European Spatial Data for Coastal and Marine Remote Sensing

Salt marshes offer a plethora of ecosystem services such as biodiversity support, ocean–climate change regulation, ornithology recreo tourism or plant gathering by hand. They undergo significant worldwide losses due to their conversion into crop fields and to their spatial compression between rising sea levels and armored shorelines. Their management requires multiscale spatiotemporal analysis to detect interrelated patterns and processes. This research innovatively studies continuous salt marsh mapping, based on normalized difference vegetation index (NDVI) ranges, across three spatial and two temporal scales. Sentinel-2 (10 m), Dove (3 m), and unmanned airborne vehicle (UAV) (0.03 m) imagery were used to progressively refine spatial resolutions over dynamic areas (extending from hundreds, tens, and a couple of km², respectively). NDVI ranges from Sentinel-2 and Dove imagery were monitored with a lag of 5 and 4 years, respectively. Contrary to spaceborne imagery, UAV imagery lacked a near-infrared (NIR) band. The UAV NIR band was thus modelled (R²_NIR = 0.98) using a three-layered neutral network (NN) prediction based on red, green, and blue reflectance imagery, itself calibrated/validated/tested by Dove imagery bands (R²_red = 0.88, R²_green = 0.84, and R²_blue = 0.90). The 100-fold increase in pixel size allowed to detect the decimeter-scale objects of salt marshes and tidal flats. The multiscale NDVI ranges were associated with microphytobenthos and topographically low, medium, and high salt marsh vegetation, including the opportunistic Elymus genus. The combination of the NDVI values derived from the Sentinel-2, Dove, and UAV imagery enabled to survey a region while detecting subtle features of salt marshes, providing an updated toolbox for managers.

Coastal ecogeomorphological features support remarkable biodiversity and provide a wide variety of ecosystem services: cultural services (recreation, tourism facilities), provisioning services (agricultural production, pastoralism), and regulating services, including carbon sequestration and natural protection against coastal erosion and marine flooding. Therefore, mapping these coastal features with very high spatial resolution is a major challenge to their preservation and to face the challenges of global change. In this study, the contribution of the near-infrared (NIR) and mid-infrared (MIR) bands from multispectral drone and super-spectral (SS) WorldView-3 (WV-3) satellite images was used to map coastal ecogeomorphological features using two supervised classification algorithms: maximum likelihood (ML) and support vector machine (SVM). Various combinations of spectral bands, visible + NIR and visible + MIR, evaluated through the overall accuracy (OA) scores, for the classification of ecogeomorphological features revealed the significant contribution of the NIR and MIR bands to the mapping of coastal features. The addition of the NIR bands to the RGB band combination significantly increased the OA scores of the classifications (by +4.99% and +6.54%, with the ML and SVM algorithms, respectively). The addition of MIR bands to the combination of these bands provides classifications with even higher OAs (up to 99.1% and 98.4%), demonstrating the relevance of MIR bands for the mapping of coastal ecogeomorphological features.

The political, economic, and social changes associated with the collapse of the Soviet Union at the end of the 1980s led to major land cover and land-use changes in the southeastern Baltic Sea coastal regions. These changes (demilitarization of the coasts, end of collective ownership, specialization of economic activities, etc.) are characterized by a fast process of coastalization with the growth of urban areas, coast suburbanization, and the decrease of agricultural land. At the same time, we observe the implementation of protected natural areas at the regional level and through cross-border cooperation with international organizations (UNESCO, European Union [EU], etc.). Both processes have an important impact on the management of the coastlines of Latvia, Lithuania, and Russia. The analysis of the coastal changes is based on the use of Landsat remote sensing data series from the 1980s to 2020 combined with EU geographic databases and the land-use plans. The comparative analysis of the land cover changes in the Oblast of Kaliningrad, Lithuanian, and Latvian coastal zones allows us to understand the impacts of the three different planning policies since the end of the 1980s. The territorial dynamics are modelled using the GEOBIA package with object-oriented classification and machine-learning algorithms (maximum likelihood, minimum distance to means, parallelepiped classifiers) applied to the Landsat 5 TM and Landsat 8 OLI satellite multispectral images. The produced land cover maps are compared with the Climate Change Initiative Land Cover of the European Space Agency from 1995 to 2015.

The coastal city of Ténès, located in northwestern Algeria, is exposed to several natural hazards, such as floods, earthquakes, landslides, and forest fires. Due to human activities, socio-economic constructions, agricultural activities, and the resulting population acceleration, land cover and land use (LULC) dynamics in the city are changing over time. Hence, the understanding of LULC changes and its interactions with human activities and natural hazards is essential for appropriate land management and decision-making. In this study, we investigate LULC changes in the Allala watershed, including the city of Ténès, using remote sensing methods and Geographic Information System (GIS) tools. Object-based image analysis (OBIA) based on random forest (RF) and support vector machine (SVM) machine learning algorithms was performed to provide LULC classification maps, and then, LULC changes were assessed using GIS. In order to assess LULC changes, we used three images acquired using remote sensing, corresponding to 3 years; 1999, 2009, and 2020. A Sentinel-2 image and two Landsat images were used as input data in our methodology. Our LULC classification results showed that RF outperformed SVM on the three input data periods, with an overall accuracy of 95.6% obtained with the Sentinel-2 image. Given the changes over time, it is clear that the Allala watershed has undergone significant changes over the years, particularly an increase in building infrastructure and agricultural land due to population and urbanization growth. Analyzing and mapping the trends of LULC changes in the study area provide a basis for strategic planning and managing, and results of LULC changes can be used as a decision support tool and provide further help in regional and national land management.

Bathymetry studies are important to monitor the changes occurring in coastal topographies, to update navigation charts, and to understand the dynamics of the marine environment. Satellite-derived bathymetry enables rapid mapping of large coastal areas through measurement of optical penetration of the water column. In this study, bathymetry prediction is investigated using Pleiades multispectral satellite data. This research work explores the possibility of using very-high-resolution multispectral satellite data with a deep learning U-Net-inspired neural network architecture to infer bathymetry estimates around Europa Island (22^o20′S, 40^o22′E), which is a coralline island in the Mozambique Channel. This study is among the first to provide an overview suitable for bathymetry mapping using a deep learning approach based on optical satellite data. An airborne light detection and ranging (LiDAR) dataset of 1 m resolution is used as ground truth to train the model. From experiments, the overall accuracy evaluation of the model shows a good relationship (R² = 0.99, standard error = 0.492) between the predicted and reference depth values that satisfy the International Hydrographic Organization (IHO) S-57 Category of Zone of Confidence (CATZOC) levels A1, A2, B, and C (IHO, 2014). These predicted bathymetry values could potentially be incorporated into electronic navigational charts. The image reconstruction shows accurate results to estimate bathymetry in the shallow waters with mean absolute error not exceeding 1.5 m in that case. The U-Net-inspired deep learning technique exhibits promising outcomes to predict water depth from very-high-resolution satellite data to operate bathymetry mapping automatically over a wide area.

Located in the southeast of Tunisia, Jerba is considered a premier tourist destination offering beautiful sandy beaches. Since 1960, the island has undergone significant socio-economic transformations due to its tourism boom. Beach tourism is extremely popular among the population, which has contributed to an intensification of coastal vulnerability, wherein the beaches are threatened with disappearance. This work aims to identify the causes of coastal vulnerability and measure it based on different geophysical and socio-economic variables using the coastal vulnerability index (CVI) developed by Gornitz (Vulnerability of the US to future sea level rise, Coastal Zone, Proceedings of the 7th Symposium on Coastal and Ocean Management, American Society of Civil Engineers, 1345–1359, 1991). This allows us to identify the most vulnerable sites and to establish maps and data for coastal management purposes. The results obtained show 63% (14 km) of the coastline of the northeast coast of the island has a high to very high degree of vulnerability. Moreover, 37% of the coastline of the southeast of the island has a low to moderate vulnerability or about 22 km of the entire coastline. The quantitative measures relating to this coastal vulnerability can aid to fortify the coast against a rise in sea level.

The dike surrounding the coast of the Vietnamese Mekong Delta (VMD) helps prevent saltwater intrusion and coastal erosion. In order to preserve these dikes, an area of protective forest is planted outside them, helping to maintain and strengthen them. Therefore, monitoring the change in the area of planted forest outside the dike will help to assess the stability of and possible threats to the coastal area and ecosystem. In this paper, using Sentinel remote sensing images, we propose a new approach; applying the Random Forest technique to assess the changes in the planted forest outside the dike. The experimental results obtained on a typical coastal area of the Vietnamese Mekong Delta will help to clearly identify the threats and the evolution of the coastline through the changes in forest area outside the dike.

Seaside tourism is one of the most accessible summer activities for the population, especially on the coast of metropolitan cities. A typical example from the southern Mediterranean is Palm Beach, Algeria, one of the most populated beaches of the Algerian capital, where intense seasonal human traffic combined with the depletion of local sediments contributes to an intensification of coastal erosion. The entire beach was affected until the authorities started to build protective breakwaters. The objective of this work is, therefore, to establish a spatial monitoring of the dynamics of this beach, thanks to the Algerian Alsat2 satellite archives. The evaluation will be applied “before,” “during,” and “after” the construction using several spectral algorithms of detection of the shoreline. The monitoring will allow the quantitative observation of the behavior of the beach and thus to qualify the degrees of effectiveness of these breakwaters. It is also a question of determining the effectiveness of the high spatial resolution imagery (2.5 m) of this national optical sensor in the measurement of macro-erosion phenomena in order to generalize this approach to the entire Algerian coast in the future.

This study aimed to assess the current state of the shoreline and the effects of erosion on the shoreline in Vinh Chau Town, Soc Trang Province, Vietnam. Satellite image overlays were used to quantify the variation in the shoreline as a result of erosion caused by changes in wave action. In addition, the wave measurements were implemented at three representative shoreline protection sections (sea dike, mangrove forest belt, and the breakwater) to evaluate the wave height reduction at the shoreline. The results showed that erosion affected approximately 23 km (32%) of 72 km coastline of the study area. The erosion penetrates the land area from −16.9 to −3.0 m/year (Landsat images) and −11.68 to −7.95 m/year (Google Earth images); the coastline recession increases every year, leading to the gradual loss of mangrove forests and also farmland. The wave measurement shows the effectiveness in wave height reduction of the mangrove forest and the constructed breakwater to protect the sea dike. Wave height reduces more than 50% when passing through the mangrove forest belt, corresponding to a maximum height (H_max) of 62.3%, 1/10 H_max at 55.3%, and 1/3 H_max at 54%. For the constructed breakwater, the wave reduction efficiencies recorded due to Hmean are 72.18% and 1/10 H_max are 73.16% and reach 72.47% with 1/3 H_max. The results are based on wave measurement over a short time period; thus, it is not possible to conclude about the wave reduction efficiency of the current measures in the long term. It is necessary to monitor continuously and with different wind seasons to have a more accurate assessment of wave reduction efficiency.

The launch of satellites equipped with sensors in the optical range of the electromagnetic spectrum has greatly facilitated the mapping and monitoring of coastal areas for risk prediction. Thus, the frequent updating of information for monitoring purposes is possible. It is, therefore, a modern alternative to traditional methods, namely, photogrammetry and in situ investigation. The objective of this work is to define an efficient and validated method for the detection and extraction of shoreline indicators. It is the first indication of validation for a satellite image classification approach, based on a deep learning algorithm, optimized and adapted to the extraction, a hydrodynamic and biological indicator of the shoreline. The convolutional neural network (CNN) architecture was designed and adapted in order to extract the target shoreline indicators. A Pleiades image of very high resolution was used, sliced into sub-regions, and analyzed by a convolution kernel of size 3*3. The classification results have revealed a very high accuracy of 92%. A validation process was undertaken by comparing the results to field surveys (reference) acquired on the same day as the satellite image acquisition. With a run-up (horizontal wave excursion) of 0.6 m, the confidence interval for the deep learning method was estimated to be ± 0.42 m, which is quite small, revealing the good accuracy of the method tested. A large panel of users could reproduce these methods in an automatic and standard way, which should allow the updating of a possible database shared between involved parties in an efficient way.

Since the Middle Ages, man has tried to expand the surface of his territory on the sea by draining the coastal marshes through the polderization process [1]. These recently-conquered lands were intended for agriculture and grazing. However, the current climate change and the latest projections of the IPCC (AR6) suggest an issue in this territorialization [4].

Sandy coasts are often marked by an intense recession of the shoreline [1]. Recent advances in the radiometric, spatial, temporal, and spectral resolution of sensors have provided a valuable tool set for innovative coastal data processing methods. It has been demonstrated that satellite imagery, as well as new remote sensing methods, can provide more practical approaches to the mapping and monitoring of coastal environments.

A detailed morpho-sedimentological map of this bay, derived from the calibrated lidar rasterization with field data, provides new insights into the relationships between the nature and morphology of the sediment bodies and their overall distribution within this megatidal bay protected by numerous islands and rocky shoals.

The rich biodiversity of bays and estuary areas provides numerous services to human populations including food support, agricultural amendments [3], and tourism. These ecosystems are also the first players in coastal protection and erosion control [1]. In a global climate change context, the loss of biodiversity is critically depleting these ecosystem contributions. On the Channel French coast, bays and estuaries show large sediment cover variations especially in unique human-modified areas like the Rance estuary [5]. Calculating a discrete Shannon index eases the modelling of the benthic diversity by quantifying the proportion of each biological and geological class. Accurate descriptions of coastal basin structures are thus proposed using Shannon index evaluations [4].

Fish face multiple environmental pressures acting as multiscale filters structuring the community [6]. Study of the effect of local (~100 m²) habitat components such as habitat architecture, substrate composition, and benthic community on fish community is still limited because of the technical difficulties to sample reliable descriptors of all these habitat components. Nevertheless, the effect of the 3D architecture, especially the complexity, has been highlighted to act as an important variable, locally structuring fish communities and leading to an increase in diversity (species richness and Shannon index) and quantity (total density and biomass) of fish [1]. Photogrammetry is extending to submarine environment and allows to produce very fine information of the architecture and substrate composition [5], as well as of the benthic community [4]. The influence of the different habitat components (i.e., architecture, substrate composition, and benthic community) on the structure of fish community is here investigated, as well as the interest of photogrammetry in comparison to visual observation (Fig. 1).

The coastal areas are the theatre of increasing erosion and submersion risks by gathering growing hazards and exposures. Nature-based resilience is here mapped at 2 m spatial resolution using a novel fusion of morpho-bathymetry classification, derived from airborne bathymetric LiDAR, and graph-based network modelling. Connectivity results were discussed in light of coastal management.