Conference Agenda

Coastal marine areas form some of the densest biodiversity hotspots, with intertidal wetlands, such as seagrasses, mangroves and saltmarshes, covering vast portions of the intertidal area. Seagrass meadows directly and indirectly provide a wide range of ecosystem services (e.g. recreation; key forage, refuge and nursery habitats for fisheries species and non-targeted species; climate regulation; coastal stabilisation and water quality mediation). Unlike subtidal seagrasses, intertidal seagrass meadows directly provide services to both marine and terrestrial ecosystems, so monitoring their occurrence, extent, condition and diversity can be used to indicate the biodiversity and health of local ecosystems. The process of monitoring large intertidal areas is, however, resource intensive and unfeasible in many regions. Current global estimates of seagrass extent and recent comprehensive seagrass reviews either do not mention intertidal seagrasses and their seasonal variation, or combine them with subtidal seagrasses. Here, using cloud based composites of high-resolution satellite data acquired by the Sentinel-2 Multispectral Instrument (MSI) alongside a highly accurate neural network, we present the first full map of intertidal seagrasses in Europe. We found that cumulatively seagrasses cover an area similar to the land area of Luxembourg: 2110 ± 344 km². Although many Northern European countries have large intertidal seagrass total extents, the proportion of intertidal areas covered by intertidal seagrass decreased with latitude (from ~32 % at 58° to ~62 % at 35°). Furthermore, we showed clear latitudinal gradients in seagrass density, with high densities of seagrass being more prevalent in low latitudes and low densities being more prevelant in high latitudes. Finally, we showed a clear relationship between intertidal seagrass peak timing and latitude, going from 10 June at 58° to 27 November at 35°. This work has provided the first Europe wide intertidal seagrass map. Our seagrass map provides critical data for prioritising and developing policies, management and protection mechanisms across local, regional or international scales to safeguard these important ecosystems and the societies that dependent upon them.

Biodiversity is a vital component of natural capital that significantly influences ecosystem functions and provides essential services and benefits, ranging from food security to cultural heritage. However, species are currently disappearing at a rate 100 to 1,000 times higher than the natural extinction rate. Coastal ecosystems are particularly concerning: they are among the most vulnerable due to their exposure to cumulative anthropogenic pressures while biodiversity knowledge is lacking.

Supported by the French National Space Agency (CNES) and endorsed by the Space Climate Observatory (SCO), the BioEOS project aims to develop observation tools to characterize the spatiotemporal dynamics of coastal biodiversity. This initiative will map changes and produce operational indicators to assist in conservation and restoration efforts in the Marine Protected Area (MPA). The project primarily takes advantage of image time series from multispectral (Pleiades, Sentinel-2, Venus) and hyperspectral (EnMAP, PRISMA) satellite systems. A set of selected biodiversity proxy metrics are extracted using high SRL (Scientific Readiness Level) algorithms that have been widely used by the benthic scientific community. These algorithms encompass the inversion of radiative transfer models, machine learning-based scene segmentation, spectral unmixing, pansharpening, and the calculation of spectral indices. This approach enables to generate valuable information on bathymetry, bottom/habitat type abundances and distributions, as well as water column properties estimations. Coral reef and seagrasses of Southwestern Indian Ocean region (La Réunion, Mayotte, Glorieuses and Bassas da India) are the first targeted ecosystems for this experimentation.

We present the main advancements of a demonstrator providing key essential variables contributing to various end uses through four distinct use cases. Additionally, we will discuss the strengths and limitations of the satellite systems employed, in light of the initial objectives set forth.

As part of the ESA Coastal Blue Carbon project, our goal is to map and monitor caracteristics of blue carbon ecosystems (BCEs), such as extent, and subsequently estimate their biomass production and carbon storage potential using Earth observation data. To achieve this, we aim to ensure that the knowledge and techniques developed and tested at the local scale using very high spatial resolution imagery (<2m) remain reliable when scaled up with lower-resolution imagery such as Sentinel-2 (10m). To prepare for this scaling up, we propose an innovative approach designed to compare performance in terms of habitat mapping and vegetation biomass estimation across different image sources and remote sensing features. This approach also enables a rigorous benchmarking of various supervised classification methods (for habitat mapping) and multivariate regression methods (for biomass mapping).

We applied this approach to the salt marshes in the Arcachon Basin, France, in 2024, using a comprehensive experiment based on a time series of Sentinel-2 and Pléiades images. The method is structured around a fixed hexagonal grid, in this case with a 20-meter side length, which allows us to build reference data integrated into this area and usable at both resolutions. The advantage of this fixed grid is that it enables rigorous comparison of simple pixel-level classifiers (such as RandomForest) with more complex patch-level classifiers (such as CNNs and CNN-RNNs), which include neural networks tailored to analyze spatial information (e.g., textures and shapes) and temporal information (e.g., phenological trajectory patterns). This setup also allows for rigorous testing of architectures that combine Sentinel-2’s temporal information with Pléiades’ spatial resolution, offering a promising hybrid model. All predictions are then analyzed within these fixed grids, allowing for a precise interpretation and assessment of results, which in turn informs methodological decisions for scaling up.

Mangrove forests play a pivotal role in maintaining coastal biodiversity and supporting local livelihoods. They are among the most productive ecosystems on Earth, with a potential storage of organic carbon reaching 693 Mg C ha-1. Mapping and monitoring of mangrove carbon stocks over time represents a significant challenge for remote sensing studies. Indeed, greater consideration of the structural and functional diversity of mangrove stands is required to improve the accuracy of carbon maps.
As part of the ESA-funded Coastal Blue Carbon project (2024-2026), we have incorporated mangrove habitat diversity into a mapping model of aerial carbon stocks. Our approach uses extensive field data from forest inventories conducted in a diverse range of mangrove habitats since 1995. Subsequently, tree growth equations are employed to calculate the above-ground biomass (AGB) and carbon stocks of numerous forest stands at the time of acquisition of a large dataset of very high-resolution Pleiades satellite imagery (50 cm) over pilot sites in French Guiana, Amapá in Brazil, Suriname and Guyana.
The FOTO texture-based methodology (Fourier-based Textural Ordination algorithm, Proisy et al., 2007) was then applied to all Pleiades mangrove images. The objective was twofold: first, to map and label the diversity of mangrove habitats in terms of canopy properties; second, to predict the associated AGB at a 1-ha scale. The resulting AGB maps are transformed into carbon maps based on the total (soil, below- and above-ground) carbon storage model developed by Walcker et al. (2018) in French Guiana with the same field dataset.
Very high-resolution Earth Observation imagery for carbon stock assessment is critical for mapping and monitoring the blue carbon capacity of coastal ecosystems. The results support local decision-making for conservation and can inform global climate policy. However, these new results also highlight the need for new field data and new models of mangrove functioning.

Remote sensing is a key area of research that will strengthen the link between the different scientific disciplines involved in the Nature-based Solutions (NbS) framework. The present review covers three decades of remote sensing studies of healthy mangrove ecosystems in French Guiana. This effort is of great value in the context of the ESA-funded Coastal Blue Carbon project and two French national programmes (Solu-Biod and FairCarbon). We have identified three typical NbS themes associated with mangroves in French Guiana.

The first theme is the prediction of coastal dynamics and erosion based on operational mangrove spatial monitoring, modelled using time series of moderate-resolution satellite imagery. These predictions are then used to support coastal planning.

The second NbS theme concerns biodiversity, ecosystem functioning, resources and uses. We address this with the support of very high spatial resolution imagery, which is key for assessing mangrove habitats. We produce biomass and carbon maps, model fine-scale socio-economic surveys and describe the use of mangrove-derived resources to inform their management.

The third theme addresses the health of mangrove coasts, including ecosystem state and the associated potential risks for human health. Work on this theme uses the fine-scale characterization and monitoring of mangrove habitats to enable early detection of threats to the ecosystems, such as defoliation during caterpillar outbreaks. Furthermore, it permits investigating the hitherto largely unstudied risk posed by mosquitoes and culicoides in mangroves, which differ from vector communities found in urban areas.

It can be concluded from this research that remote sensing provides a strategic, operational and pioneering approach to anticipating coastal change in tropical regions. This allows for the rapid detection and public awareness of socio-environmental issues, as well as informing decision-making processes. Indeed, time series and remote sensing images facilitate understanding of global change and inform decisions about mangrove-dependent social-ecological systems.

Kelp forests (Order Laminariales) create incredibly complex and productive marine habitats which support marine biodiversity along 25% of global coastal shorelines. However, these critical ecosystems are in decline in many regions around the world. In order to enhance our understanding of kelp forest ecosystem dynamics, investigate drivers of change and assess conservation and restoration actions, spatial datasets and monitoring tools are needed. In British Columbia (BC), Canada, bull kelp (Nereocystis luetkeana) and giant kelp (Macrocystis pyrifera), are the two dominant kelp species and are located along a very complex coastal environment that presents unique mapping challenges. These challenges have led to the development of local, regional and coast-wide kelp mapping methods for a suite of remote sensing sensors and platforms (drones, aerial platforms and satellites) to map kelp forest extent and change through time. As part of a global kelp mapping community of practice, a time series of kelp extent data is being derived from the Landsat series of satellite sensors to create a coast-wide dataset from 1984 to present day in BC. In this work we describe how kelp forest extent data are derived from the Landsat imagery and how we are using local and regional spatial datasets to inform mapping accuracy and species-level considerations. This research informs ongoing work related to linking remote sensing data with available datasets for assessing kelp forest ecosystem productivity and biodiversity.

Coastal ecosystems can contribute significally to the carbon budget and climate change, particularly trough the concept of blue carbon. The Gross Primary Productivity (GPP) of mudflat is primarily due to the activity of microphytobenthos (MPB), a community of microscopic photosynthetic organisms that inhabit the upper layer of mudflats. Remote sensing of GPP contributes considerably in monitoring and upscaling the carbon fluxes for understanding their impact in climate change. From this perspective, this study conducted in estuarine environments in France, aims (1) to evaluate the spatio-temporal variation of GPP across different seasons and locations, as well as (2) to model GPP using hyperspectral indices coupled with environmental variables and direct carbon flux measurements. For this purpose, this research combines the hyperspectral remote sensing indices and environmental variables, including photosynthetically active radiation (PAR) and mudflat temperature with the CO2 chamber-based measurements of Net Ecosystem Exchange (NEE) and Respiration (R) to link direct measurements of GPP with remote sensing and environmental indices. The results show that the GPP measured values of MPB vary across seasons and locations, ranging from 144.26 mgC/m²/h to 289.08 mgC/m²/h. Remote sensing indices coupled with environmental variables capture these seasonal and spatial variations, allowing for reliable estimates of GPP.

Seagrasses are critical to coastal ecosystems, providing habitat, stabilizing sediments, and aiding carbon sequestration. Climate change has increased the frequency and intensity of heatwaves, potentially threatening seagrass health. This study investigates the impact of marine and atmospheric heatwaves on the pigment composition and reflectance of the intertidal seagrass Nanozostera noltei. We performed laboratory experiments, exposing N. noltei samples to controlled heatwave conditions and measured hyperspectral reflectance and pigment concentration to assess its impact over time. Results revealed that heatwaves induce significant declines in seagrass reflectance, particularly in the green and near-infrared regions, linked to (likely due to) pigment degradation. Key vegetation indices, such as the Normalized Difference Vegetation Index (NDVI) and Green Leaf Index (GLI), also displayed marked reductions under heatwave stress, with NDVI values decreasing by up to 34% and GLI by 57%. A novel Seagrass Darkening Index (SDI) was developed to identify seagrass darkening, showing a strong correlation with heatwave exposure. This research suggests that spectral monitoring can effectively track the early impacts of heatwaves on seagrasses, providing a valuable tool for remote sensing-based habitat assessment. Satellite observations confirmed these findings, showing widespread seagrass darkening during atmospheric and marine heatwave events in Quiberon, France. Darkened seagrasses observed after heatwaves were exposed more than 13.5 hours daily. This work highlights the need for continuous monitoring of seagrass meadows under the current climate regime, underscoring the potential of remote sensing in capturing rapid environmental changes in intertidal zones.