1. Monitoring Intertidal and coast
With the establishment of the National Park Schleswig-Holstein Wadden Sea, being a UNESCO World Heritage since 2011, the monitoring of important environmental parameters became necessary for the verification of the status and the planning of further development. Finally, a coordinated monitoring program could be agreed upon with the Wadden Sea riparian states The Netherlands, Germany and Denmark. The roughly 80 environmental parameters include sediment and seagrass abundance, both parameters that have to be collected for numerous other guidelines and have repercussions on the other biotopes. Sediment type, currents and exposure times are important factors for the biology and productivity of intertidal flat areas.
Seagrasses are the only flowering plants in the European Wadden Sea. They occur in the intertidal zones and constitute an important part of the food web due to their high productivity. Besides serving as a food source for migrating geese, seagrasses also provide a habitat for a variety of other species. In the North Frisian Wadden Sea, where the largest seagrass stocks can be found, seagrass meadows have been regularly monitored since 1994 using aerial surveys. After a significant decline in the 1930s, monitoring results have shown a steady increase in stock size until 2020. In almost all other areas of the world, however, seagrass is declining.
One main goal of the current work is to provide comparable measurements from remote and the ground. Different techniques are used, each showing advantages and disadvantages such as spatial inaccuracy (airborne), spatial coverage (ground based), individual interpretation (airborne, ground based) or spectral similarity of surfaces and atmospheric influences (satellite data). Thus, the comparison of the different data sets is difficult.

2. Satellite Acquisitions
Methods for detecting different habitats in intertidal flat areas based on satellite data have been developed since many years. They are currently applied in test phase for the operational monitoring of sediment and seagrass meadows. The distribution of sediment types - sand, mixed and muddy sediments - provide valuable information about potential distribution of benthic organisms. The changes of sediment and morphologic structures provide information about the adjustment of the system to external influences, such as climate change. For the detection of sediments, linear spectral unmixing is used and for the identification different vegetation indices are applied. Spectral and textural features are used within decision trees to classify and assess the different habitats. Water coverage is permanently changing on intertidal flats and therefore, the influence of water coverage as well as different degree of wetness has to be taken into account from the spectral signal of the target surfaces. Data availability is limited due to pre-conditions of low tide and low cloud coverage, but has improved since the launch of Sentinel-2 A and B. Satellite data are used in parallel to long-term monitoring programs since several years in order to assess the differences of the acquisition techniques, so that time series can be continued - and adjusted - for future programmes.

3. Verification methods for Seagrass

a. Airborne: Mapping by observation
For more than 30 years, an airborne low altitude survey of the Wadden Sea with observers has been carried out for rapid assessment. The procedure of airborne mapping provides maps of seagrass and green algae three times a year. The technique leads to a strong simplification of the observed occurrences, size and position inaccuracies, especially on outer tidal flats with low structural features.
Today's satellite image analyses are qualitatively much more accurate than the airborne mapping, but the new remote sensing method cannot distinguish spectrally between green algae and seagrass stands or provide indications of their species diversity. So far, the three annual aerial surveys are still used to determine the maximum spread - the characteristic value for the long-term population development - and intra-annual development over time.
b. Ground truth existing: References by estimation
In addition to airborne mapping, surveys of seagrass cover on the bottom is conducted. Due to the size of the area, only 1/6 of the full SH Wadden Sea area can be mapped within one year. Besides the surrounding of the seagrass meadows along the 5% coverage limit, transects crossing the meadows provide a state of the varying stand density, during which vital parameters, among others, are also observed. The data are used for verification of both aerial observations and satellite image analyses.
In addition, specific transects are captured that are used to optimize image classification or transform image classes into real values to be reported for monitoring. For this transects, photos are taken in different viewing angles and directions for later visual inspection.
c. Ground truth evolution
The assessments performed in the field are often subjective and can lead to inaccurate results due to the difficult conditions in the field.
The data collection during a transect survey leads up to 9 km through silty mudflats under often variable, sometimes harsh weather and light conditions. The coverage estimates are assessed by the mappers, yet the low tide periods are too short to make repeated observations for verification. Therefore, a measurement independent of the observers is desired.
In order to support ground truth field mapping, an automated method was developed using the RGB control images taken during a transect survey. For this, a series of images is shot in a top-down view, covering the area of the transect point. These are then split by a segmentation algorithm (SLIC). A neural network was trained to classify the resulting sub images into either three seagrass or several background categories. The seagrass segments are subjected to three independent methods to determine the coverage: a ‘simple’ method assigning the different seagrass classes a flat coverage percentage; the Green Leaf Index (GLI), a vegetation index utilizing the characteristics of light reflected by vegetation without having access to infra-red; and ‘Otsu’, an algorithm to differentiate between light and dark areas. The interpretation of the photos has to cope for the same problems as the remote acquired data: Each of these methods have different advantages and disadvantages in the presence of disruptive factors like water coverage, reflections, or unfavourable light conditions. Finally, the resulting relative coverage area of the transect point is given by the mean of the result of all contributing control images.
d. Measurements by AI (Ground truth and Drones)
In the future, moving to drones for image taking might increase efficiency, since it would be possible to cover larger areas in less time. It remains to be tested if the required image quality can be reliably produced in the often harsh weather conditions of the Wadden Sea, and if the proposed methods produce credible results even in the absence of parameters that can only be measured by an observer on the ground.

4. Future Reporting including EO
A crucial question in the use of remote sensing for legally relevant information is the transformation of image classification into real values - combined with information on the size of the uncertainty. In order to assess the transferability of assessment from image to image, the absolute reference provided by the photo interpretation as ground reference is an important step forward.
The combination of the various monitoring methods into a new permanent monitoring system and the development of a largely automated data flow from the satellite image to the reporting product is our mid-term goal for a reliable monitoring of the unique Wadden Sea and its habitats.


Bibliography
Duffy, James & Pratt, Laura & Anderson, Karen & Land, Peter & Shutler, Jamie. (2017). Spatial assessment of intertidal seagrass meadows using optical imaging systems and a lightweight drone. Estuarine, Coastal and Shelf Science. 200. 10.1016/j.ecss.2017.11.001.
Gade, K., Stelzer, K. & J. Kohlus (2013): Towards an Improved Classification System of Intertidal Flat Surfaces Based on Satellite Optical and Radar Data. Vortrag und abstract (p. 63), 33rd EARSeL Symposium 3. - 6.June 2013, 6th Workshop on Remote Sensing of the Coastal Zone, Matera, It..
Kohlus, Jörn; Stelzer,Kerstin; Müller, Gabriele; Smollich, Susan (2020): Mapping seagrass (Zostera) by remote sensing in the Schleswig-Holstein Wadden Sea, In: Estuarine, Coastal and Shelf Science; Spezial Issue: Asmus,Ragnhild; Schueckel, Ulrike; Eskildsen,Kai; Ricklefs, Klaus; Garthe, Stephan: From single ecological interactions to holistic assessments of coastal habitats in the Wadden Sea.
https://doi.org/10.1016/j.ecss.2020.106699
Müller, G, Stelzer, K., Smollich, S., Gade, M., Adolph, W., Melchionna, S., Kemme, L., Geißler, J., Millat, G., Reimers, H-C., Kohlus, J., Eskildsen, K. (2016): Remotely sensing the German Wadden Sea - a new approach to address national and international environmental legislation, In: Environmental Monitoring and Assessment, 188(10), 1-17; DOI 10.1007/s10661-016-5591-x. The article is available electronically
Reus, G. et al., "Looking for Seagrass: Deep Learning for Visual Coverage Estimation," 2018 OCEANS - MTS/IEEE Kobe Techno-Oceans (OTO), 2018, pp. 1-6, doi: 10.1109/OCEANSKOBE.2018.8559302.
Stelzer, K., Gade, M., Adolph, W. & J. Kohlus (2013): DeMarine-Environment - Suitability of RapidEye Data for Monitoring of the Wadden Sea. Vortrag und abstract, 5. RESA Workshop 20.03./ 21.03.2013 in Neustrelitz am Standort des DLR Neustrelitz.

Key information derived from multi-source optical imagery for supporting coastal territories management
Konrad Rolland, Anaïs Teissonnier, Anne Colliez, Emile Naiken
Collecte Localisation Satellite, PIGMA, région Nouvelle Aquitaine, DINAMIS

Coastal areas have been developing continuously for the past 40 years in terms of urban planning, demographics and economics; their attractiveness will continue in light of the latest prospective studies. A fragile, highly attractive area that drives the economy, the coastline is a geographical space where specific planning and management policies are deployed.
The imperatives of territorial recomposition and reconversion, whether driven by a logic of adaptation to the risks of climate change or a response to social and environmental changes, force us to question new models for thinking about and inventing a more resilient coastline. In a context of strong residential, recreational and tourist demands, of global changes in society, of increasing climatic and environmental challenges, and in order to succeed in the transition, it is necessary to rely on a detailed knowledge of developments in order to anticipate future changes and risks.
Within this framework, CLS has established a multi-date (1980s to 2020), accurate, reliable and homogeneous large-scale land use/Land cover database for the Nouvelle Aquitaine region (84,000 km² and a coastline of over 1,000 km). This complex and innovative project covers the entire spectrum of land observation, from the acquisition of image data to the analysis of the maps produced. The objective is to understand the signals of change in the territory and to anticipate future changes.
To build up this base, we researched, processed, and combined different image sources (panchromatic and infrared aerial images, but also VHR satellite images) but also DTM. We defined a production methodology in line with the needs of the territory and compatible with the timeframe. This solution used a mixed approach between automatic data and image processing and photointerpretation. It also made it possible to produce, for example, the latest 2020 land use/land cover databases based on low-cost satellite sources in a few months with reliability levels > 90%.
The quality and temporality (5 dates) of the land use/ land cover data allow us to establish scenarios of the state of the territory in 10, 20 or 50 years according to climate change models.
For this conference, we will present the methodology implemented but also the main results of the project to demonstrate how multi-source land observation data can be used in combination with other business data to understand and manage the development of a coastal territory.

Coastal marine environments, being invaluable ecosystems and host to many species, are under increasing pressure caused by anthropogenic impacts such as, among others, growing economic use, coastline changes and recreational activities. A continuous monitoring of those environments is of key importance for the identification of natural and manmade hazards, for an understanding of oceanic and atmospheric coastal processes, and eventually for a sustainable use of those vulnerable areas.

The joint Sino-European project “Remote Sensing of Changing Coastal Marine Environments” (ReSCCoME), as part of ESA’s DRAGON 5 Programme, addresses research and development activities that focus on the way, in which the rapidly increasing amount of high-resolution EO data can be used for the surveillance of marine coastal environments, and how EO sensors can detect and quantify processes and phenomena that are crucial for the local fauna and flora, for coastal residents and local authorities. We will present results from three ongoing ReSCCoME research activities, during which synthetic aperture radar (SAR) data are being used for different monitoring purposes in vulnerable costal marine environments.

Within the first research activity a classification scheme for sediments and habitats on intertidal flats in the German Wadden Sea was developed, whose feature set consists of the Freeman-Durden (F) and Cloude-Pottier (C) decomposition components, the Double-Bounce Eigenvalue Relative Difference (D) parameter, and two parameters derived from elements of the Kennaugh matrix (K). These feature sets are used as input for a Random Forst (RF) classification, and hence, the classification scheme is abbreviated FCDK-RF. Fully polarimetric SAR data acquired at L-, C- and X-band were used, and a comparison of the classification results with reference data obtained from optical data and field campaigns revealed that the FCDK feature set has good potential for the identification of sandy and mixed sediments on intertidal flats, even when they merge into, or mix with, each other, and for the detection of bivalve beds. A simplified FCDK-RF scheme for the use of dual-co-pol SAR data showed lower accuracies in the discrimination of mixed and sandy surfaces, but was well suited for the detection of bivalve beds.

The second research activity focuses on the impact of large offshore wind farms on the local and regional wind climate, with special emphasis on the wind farm wake effect. The wake region on the downstream side of a wind farm is typically characterized by reduced wind speeds and increased turbulence intensities lasting for tens of km. Within the coastal zone, wake effects can be mixed with other effects, e.g., horizontal wind speed gradients caused by the roughness transition between land and sea. Based on a large archive of wind maps retrieved from Sentinel-1 A/B SAR-C and Envisat ASAR scenes of the North Sea and the China Seas, sector-wise mean wind speeds were analyzed upstream vs. downstream of large offshore wind farms. SAR acquisitions before and after the wind farm construction were analyzed separately in order to quantify the velocity deficits caused by wind farms. A correction for horizontal wind speed gradients was applied to take the ‘background’ wind speed variability into account. Further, a correlation was found between the velocity deficits and the wind farm capacity per km², with larger wind farms leading to more pronounced velocity deficits.

Finally, the third research activity aims at different statistical analyses based on visual inspections of SAR data with respect to imprints of marine oil pollution. More than 2000 Sentinel-1 SAR-C and Envisat ASAR scenes of the Western Java Sea were analyzed and the ‘normalized oil pollution’ density was defined, which takes into account the SAR image coverage and local wind speed, both of which may affect the total number of oil spills observed at a certain location. This density was highest along the main shipping routes and at major oil production sites. Approximately half of the spills were smaller than 1 km², though spills of more than six times that size were also found. Since visual observations are always subjective, results from independent operators examining the same set of SAR images were compared and showed good qualitative agreement. The observed quantitative differences became smaller, when only SAR images acquired at higher wind speeds were considered, indicating that a confusion with biogenic slicks and low-wind areas is the main source of errors in oil pollution detection.

Inter-tidal zones globally are currently decreasing due factors including coastal development and sea-level rise that have put increasing pressures on these fragile environments (Murray et al, 2019). Such zones have implications for natural processes including bio-diversity, blue-carbon and coastal flood protection as well as for human processes including blue-economies, port-authorities and ecosystem services. Often inter-tidal areas are under-studied due to the inherent difficulties and danger of directly accessing them and the expense of existing mapping techniques. This talk will show how time-averaged inter-tidal bathymetry can be estimated by the Temporal Waterline (TWL) method and used to routinely monitor these zones.
Research at the UK National Oceanography Centre (NOC) (Bell et al, 2016; Bird et al 2018) has led to the development of a novel method for estimating bathymetry across the intertidal zone known as the Temporal Waterline Method (TWL). TWL processing, like spatial waterline methods requires a time-series of images spanning the tidal range and an estimate of the water-level at the time of each image. As well as being an active research topic, methods based on the use of fixed position X-Band Radar have been developed into a commercial service delivered by Marlan Maritime Technologies Ltd. Recent work at NOC has further developed the TWL methodology to exploit SAR (Synthetic Aperture Radar) satellite data, unlocking the potential for new (potentially global) intertidal zone monitoring solutions. Initially the UK tide-gauge archive was used to estimate the water-levels but this only allows processing of areas within 10 to 20 km of the tide-gauge. The most recent work has been to extend the method to use outputs from the FES2014 global tidal model, so that the TWL method can be applied globally.
The talk will present the methodology and show several UK case studies at both local and regional scales with examples of validation against LiDAR surveys. The potential for the routine monitoring of regional areas will be shown. The limitations and areas for improvement will also be discussed.

References:
Bell, Paul S.; Bird, Cai O.; Plater, Andrew J. 2016 A temporal waterline approach to mapping intertidal areas using X-band marine radar. Coastal Engineering, 107. 84-101. https://doi.org/10.1016/j.coastaleng.2015.09.009
Bird, Cai O.; Bell, Paul S.; Plater, Andrew J. 2017 Application of marine radar to monitoring seasonal and event-based changes in intertidal morphology. Geomorphology, 285. 1-15. https://doi.org/10.1016/j.geomorph.2017.02.002
Murray, Nicholas J.; Phinn, Stuart R.; DeWitt, Michael; Ferrari, Renata; Johnston, Renee; Lyons, Mitchell B.; Clinton, Nicholas; Thau, David; Fuller, Richard A. 2019 The global distribution and trajectory of tidal flats. Nature 565, no. 7738. 222-225.
https://www.nature.com/articles/s41586-018-0805-8?unique_ID=636808896127280679

Green macroalgae blooms have been persistently affecting the coasts of Brittany (France) since the 1970’s, causing losses of income to the fishing and touristic sectors. Macroalgae typically proliferate in confined coastal waters when nitrogen inputs, carried by rivers, exceed the assimilative capacity of the ecosystem. The current and tides uproot the macroalgae, leaving them stranded on beaches, where their decomposition might cause serious threats to animal and human’s health. Since 2002, the French Algae Technology and Innovation Center (CEVA) has been monitoring macroalgae surfaces on beaches at 95 sites using aerial photography. Along with this monitoring effort, the French government launched in 2010 anti-algae plans, with the aim of reducing nitrogen inputs and ultimately containing green macroalgae blooms frequency and severity.

Long-term estimates of macroalgae surfaces are however not available, limiting the understanding of temporal trends in green tides and the efficiency of macroalgae bloom reduction measures. Using freely and openly available Landsat imagery archives over 35 years (1984-2019), we automatically detected and quantified green macroalgae surfaces at four highly affected sites in Northern Brittany (Schreyers et al., 2021). Mean macroalgae coverage were characterized at annual and monthly scales. We demonstrate important interannual and seasonal fluctuations in macroalgae surfaces. Over the studied period, green macroalgae blooms did not show a decrease in extent at three out of the four studied sites, despite an observed decrease in nitrogen concentrations for the rivers draining the study sites.
In addition to this long-term trend analysis, we explored the potential of Sentinel-2 imagery for macroalgae surface detection and surface quantification. Sentinel-2 provides higher-spatial resolution (10 to 20 m) as well as more frequent observations than the Landsat sensors. Given the high temporal variability and persistent cloud coverage in Brittany, increasing the number of imagery available for detection can improve the accuracy of monitoring. We compare Sentinel-2 MSI derived macroalgae surfaces with i) Landsat-8 OLI estimates and ii) the aerial photography estimates from CEVA. We ultimately discuss the advantages and limitations of satellite-based monitoring of macroalgae proliferations and potential applications to other affected systems, for example in the Caribbean Sea, West Africa and the Gulf of Mexico where Sargassum bloom events have intensified since 2011.

References:
Louise Schreyers, Tim van Emmerik, Lauren Biermann, and Yves-François Le Lay. 2021. "Spotting Green Tides over Brittany from Space: Three Decades of Monitoring with Landsat Imagery" Remote Sensing 13, no. 8: 1408. https://doi.org/10.3390/rs13081408