Multiscale remote sensing of intertidal vegetation of European coasts in response to natural and anthropogenic pressures.

Habitats formed by marine vegetation in intertidal zones that are exposed at low tide (seagrass beds, microphytobenthos, macroalgae) are heavily impacted by human activities. Seagrass beds are threatened by numerous anthropogenic activities (McKenzie et al., 2020), microphytobenthos is affected by the global decrease in intertidal mudflats (Murray et al., 2019), and the expansion of wild oysters may have reduced the areas colonized by macroalgae (Le Bris et al., 2016). These habitats fulfill recognized ecological functions, including protection against coastal erosion, mitigation of eutrophication effects, absorption of atmospheric CO2, and acting as biodiversity hotspots that support specific flora and fauna.

However, intertidal zones, such as mudflats, are challenging to access, and traditional field sampling would require excessive time and effort to cover the target areas. Nonetheless, regulatory requirements for monitoring the ecological status of coastal marine habitats necessitate regular mapping. This is the case with the Water Framework Directive (WFD) or the Marine Strategy Framework Directive (MSFD), which use the diversity of marine habitats as bioindicators of coastal or estuarine water quality (Borja et al., 2013; Zoffoli et al., 2021).

Spatial remote sensing is an innovative tool for studying essential variables for the biodiversity of these habitats (Peirera et al., 2013; Skidmore et al., 2015). However, past and current satellite missions do not possess optimal technical characteristics (spatial, spectral, and temporal resolution) to be fully operational (Muller-Karger et al., 2018). For some habitats, multispectral resolution may be sufficient under certain conditions (Zoffoli et al., 2020), but there are still risks of confusion. For others, higher resolution or the presence of specific spectral bands is essential for distinguishing taxonomically distinct organism classes (Fyfe et al., 2003; Launeau et al., 2018; Méléder et al., 2018). The identification principle relies, in part, on the presence of absorption bands in the visible spectrum related to the presence of photosynthetic and accessory pigments, which can be detected and quantified using high-pressure liquid chromatography (Méléder et al., 2003, 2005; Bargain et al., 2013; Jesus et al., 2014). The analysis of the phenology of these organisms could also be exploited to aid in their discrimination.

With its revisit time of 3 to 5 days, the Sentinel 2 satellite can also help identify seasonal cycles of marine vegetation (Zoffoli et al., 2020). Temporal variations in intertidal vegetation have been less documented than spatial variations, mainly due to the limited availability of high-resolution satellite images (e.g., SPOT, LANDSAT) and the constraints of acquiring images at low tide related to the study object. Therefore, there is currently no sensor that can provide high spatial, spectral, and temporal resolutions simultaneously. Consequently, this thesis will employ a multi-sensor analysis (Sentinel 2; MODIS) and will notably utilize very high-resolution spatial mapping using drones.

The objectives are related to mapping the spatiotemporal variations of plant biodiversity in intertidal zones through remote sensing, particularly using the Sentinel 2 satellite. The aim is to analyze the main natural and anthropogenic factors responsible for these changes. The ultimate goal is to demonstrate that remote sensing can meet the regulatory requirement for monitoring the ecological status of habitats of community interest (Papathanasopoulou et al., 2019) and that marine angiosperms could be used as an indicator of water quality.

My PhD

The aim of my doctoral research is to map the vegetation that grows in the zone comprised between high and low tide, known as the intertidal zone. This area is primarily vegetated by algae and a few flowering plant species. The problem lies in the fact that some of these plants share a similar pigment composition. This means that they reflect more or less the same colors, which is technically referred to as having the same spectral signature. This poses a challenge when attempting to differentiate between these vegetation types using remote sensing because color is the only information that is measured by this technique.

Therefore,my objective is to find advanced methods based machine learning to effectively distinguish between intertidal vegetation.

What is remote sensing ?

Remote sensing is a method of gathering data about an object without being in direct contact with it. It’s like using satellites or drones to capture information from a distance. This could involve taking pictures (either in the visible or the infrared spectrum), measuring temperature, or even penetrating through clouds and vegetation to understand the terrain (with Lidar or Radar, for example). It’s a useful technique for studying the Earth, monitoring the environment, and making informed decisions about activities like agriculture, urban development, or environmental management.

In my case, I primarily use remote sensing to observe phenomena in the visible and near-infrared regions of the electromagnetic spectrum, specifically the wavelengths where the sun emits the majority of its energy. It is precisely within these wavelengths, know as Photosynthetically Active Radiation (PAR), that plant pigments absorb light to carry out photosynthesis.

Spectral resolution

When discussing remote sensing, one of the most important technical characteristics is known as spectral resolution. Spectral resolution refers to the number of different “colors” (wavelengths) that a sensor can measure. We can distinguish between multispectral sensors, which can measure energy only in a few bands, and hyperspectral sensors, which are capable of recording energy in hundreds of bands. While hyperspectral data provides a more extensive set of information, it tends to be more difficult to process and typically noisier than multispectral data.

In my case, I mainly work with multispectral data because it is the most common type of data systematically acquired by satellite. Some products, like Sentinel-2, are freely accessible. When it comes to drones, the acquisition and processing of multispectral data are much less demanding compared to hyperspectral data. Furthermore, a recent publication has shown that the contribution of hyperspectral data was negligible in accurately discriminating some types of intertidal macrophytes.

Multiscale remote sensing of intertidal vegetation of European coasts in response to natural and anthropogenic pressures.

Habitats formed by marine vegetation in intertidal zones that are exposed at low tide (seagrass beds, microphytobenthos, macroalgae) are heavily impacted by human activities. Seagrass beds are threatened by numerous anthropogenic activities (McKenzie et al., 2020), microphytobenthos is affected by the global decrease in intertidal mudflats (Murray et al., 2019), and the expansion of wild oysters may have reduced the areas colonized by macroalgae (Le Bris et al., 2016). These habitats fulfill recognized ecological functions, including protection against coastal erosion, mitigation of eutrophication effects, absorption of atmospheric CO2, and acting as biodiversity hotspots that support specific flora and fauna.

However, intertidal zones, such as mudflats, are challenging to access, and traditional field sampling would require excessive time and effort to cover the target areas. Nonetheless, regulatory requirements for monitoring the ecological status of coastal marine habitats necessitate regular mapping. This is the case with the Water Framework Directive (WFD) or the Marine Strategy Framework Directive (MSFD), which use the diversity of marine habitats as bioindicators of coastal or estuarine water quality (Borja et al., 2013; Zoffoli et al., 2021).

Spatial remote sensing is an innovative tool for studying essential variables for the biodiversity of these habitats (Peirera et al., 2013; Skidmore et al., 2015). However, past and current satellite missions do not possess optimal technical characteristics (spatial, spectral, and temporal resolution) to be fully operational (Muller-Karger et al., 2018). For some habitats, multispectral resolution may be sufficient under certain conditions (Zoffoli et al., 2020), but there are still risks of confusion. For others, higher resolution or the presence of specific spectral bands is essential for distinguishing taxonomically distinct organism classes (Fyfe et al., 2003; Launeau et al., 2018; Méléder et al., 2018). The identification principle relies, in part, on the presence of absorption bands in the visible spectrum related to the presence of photosynthetic and accessory pigments, which can be detected and quantified using high-pressure liquid chromatography (Méléder et al., 2003, 2005; Bargain et al., 2013; Jesus et al., 2014). The analysis of the phenology of these organisms could also be exploited to aid in their discrimination.

With its revisit time of 3 to 5 days, the Sentinel 2 satellite can also help identify seasonal cycles of marine vegetation (Zoffoli et al., 2020). Temporal variations in intertidal vegetation have been less documented than spatial variations, mainly due to the limited availability of high-resolution satellite images (e.g., SPOT, LANDSAT) and the constraints of acquiring images at low tide related to the study object. Therefore, there is currently no sensor that can provide high spatial, spectral, and temporal resolutions simultaneously. Consequently, this thesis will employ a multi-sensor analysis (Sentinel 2; MODIS) and will notably utilize very high-resolution spatial mapping using drones.

The objectives are related to mapping the spatiotemporal variations of plant biodiversity in intertidal zones through remote sensing, particularly using the Sentinel 2 satellite. The aim is to analyze the main natural and anthropogenic factors responsible for these changes. The ultimate goal is to demonstrate that remote sensing can meet the regulatory requirement for monitoring the ecological status of habitats of community interest (Papathanasopoulou et al., 2019) and that marine angiosperms could be used as an indicator of water quality.