Hyperspectral PRISMA images are new and have not yet been evaluated for their ability to detect marine plastic litter. The hyperspectral PRISMA images have a fine spectral resolution, however, their spatial resolution is not high enough to enable the discrimination of small plastic objects in the ocean. Pansharpening with the panchromatic data enhances their spatial resolution and makes their detection capabilities a technological challenge. This study exploits, for the first time, the potential of using satellite hyperspectral data in detecting small-sized marine plastic litter. Controlled experiments with plastic targets of various sizes constructed from several materials have been conducted. The required pre-processing steps have been defined and 13 pansharpening methods have been applied and evaluated for their ability to spectrally discriminate plastics from water. Among them, the PCA-based substitution efficiently separates plastic spectra from water without producing duplicate edges, or pixelation. Plastic targets with size equivalent to 8% of the original hyperspectral image pixel coverage are easily detected. The same targets can also be observed in the panchromatic image, however, they cannot be detected solely by the panchromatic information as they are confused with other appearances. Exploiting spectra derived from the pan-sharpened hyperspectral images, an index combining methodology has been developed, which enables the detection of plastic objects. Although spectra of plastic materials present similarities with water spectra, some spectral characteristics can be utilized for producing marine plastic litter indexes. Based on these indexes, the index combining methodology has successfully detected the plastic targets and differentiated them from other materials.

A significant amount of the produced solid waste reaching the oceans is made of plastics. The amount of plastic debris in the ocean and coastal areas is steadily increasing and is now a major global environmental issue. The monitoring of marine plastic litter, ground-based monitoring systems and/or field campaigns are time-consuming, expensive, require great organisational efforts, and provide very limited information in terms of the spatial and temporal dynamics of marine debris. Earth Observation (EO) by satellite can contribute significantly to marine plastic litter detection. In 2019, a new hyperspectral satellite, called PRISMA, was launched by the Italian Space Agency. The high spectral resolution of PRISMA may allow for better detection of floating plastic materials. At the same time, Machine Learning (ML) algorithms have the potential to find hidden patterns and identify complex relations among data and are increasingly employed in EO. This paper presents the development of a new method of identifying floating plastic objects in coastal areas by exploiting pan-sharpened hyperspectral PRISMA data, based on the combination of unsupervised and supervised ML algorithms. The study consisted of a configuration phase, during which the algorithms were trained in a fully controlled test, and a validation phase, in which the pre-trained algorithms were applied to satellite data collected at different sites and in different periods of the year. Despite the limited input data, results suggest that the tested ML approach, applied to pan-sharpened PRISMA data, can effectively recognise floating objects and plastic targets. The study indicates that increasing input datasets can help achieve higher-quality results.

Cloudy regions in optical satellite images prevent the extraction of valuable information by image processing techniques. Several threshold, multi-temporal and machine learning approaches have been developed for the separation of clouds in land and ocean applications, but this task still remains a challenge. Concerning deep water marine applications, the main difficulties are imposed in regions with high noise levels and sunglint. In this study, artificial neural networks (ANNs) with different configurations are evaluated for the detection of clouds in Sentinel-2 images depicting deep water regions with several noise levels. The ANNs are trained on a manual public dataset and on a manual dataset created for the needs of this study, which authors intend to make publicly available. Results are compared with the cloud masks produced by three state-of-the-art algorithms: Fmask, MAJA, and Sen2Cor. It was shown that the ANNs trained on the second dataset perform very favourably, in contrast to the ANNs trained on the first dataset that fails to adequately represent the spectra of the noisy Sentinel-2 images. This study further reinforces the value of the ‘cirrus’ band and indicates the bands that mitigate the influence of noisy spectra, by defining and examining an index that characterizes the importance of the bands according to the weights produced by the ANNs. Finally, the possibility of improving results by making predictions using the feature scaling parameters of the test set instead of those of the training set is also investigated in cases where the test set cannot be adequately represented by the training set.

Change detection (CD) through Earth observation techniques can offer very significant information for monitoring tasks in a time-efficient manner. Very high-resolution (VHR) images can display objects in fine detail, thus making it possible to rapidly perceive isolated changes. However, this is a challenging task because of the increased within-class variance and geometric registration errors caused by different satellite view directions and angles. Lately, deep learning (DL) CD methods have proven very appealing for the CD problem because of their flexibility to combine and process different types of information along with the increased availability of higher processing power systems. Even though previous research has developed several notable DL methodologies, it has mostly focused on images with minor co-registration errors. Based on that, the goal of this study is to evaluate the performance of five state-of-the-art DL CD methods, two unsupervised and three supervised, on VHR images with severe co-registration errors. The methods are implemented on four urban European areas of versatile morphology. In addition, before applying the CD process, four popular automatic co-registration methods were evaluated because of the importance of this pre-processing step for the successful output of the CD problem. It was shown that phase correlation used on the Fourier-Mellin Transform produced the most satisfactory co-registration results and STANet detected building-related changes most successfully. Its success can be attributed to its particular attention mechanism and its training dataset. The rest of the co-registration and CD methods showed low performance.

Removal of cloud interference is a crucial step for the exploitation of the spectral information stored in optical satellite images. Several cloud masking approaches have been developed through time, based on direct interpretation of the spectral and temporal properties of clouds through thresholds. The problem has also been tackled by machine learning methods with artificial neural networks being among the most recent ones. Detection of bright non-cloud objects is one of the most difficult tasks in cloud masking applications since spectral information alone often proves inadequate for their separation from clouds. Scientific attention has recently been redrawn on self-organizing maps (SOMs) because of their unique ability to preserve topologic relations, added to the advantage of faster training time and more interpretative behavior compared to other types of artificial neural networks. This study evaluated a SOM for cloud masking Sentinel-2 images and proposed a fine-tuning methodology to separate clouds from bright land areas. The fine-tuning process which is based on the output of the non-fine-tuned network, at first directly locates the neurons that correspond to the misclassified pixels. Then, the incorrect labels of the neurons are altered without applying further training. The fine-tuning method follows a general procedure, thus its applicability is broad and not confined only in the field of cloud-masking. The network was trained on the largest publicly available spectral database for Sentinel-2 cloud masking applications and was tested on a truly independent database of Sentinel-2 cloud masks. It was evaluated both qualitatively and quantitatively with the interpretation of its behavior through multiple visualization techniques being a main part of the evaluation. It was shown that the fine-tuned SOM successfully recognized the bright non-cloud areas and outperformed the state-of-the-art algorithms: Sen2Cor and Fmask, as well as the version that was not fine-tuned.