts that now make it a reliable measurement technique. Sentinel-1 constellation is one of them and operates since 2016. However known limitations, including wave blurring caused by the azimuth cut-off, limit the performance of wave spectra measurement to long swells. SWIM is a new rotating Radar onboard the Chinese-French CFOSAT satellite dedicated to directional wave spectra measurement. Not suffering from the cut-off limitations, new horizons and perspectives for synergies are opened in terms of spectral limit and directionality. Sentinel-1 wave spectra measurements are limited to Wave Mode acquisitions, only available over deep ocean basins and away from the North-East Atlantic Oean. SWIM measurements offer hundreds of co-located measurements over these regions, but also extends the coverage to closed seas worldwide and to European waters. Other complementarities exist in wave measured wavelengths : Sentinel-1 extends to long swell (up to 800 m wavelength) while SWIM shows greater ability to measure wind sea components (close or below 50 m) and a significant part of the Stokes drift. Those complementarities are assessed at different levels and with different comparisons methodologies. First, partition integral parameters are compared between SWIM Level-2P products or S1 Level-2 products and numerical wave model outputs alone. Second, dynamical co-locations are performed between S1 and SWIM, using cross-overs given by Level-3 spectral products. These measurements, also referred to as Fireworks, allow to dramatically increase of the number of co-located points and better inter-compare. These new performances have applications in data assimilation and prospects for new products such as Stokes drift, a first-estimated spaceborne measurement.

International audience; Through the Synthetic Aperture Radar (SAR) embarked on the satellites Sentinel-1A and Sentinel-1B of the Copernicus program, a large quantity of observations is routinely acquired over the oceans. A wide range of features from both oceanic (e.g., biological slicks, icebergs, etc.) and meteorologic origin (e.g., rain cells, wind streaks, etc.) are distinguishable on these acquisitions. This paper studies the semantic segmentation of ten metoceanic processes either in the context of a large quantity of image-level groundtruths (i.e., weakly-supervised framework) or of scarce pixel-level groundtruths (i.e., fully-supervised framework). Our main result is that a fully-supervised model outperforms any tested weakly-supervised algorithm. Adding more segmentation examples in the training set would further increase the precision of the predictions. Trained on 20 × 20 km imagettes acquired from the WV acquisition mode of the Sentinel-1 mission, the model is shown to generalize, under some assumptions, to wide-swath SAR data, which further extents its application domain to coastal areas

The Hunga-Tonga submarine volcano produced a violent explosion on Saturday 15/01/2022 at approximately 04:15 UT (17:15 local time). The explosion generated a tsunami affecting all the Pacific Ocean. The tsunami has been detected in the multimission (Duacs) operational Sea Level products ingesting the flying altimeters Jason-3, Sentinel-3A, Sentinel-3B provided by Eumetsat, Cryosat-2, Saral, HY-2B provided by Cnes. After a specific processing carried out to remove the natural variability of the ocean, the tsunami signal was detected very clearly in the constellation, notably on HY2B, first to fly over the area 1h30 after the eruption. Comparisons to models show a position mismatch in the NRT predictions, demonstrating the interest of altimetry in tsunami analyses. The signature of the tsunami was also detected in radar images from the Copernicus Sentinel-1A synthetic aperture radar (SAR) satellite. the peak wavelength of the retrieved waves was estimated allowing us to track them back near the center of the eruption, therefore confirming that they were generated by this event. This paper describes the tsunami signal as observed by the altimetry constellation and Sentinel1A.

Rain Signatures on C-band Synthetic Aperture Radar (SAR) images acquired over ocean are common and can dominate the backscattered signal from the ocean surface. In many cases, the inability to decipher between ocean and rain signatures can disturb the analysis of SAR scenes for maritime applications. This study relies on Sentinel-1 SAR acquisitions in the Interferometric Wide swath mode and high-resolution measurements from ground-based weather radar to document the rain impact on the radar backscattered signal in both co- and cross-polarization channels. The dark and bright rain signatures are found in connection with the timeliness of the rain cells. In particular, the bright patches are demonstrated by the hydrometeors (graupels, hails) in the melting layer. In general, the radar backscatter under rain increases with rain rate for a given sea state and decreases when the sea state strengthens. The rain also has a stronger impact on the radar signal in both polarizations when the incidence angle increases. The complementary sensitivity of the SAR signal of rain in both channels is then used to derive a filter to locate the areas in SAR scenes where the signal is not dominated by rain. The filter optimized to match the rain observed by the ground-based weather radar is more efficient when both polarization channels are considered. Case studies are presented to discuss the advantages and limitations of such a filtering approach.

International audience; The above classification model (one 20x20 km imagette = one label) is then used over a full Wide Swath S-1 image using local class estimates with a convolution approach to provide a semantic segmentation. Multi-class (and multi-label , i.e. one pixel several classes) semantic segmentation is now feasible with DL , pending starting with a good and sufficient training database, and active learning process to enrich it, To this end, annotation tools and adequate framework should be consolidated and tuned to our problematic: 1) massive processing needed to raise specific issues (for instance, "sea ice" North of Madagascar should be re-tagged as internal waves), 2) additional information from ancillary metocean data might be provided , … The entire S-1 Wave Mode archive from 2016 is being processed. Below the occurrence of each class as classified by the DL model is provided on a monthly basis. This classified database could be used as input for a systematic collocation process with SWOT data. That will 1) help to understand Ka-band near nadir imaging processes for a given phenomenon, 2) serve during the Cal/val campaign, and 3) be used to build a training database with tagged SWOT images serving also DL model. With the upcoming launch of SWOT, new Ka-band near-nadir SAR image will be produced. Whereas the legacy of ocean SAR imaging is huge for L-C-or X-band SAR sensors with intermediate incidence angle, the interaction of Ka-band near-nadir EM waves and its associated SAR image formation lead to some uncertainties on how metocean features will be imaged on SWOT image. To name a few, atmospheric fronts, ocean fronts, rain cells, convective microcells, internal waves, gravity waves, biological slicks, upwelling or wind trails are phenomena that will be imaged by SWOT. These phenomena could be a source of errors and bias for SSH products. Meanwhile, they are of potential interest for the scientific communities. In this study, we aim to propose a methodology to flag and detect these phenomena to 1) mask them when generating L2 ocean products (SSH, wind…) and thus avoid any additional errors, 2) improve our understanding of Ka-band near nadir ocean SAR imaging while 3) being of interest for scientific community. This study is based on Sentinel-1 C-band SAR images for which the 3 above objectives are still relevant. Within the frame of Copernicus S-1 mission, high-resolution SAR images are produced each day, representing a daily average of 3.45 TB of published data.

International audience

Swells are found in all oceans and strongly influence the wave climate and air-sea processes. The poorly known swell dissipation is the largest source of error in wave forecasts and hindcasts. We use synthetic aperture radar data to identify swell sources and trajectories, allowing a statistically significant estimation of swell dissipation. We mined the entire Envisat mission 2003–2012 to find suitable storms with swells (13

A method for systematically tracking swells across oceanic basins is developed by taking advantage of high-quality data from space-borne altimeters and wave model output. The evolution of swells is observed over large distances based on 202 swell events with periods ranging from 12 to 18 s. An empirical attenuation rate of swell energy of about 4 × 10−7 m−1 is estimated using these observations, and the nonbreaking energy dissipation rates of swells far away from their generating areas are also estimated using a point source model. The resulting acceptance range of nonbreaking dissipation rates is −2.5 to 5.0 × 10−7 m−1, which corresponds to a dissipation e-folding scales of at least 2000 km for steep swells, to almost infinite for small-amplitude swells. These resulting rates are consistent with previous studies using in-situ and synthetic aperture radar (SAR) observations. The frequency dispersion and angular spreading effects during swell propagation are discussed by comparing the results with other studies, demonstrating that they are the two dominant processes for swell height attenuation, especially in the near field. The resulting dissipation rates from these observations can be used as a reference for ocean engineering and wave modeling, and for related studies such as air-sea and wind-wave-turbulence interactions.

The capability to observe ocean swell using spaceborne Synthetic Aperture Radar (SAR) has been demonstrated starting with ERS-1 mission in 1992. This dissertation shows how ocean swell properties can be used to combine swell observations of heterogeneous quality and acquired at various times and locations for the observation and forecast of ocean swell fieldsusing ASAR instrument on-board ENVISAT. The first section is a review of how ocean swell spectra can be derived from the SAR complex images of the ocean surface using a quasi-linear transformation. Then, significant swell heights, peak periods and peak directions from in situ measurements are used to assess the accuracy of the SAR observed swell spectra. Using linear propagation in deep ocean, a new swell field reconstruction methodologyis developed in order to gather SAR swell observations related to the same swell field. Propagated from their generation region, these observations render the spatio-temporal properties of the emanating ocean swell fields. Afterwards, a methodology is developed for the exclusion of outliers taking advantage of the swell field consistency. Also, using the irregularly sampled SAR observations, quality controlled estimations of swell field integral parameters are produced on a regular space-time grid. Validation against in situ measurements reveals the dramatic impact of the density of propagated observations on the integral parameters estimated accuracy. Specifically, this parameter is shown to be very dependent on the satellite orbit. Finally, comparisons with the numerical wave model WAVEWATCH-III prove it could potentially benefit from the SAR swell field estimates for assimilation purposes., L’imagerie satellite radar propose un point de vue intéressant pour l’étude et la compréhension des océans. Là où l’altimétrie, reconnue et utilisée mondialement, a su s’imposer comme une source de données majeure, les observations de houle issues du SAR (de l’anglais « Synthetic Aperture Radar ») restent encore largement sous exploitées. L’objet de cette thèse est de promouvoir l’utilisation de ces données en proposant un modèle pour l’analyse et la prévision de la houle à l’échelle du globe qui soit indépendant des modèles numériques classiques comme Wavewatch-III. Ce travail s’inscrit dans une logique de pérennisation de la mesure de houle depuis l’espace avec le lancement dans les trois années à venir des trois missions satellites Sentinel-1 A et B et CFOSAT. Un des principaux résultats de ce travail est la capacité de la méthode développée à fournir une information plus précise que celle des modèles existants. Cette méthode permet également une meilleure caractérisation des mesures utilisées en entrée et des pistes d’amélioration de ces dernières sont dégagées pour les futures activités de calibration/validation. Ces travaux ouvrent également des perspectives sur les possibilités d’assimilation des sorties de ce nouveau modèle dans les modèles numériques classiques.