Patrícia Carneiro Genovez

The development of scalable AI-based solutions for oil spill detection, discrimination, and characterization is crucial for advancing marine protection, disaster response, and sustainable ocean governance. Due to the attenuation of sea surface roughness, oil spills are detected as low backscattering regions in Synthetic Aperture Radar (SAR) imagery, being a strategic data source for operational services dedicated to marine pollution monitoring. Near-real-time information on the location, extent, and shape of oil-covered areas represents the primary SAR-derived input for oil spill response (OSR). In a subsequent stage, characterizing relative oil thickness variations within slicks becomes critical for improving cleanup effectiveness, which is higher over thicker oil layers commonly referred to as “actionable oil”.A strong contrast between dark features and the surrounding sea-surface is required for reliable detection and represents a key property for improving discrimination and characterization using data-driven approaches. In this context, the Damping Ratio (DR) has been demonstrated as a meaningful SAR-based feature for enhancing sea surface contrast. Compared to the Normalized Radar Cross Section (NRCS), DR is less affected by incidence angle and wind intensity, offering strong potential for the development of robust, operational AI-based systems for oil slick detection and characterization.Existing deep learning approaches for oil spill detection rely exclusively on NRCS, which has shaped available SAR datasets toward pre-processed products unsuitable for oil slick characterization. The project “Searching for Oil Spills on Sea Surfaces” (SOSeas) proposes a two-stage AI-based framework, in which deep learning is used for oil spill detection and delineation, followed by the thematic characterization of relative oil thickness within intra-slicks, both utilizing DR as the primary feature. Achieving this objective required the construction of the SOSeas.Dataset, a new, large-scale, field-validated oil spill benchmark that goes beyond NRCS, providing SAR-derived products from Sentinel-1, especially DR, in raw format to preserve the sea-surface backscattering properties important for characterization.A proof-of-concept dataset comprising 143 oil spills, field-validated by the Bonn Agreement and primarily located in the North Sea, was used to train, test, and validate a UNet–based semantic segmentation model for oil spill detection. Two identically configured models were trained using either DR or NRCS as input to directly compare their detection performance. To evaluate the discriminative power of DR versus NRCS, binary oil spill masks were used as labels to distinguish polluted water (PW) from non-oiled (NO) areas, which include clean ocean and lookalikes. Models trained with DR consistently outperformed NRCS-based models, achieving higher Intersection over Union (NRCS: 0.4058; DR: 0.5491) and F1-scores (NRCS: 0.58; DR: 0.71) for the polluted water class.The validation of the DR as a better feature for oil detection lays the foundation for developing the second stage as a future perspective, integrating DR and deep learning for oil slick characterization. Finally, the SOSeas.Dataset lays the groundwork for developing new AI-driven solutions capable of processing large volumes of SAR data, identifying patterns, and extracting useful information in near-real time, supporting operational agencies while enhancing monitoring and OSR actions for ocean protection.

Distinguishing between natural and anthropic oil slicks is a challenging task, especially in the Gulf of Mexico, where these events can be simultaneously observed and recognized as seeps or spills. In this study, a powerful data analysis provided by machine learning (ML) methods was employed to develop, test, and implement a classification model (CM) to distinguish an oil slick source (OSS) as natural or anthropic. A robust database containing 4916 validated oil samples, detected using synthetic aperture radar (SAR), was employed for this task. Six ML algorithms were evaluated, including artificial neural networks (ANN), random forest (RF), decision trees (DT), naive Bayes (NB), linear discriminant analysis (LDA), and logistic regression (LR). Using RF, the global CM achieved a maximum accuracy value of 73.15. An innovative approach evaluated how external factors, such as seasonality, satellite configurations, and the synergy between them, limit or improve OSS predictions. To accomplish this, specific classification models (SCMs) were derived from the global ones (CMs), tuning the best algorithms and parameters according to different scenarios. Median accuracies revealed winter and spring to be the best seasons and ScanSAR Narrow B (SCNB) as the best beam mode. The maximum median accuracy to distinguish seeps from spills was achieved in winter using SCNB (83.05). Among the tested algorithms, RF was the most robust, with a better performance in 81% of the investigated scenarios. The accuracy increment provided by the well-fitted models may minimize the confusion between seeps and spills. This represents a concrete contribution to reducing economic and geologic risks derived from exploration activities in offshore areas. Additionally, from an operational standpoint, specific models support specialists to select the best SAR products and seasons for new acquisitions, as well as to optimize performances according to the available data.

1141194: During oil spills response activities (OSR), discrimination and characterization of oil slicks has the potential to minimize false alarms and to indicate the location of thicker layers of oil, providing tactical and strategic information to decision makers. This specific demand for spatial intelligence during emergencies emphasizes the importance of the Multisource Image Processing System developed by the Brazilian Institute for Space Research (INPE) and described herein. Synthetic Aperture Radars (SAR) are the most common operational remote sensing data providers, acquiring images in different frequencies, incidence angles, resolutions, polarizations and formats (amplitude, intensity or complex). The system is able to integrate this product diversity including polarized (SAR), polarimetric (PolSAR) and optical data acquired by multiple sensors with different signal to noise ratios (SNR) and statistical distributions. The classifier uses a supervised approach to compare stochastic distances between different statistical distributions and statistical hypothesis testing to associate confidence levels to the results. Two study cases including mineral and biogenic oils with internal thickness variations were used to demonstrate the potential of the system to process SAR and PolSAR data, considering different formats and statistical distributions. The system was able to differentiate mineral oils from biogenic oils, as well as to detect thicker layers within the oil slicks with high global accuracies and low uncertainties. The polarimetric (dual and full-pol) and the intensity pair data had better accuracy at discriminating different types of oils; meanwhile SAR and PolSAR data returned statistically equivalent global accuracies in characterizing oil slicks. The low noise floor in UAVSAR sensor is able to detect oil thickness variations, which is feasible even if only using the VV channel. The possibility to qualitatively identify the thicker layers of the spilled oil, as well as minimize false alarms ratio is a significant contribution to ongoing efforts to improve OSR. The classification maps provided by the system may also be used to extract the relative areas for each type of oil, supporting the oil volume estimation. In this context, the integration of all information extracted from SAR and PolSAR data is useful both at strategic and tactical levels of an emergency, and is being used to guide aerial surveillance and the deployment of barriers and skimmers, aiming to increase oil recovery efficiency and minimize environmental impact.

During emergency responses to oil spills on the sea surface, quick detection and characterization of an oil slick is essential. The use of Synthetic Aperture Radar (SAR) in general and polarimetric SAR (PolSAR) in particular to detect and discriminate mineral oils from look-alikes is known. However, research exploring its potential to detect oil slick characteristics, e.g., thickness variations, is relatively new. Here a Multi-Source Image Processing System capable of processing optical, SAR and PolSAR data with proper statistical models was tested for the first time for oil slick characterization. An oil seep detected by NASA`s Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) in the Gulf of Mexico was used as a study case. This classifier uses a supervised approach to compare stochastic distances between different statistical distributions (fx) and hypothesis tests to associate confidence levels to the classification results. The classifier was able to detect zoning regions within the slick with high global accuracies and low uncertainties. Two different classes, likely associated with the thicker and thinner oil layers, were recognized. The best results, statistically equivalent, were obtained using different data formats: polarimetric, intensity pair and intensity single-channel. The presence of oceanic features in the form of oceanic fronts and internal waves created convergence zones that defined the shape, spreading and concentration of the thickest layers of oil. The statistical classifier was able to detect the thicker oil layers accumulated along these features. Identification of the relative thickness of spilled oils can increase the oil recovery efficiency, allowing better positioning of barriers and skimmers over the thickest layers. Decision makers can use this information to guide aerial surveillance, in situ oil samples collection and clean-up operations in order to minimize environmental impacts.

During the oil spills response actions, the rapid location and mapping of the oil slicks at sea surface is needed, as well as the characterization of different types of oils, avoiding false alarms whenever possible. In this way, all information provided by remote sensors can be useful to improve the strategies to cleanup oceanic regions and to protect the vulnerable coastal areas, minimizing possible environmental damages. Synthetic Aperture Radars (SAR) are widely used for this purpose, due to their potential for systematic data acquisition covering large areas with near real time delivery capability, offering different spectral and spatial resolutions, swaths, incidence angles and polarization channels. Since the nineties automatic and semiautomatic algorithms have been developed to detect low backscatter regions related with oil slicks aiming to assist large scale monitoring initiatives. Traditionally, the majority of the SAR data used have commonly been acquired with wide swaths in single or dual-pol channels. However during the emergency situations, where the position of the oil spills are known, polarimetric SAR data (Pol SAR) can be used to provide detailed information about the occurrences, as well as to minimize the false alarms interference. Recent studies, using several polarimetric attributes, have been investigating the potential of dual, quad and compact polarization modes to detect different backscatter mechanisms in biogenic and mineral oil slicks, probably related with different chemical properties and thicknesses variations. The objective of this paper is to evaluate the potential and the trade-offs provided by single, dual and quad-polarized data to detect oil slicks at sea surface. To accomplish this, a set of SAR data acquired by RADARSAT-2 [quad-polarized] and Sentinel-1 [dual-polarized] were processed and classified using a region classifier named Pol Class based on stochastic distances. The Pol Class was used because of its potential to use proper statistical modeling to process SAR data in different formats. This work uses the Intensity and the Polarimetric modules to classify the SAR data. The Polarimetric module is used to evaluate all complex information contained in the Pol SAR data, and the Intensity module provides the dual and single-polarized channels assessment, making feasible to compare the advantages of using all possible polarimetric combinations. The expectation is to contribute with the scientific community evaluation about the trade-offs provided by the single, dual and quad-polarized SAR data applied to oil detection, considering as reference the needs of the operational use during both, routine monitoring and emergency situations.