Satellite-borne Synthetic Aperture Radar Images Research Articles

A domain specific knowledge extraction transformer method for multisource satellite-borne SAR images ship detection

Multisource satellite-borne synthetic aperture radar (SAR) images have different probability distributions. Traditional supervised learning, consequently, cannot achieve good test performance on one novel satellite-borne SAR dataset while training a good model on another existing satellite-borne SAR dataset. In this article, a domain adaptation (DA) Transformer object detection method is proposed to solve the unlabeled multisource satellite-borne SAR image object detection problem. Unlike existing DA methods based on convolutional neural network (CNN) that focus more on multi-level local feature extraction, we choose to use Vision Transformer (ViT) Faster region CNN (FRCNN) as the baseline network to cope with the extraction of global features of SAR images. Then, two classification tokens are used to learn the mapping of different domains and fully extract domain-specific knowledge, generating two different feature spaces that rely on the original label and pseudo-label to train the source and target domains feature spaces, respectively. Besides, the pseudo-label of target domain is also refined and reconstructed by feature clustering in order to improve the accuracy of target domain knowledge. Finally, the original detection head of FRCNN is employed to detect the target domain SAR image objects. Extensive experiments on image datasets from multisource satellite-borne SAR such as Gaofen-3, TerraSAR-X, Sentinel-1, and RadarSAT-2 show that compared to the other state of the art (SOTA) methods, the proposed method can achieve the greatest object detection accuracy. Especially, taking the recently proposed Transformer-based method as an example, our method has more than 5% improvement in accuracy and more than 16% reduction in training time.

Multiparametric sea state fields from synthetic aperture radar for maritime situational awareness

This paper introduces a method for estimating a series of sea state parameters from satellite-borne synthetic aperture radar (SAR). The method was realized in a near real time (NRT) application which allows for the processing of data from different satellites and modes. The algorithm estimates the total significant wave height Hs, dominant and secondary swell and windsea wave heights, first, and second moment wave periods, the mean wave period and the period of wind sea. The algorithm was applied to the Sentinel-1 (S1) C-band Interferometric Wide Swath Mode (IW), Extra Wide (EW) and Wave Mode (WV) Level-1 (L1) products and also extended to X-band TerraSAR-X (TS-X) StripMap (SM) products. The scenes are processed in raster and result in continuous sea state fields, with the exception of S1 WV, where averaged values for sea state parameters for along-orbit imagettes of 20 km × 20 km are presented.The developed empirical algorithm consists of two parts: a first CWAVE_EX (extended CWAVE) part, based on a linear regression approach, and a subsequent machine learning part using the support vector machine (SVM) technique. A series of new data preparation steps (i.e. filtering, denoising) and new features estimated from SAR images are also introduced. The algorithm was tuned and validated using two independent global wave model hindcasts, WaveWatch-3 and MFWAM as well as National Data Buoy Center (NDBC) measurements. The achieved root mean squared errors (RMSE) for CWAVE_EX for the total Hs are 0.60 m for low-resolution modes S1 IW (10 m pixel) and EW (40 m pixel) and 0.35 m for S1 WV and TS-X SM (pixel spacing ca.1–4 m) in comparison to model predictions. The RMSEs of the retrieved wave periods are in the range of 0.45–0.90 s for all of the satellites and models considered. Similarly, the dominant and secondary swell, and wind sea wave height RMSEs are in the range of 0.35–0.80 m. The SVM postprocessing improves the accuracy of the initial results of CWAVE_EX for Hs and reaches an RMSE of 0.25 m for S1 WV. Comparisons to 64 NDBC buoys, collocated at distances shorter than 50 km to S1 WV imagettes worldwide, result in an RMSE of 0.41 m. All results and the methods presented are novel in terms of the accuracy achieved, combining the classical approach with machine learning techniques, and performing an automatic NRT processing of multiparametric sea state fields from L1 data with automatic switching for different satellites and modes.The complete archive of S1 WV L1 Single Look Complex products from December 2014 until February 2021 was processed to create a sea state parameter database and validated using model hindcast and buoy measurements. The derived parameters are available to the public within the scope of the European Space Agency's Climate Change Initiative.

Ocean surface wind speed retrieval from C-band quad-polarized SAR measurements at optimal spatial resolution

ABSTRACT This work investigates the effect of the resampling spatial resolution (RSR) on the retrieval of ocean surface wind speed from satellite-borne Synthetic Aperture Radar (SAR) images. For this purpose, 358 RADARSAT-2(RS-2) fine-beam quad-polarized SAR images and in-situ National Data Buoy Centre (NDBC) and China State Oceanic Administration (SOA) buoy measurements from selected geographic locations are used in this paper. In order to obtain better accuracy SAR-retrieved wind speeds, CMOD5.N function, Z2011 polarization ratio (PR) model and C2PO model constitute the best models for VV-, HH- and cross-polarized SAR images, respectively. The resulting statistical analysis suggests that better accuracy wind speeds can be retrieved at somewhat coarser resolutions. The optimal RSR for the VV-, HH-, VH- and HV-polarized channels are 850, 1050,1350 and 1450 m, respectively, with RMSEs of 1.74, 2.03, 1.61 and 1.83 m/s.

Estimation of sea state from Sentinel-1 Synthetic aperture radar imagery for maritime situation awareness

ABSTRACTA new empirical algorithm CWAVE_S1-IW for estimation of significant wave height Hs including swell and wind sea from C-band satellite-borne Synthetic Aperture Radar (SAR) data has been developed for Sentinel-1 (S1) Interferometric Wide Swatrh Mode (IW) imagery. The algorithm was implemented into the Sea Sate Processor (SSP) for fully automatic processing for near real time (NRT) services and allow the estimation of wave fields of thousands of kilometres in the flight direction and 250 km swath from S1 IW scenes consisting of a sequence of individual images.The priority of CWAVE_S1-IW development was an automatic, fast and robust raster processing independent of wave patterns, applicable even when only clutter is visible in the SAR images. The algorithm is based on the spectral analysis of subscenes in wavenumber space. The empirical function allows direct Hs estimation from image spectra without first converting them into wave spectra and uses integrated image spectra parameters as well as estimated local wind information. A texture analysis based on Grey Level Co-occurrence Matrices (GLCM) is also applied. In this way, also the parameters of short waves can be estimated, which are not visible in S1 IW images and are only represented by clutter.The algorithm was tuned worldwide using in-situ collocated measurements of 92 buoys with more than 2500 acquisitions. The validated SSP allows automatic processing of worldwide S1 IW images in VV or HH polarization, including Atlantic storms, cyclones, and huge storms in the Gulf of Alaska with a root-mean-square (RMSE) error of 80 cm for Hs. For the closed seas like the North Sea, Baltic Seas and Black Sea the accuracy is higher with an RMSE = 55 cm. The algorithm is integrated into a demonstration service, used for further validation at the DLR ground station in Neustrelitz. The NRT processing has been tested by supporting a research ship cruise in the Antarctic Sea.

Meteo-marine parameters for highly variable environment in coastal regions from satellite radar images

The German Bight of the North Sea is the area with highly variable sea state conditions, intensive ship traffic and with a high density of offshore installations, e.g. wind farms in use and under construction. Ship navigation and the docking on offshore constructions is impeded by significant wave heights HS>1.3m. For these reasons, improvements are required in recognition and forecasting of sea state HS in the range 0–3m. Thus, this necessitates the development of new methods to determine the distribution of meteo-marine parameters from remote sensing data with an accuracy of decimetres for HS. The operationalization of these methods then allows the robust automatic processing in near real time (NRT) to support forecast agencies by providing validations for model results.A new empirical algorithm XWAVE_C (C=coastal) for estimation of significant wave height from X-band satellite-borne Synthetic Aperture Radar (SAR) data has been developed, adopted for coastal applications using TerraSAR-X (TS-X) and Tandem-X (TD-X) satellites in the German Bight and implemented into the Sea Sate Processor (SSP) for fully automatic processing for NRT services.The algorithm is based on the spectral analysis of subscenes and the model function uses integrated image spectra parameters as well as local wind information from the analyzed subscene. The algorithm is able to recognize and remove the influence of non-sea state produced signals in the Wadden Sea areas such as dry sandbars as well as nonlinear SAR image distortions produced by e.g. short wind waves and breaking waves. Also parameters of very short waves, which are not visible in SAR images and produce only unsystematic clutter, can be accurately estimated. The SSP includes XWAVE_C, a pre-filtering procedure for removing artefacts such as ships, seamarks, buoys, offshore constructions and slicks, and an additional procedure performing a check of results based on the statistics of the whole scene.The SSP allows an automatic processing of TS-X images with an error RMSE=25cm and Scatter Index SI=20% for total significant wave height HS from sequences of TS-X StripMap images with a coverage of ∼30km×300km across the German Bight. The SSP was tuned spatially with model data of the German Weather Service’s (DWD) CWAM (Coastal WAve Model) with 900m horizontal resolution and tuned in situ with 6 buoys located in DWD model domain in the German Bight. The collected, processed and analyzed data base for the German Bight consists of more than 60 TS-X StripMap scenes/overflights with more than 200 images since 2013 with sea state acquired in the domain HS=0–7m with a mean value of 1.25m over all available scenes at buoy locations.The paper addresses the development and implementation of XWAVE_C, and presents the possibilities of SSP delivering sea state fields by reproducing local HS variations connected with local wind gusts, variable bathymetry and moving wind fronts under different weather conditions.

Ship Detection with Spectral Analysis of Synthetic Aperture Radar: A Comparison of New and Well-Known Algorithms

The surveillance of maritime areas with remote sensing is vital for security reasons, as well as for the protection of the environment. Satellite-borne synthetic aperture radar (SAR) offers large-scale surveillance, which is not reliant on solar illumination and is rather independent of weather conditions. The main feature of vessels in SAR images is a higher backscattering compared to the sea background. This peculiarity has led to the development of several ship detectors focused on identifying anomalies in the intensity of SAR images. More recently, different approaches relying on the information kept in the spectrum of a single-look complex (SLC) SAR image were proposed. This paper is focused on two main issues. Firstly, two recently developed sub-look detectors are applied for the first time to ship detection. Secondly, new and well-known ship detection algorithms are compared in order to understand which has the best performance under certain circumstances and if the sub-look analysis improves ship detection. The comparison is done on real SAR data exploiting diversity in frequency and polarization. Specifically, the employed data consist of six RADARSAT-2 fine quad-polacquisitions over the North Sea, five TerraSAR-X HH/VV dual-polarimetric data-takes, also over the North Sea, and one ALOS-PALSAR quad-polarimetric dataset over Tokyo Bay. Simultaneously to the SAR images, validation data were collected, which include the automatic identification system (AIS) position of ships and wind speeds. The results of the analysis show that the performance of the different sub-look algorithms considered here is strongly dependent on polarization, frequency and resolution. Interestingly, these sub-look detectors are able to outperform the classical SAR intensity detector when the sea state is particularly high, leading to a strong clutter contribution. It was also observed that there are situations where the performance improvement thanks to the sub-look analysis is not so noticeable.

Chronic, Anthropogenic Hydrocarbon Discharges in the Gulf of Mexico

Satellite-borne Synthetic Aperture Radar (SAR) was used to obtain more precise estimates of the magnitude of the chronic hydrocarbon discharges described in qualitative pollution reports associated with the production and transportation network of the U.S. coast of the Gulf of Mexico. The National Response Center (NRCen) oil pollution reports were collected and filtered for the period of 2001 to 2012 to determine which of the reports coincided with archived SAR images. Some of the images covered multiple reports and some of the oil discharges described in one report could be observed in more than one image. In all, 177 reports could be investigated from 137 SAR images collected on or near the corresponding report dates. Further analysis found that oil slicks observed in 66 of these SAR images could be attributed to 67 of the reported incidents. Objective measurements indicated that the area of these transient oil slicks visible in SAR images was, on average, significantly larger than what was reported to the NRCen. The only recurring point source for oil slicks was the former site of the Taylor offshore platform. Here chronic, oil slicks were observed that were consistently much larger than other anthropogenic discharges. The SAR images of floating oil discharged from the Taylor site were verified by visual inspection from a boat and aerial photography. For some of the oil slicks discharged from the Taylor site, the accuracy of SAR images for detecting oil slick areas was validated by comparing SAR results to Landsat 7 Enhanced Thematic Mapper Plus (ETM+) and Moderate Resolution Imaging Spectroradiometer (MODIS) images. These results show that surveillance by SAR would improve accuracy for estimates of chronic anthropogenic oil pollution, particularly where continuous discharges are on-going.

Ocean Feature Analysis Using Automated Detection and Classification of Sea-Surface Temperature Front Signatures inRADARSAT-2Images

The Royal Canadian Navy produces a semiweekly map of major water mass boundaries in the Western North Atlantic using temperature measurements from several data sources, including satellite sea surface temperature (SST) images from the Advanced Very High Resolution Radiometer (AVHRR). Temporal–spatial detail that can be provided by AVHRR of the location of important SST boundaries such as the Gulf Stream North Wall is limited due to pervasive cloud cover. The ability of satellite-borne synthetic aperture radar (SAR) to image SST front signatures unrestrained by cloud cover makes it a potentially significant additional data source. The Spaceborne Ocean Intelligence Network project has developed an automated procedure to detect candidate SST front signatures in RADARSAT-2 SAR images of the ocean surface and classify them with greater than 80% accuracy.

Speckle reduction for the forest mapping analysis of multi-temporal Radarsat-1 images

As the number of satellite-borne synthetic aperture radar (SAR) systems increases, both the availability and the length of multi-temporal (MT) sequences of SAR images have also increased. Previous research on MT SAR sequences suggests that they increase the classification accuracy for all applications over single date images. Yet the presence of speckle noise remains a problem and all images in the sequence must be speckle filtered before acceptable classification accuracy can be attained. Several speckle filters designed specifically for MT sequences have been reported in the literature. Filtering in the spatial domain, as is usually done, reduces the effective spatial resolution of the filtered image. MT speckle filters operate in both the spatial and temporal dimensions, thus the reduction in resolution is not likely to be as severe (although a comparison between MT and spatial filters has not been reported). While this advantage may be useful when extracting spatial features from the image sequence, it is not quite as apparent for classification applications. This research explores the relative performance of spatial and MT speckle filtering for a particular classification application: mapping boreal forest types. We report filter performance using the radiometric resolution as measured by the equivalent number of looks (NL), and classification performance as measured by the classification accuracy. We chose representative spatial and MT filters and found that spatial speckle filters offer the advantage of higher radiometric resolution and higher classification accuracy with lower algorithm complexity. Thus, we confirm that MT filtering offers no advantage for classification applications; spatial speckle filters yield higher overall performance.

Using SAR images to delineate ocean oil slicks with a texture-classifying neural network algorithm (TCNNA)

Satellite-borne synthetic aperture radar (SAR) data are widely used for detection of hydrocarbon resources, pollution, and oil spills. These applications require recognition of particular spatial patterns in SAR data. We developed a texture-classifying neural network algorithm (TCNNA), which processes SAR data from a wide selection of beam modes, to extract these patterns from SAR imagery in a semisupervised procedure. Our approach uses a combination of edge-detection filters, descriptors of texture, collection information (e.g., beam mode), and environmental data, which are processed with a neural network. Examples of pattern extraction for detecting natural oil seeps in the Gulf of Mexico are provided. The TCNNA was successful at extracting targets and rapidly interpreting images collected under a wide range of environmental conditions. The results allowed us to evaluate the effects of different environmental conditions on the expressions of oil slicks detected by the SAR data. By processing hundreds of SAR images, we have also found that the optimum wind speed range to study surfactant films is from 3.5 to 7.0 m·s–1, and the best incidence angle range for surfactant detection in C-band is from 22° to 40°. Minor postprocessing supervision is required to check TCNNA output. Interpreted images produce binary arrays with imbedded georeference data that are easily stored and manipulated in geographic information system (GIS) data layers.

Measurement of Ocean Surface Winds Using Synthetic Aperture Radars

A methodology for retrieving high-resolution ocean surface wind fields from satellite-borne synthetic aperture radar (SAR) data is introduced and validated. The algorithms developed are suited for ocean SAR data, which were acquired at the C band of either vertical (VV) or horizontal (HH) polarization in transmission and reception. Wind directions are extracted from wind-induced streaks that are visible in SAR images of the ocean at horizontal scales greater than 200 m. These wind streaks are very well aligned with the mean surface wind direction. To extract the orientation of these streaks, two algorithms are introduced, which are applied either in the spatial or spectral domain. Ocean surface wind speeds are derived from the normalized radar cross section (NRCS) and image geometry of the calibrated SAR images, together with the local SAR-retrieved wind direction. Therefore, several C-band models (CMOD IFR2, CMOD4, and CMODS) are available, which were developed for VV polarization, and have to be extended for HH polarization. To compare the different algorithms and C-band models as well as demonstrate their applicability, SAR-retrieved wind fields are compared to numerical-model results considering advanced SAR (ASAR) data from Environmental Satellite (ENVISAT), a European satellite.