Earth observation applications for coastal sustainability: potential and challenges for implementation

The coast is home to unique ecosystems, where complex ecological processes take place through the interaction of terrestrial, aquatic, atmospheric, and human landscapes. However, there are considerable knowledge and data gaps in achieving effective and future change-proof sustainable management of coastal zones around the world due to both technical and social barriers, as well as governance challenges. Currently, the role of Earth observation (EO) in addressing many of the recognised information gaps is small and under-utilised. While EO can provide much of the spatiotemporal information required for historical analysis and current status mapping, and offers the advantage of global coverage; its uptake can be limited by technical and methodological challenges associated mostly with lack of capacity and infrastructure, product accuracy and accessibility, costs, and institutional acceptance. While new initiatives and recent technological progress in the EO and information technology arena aim to tackle some of these issues so that EO products can be more easily used by non-EO experts, uptake is still limited. This paper discusses how EO can potentially inform transformative practices of planning in the coastal water zone, by using examples to demonstrate the EO potential in providing information relevant to decisionmaking framed by international agreements, such as the United Nations Agenda 2030, the Convention on Biological Diversity, and the Sendai Framework for Risk Reduction. By presenting evidence for how EO can contribute to innovative opportunities and data synergies at scale, the paper discusses opportunities and challenges for a more solution-led approach to sustainable coastal management.


Introduction
International efforts, such as aspirational global conventions including the United Nations 2030 Agenda for Sustainable Development (hereinafter 2030 Agenda), the Convention on Biological Diversity (CBD), the Sendai Framework for Disaster Risk Reduction, and the Conference of the Parties (COP21) Paris Agreement, call for progress to be made towards a more resilient and sustainable future worldwide. Delivering progress towards these conventions demands large amounts of different types of data from multiple sources. For the 2030 Agenda alone, a global indicator framework was developed by the Inter-Agency and Expert Group on Sustainable Development Goal Indicators (IAEG-SDGs) in 2017 that includes 232 indicators across 17 goals (United Nations 2018; www.un.org/ sustainabledevelopment/sustainable-development-goals/). The Sendai Framework, designed to function as a management tool to help countries develop disaster risk reduction strategies, make risk-informed policy decisions, and allocate resources to prevent new disaster risks, identifies another 38 indicators for four priority areas and seven targets (UNISDR 2015;www.unisdr.org/we/monitor/indicators). The data requirements for these two conventions alone are extensive, but when coupled with more well-established conventions, such as the CBD, which itself has five goals and 20 targets (CBD 1992;www.cbd.int/sp/targets/), the magnitude of the challenge becomes clearer. This unprecedented demand for data, coupled with an unparalleled rate of innovation in data collection techniques and technologies, and the capacity to distribute data widely and freely, has expanded the horizon of possibility for solutions. The adoption of 2030 Agenda and the Sendai Framework, both in 2015, present strategic opportunities to build on the momentum of the data revolution and demonstrate the centrality of data for development (Espey et al. 2015).
The need to recognise and contribute to data collection is particularly pertinent in a coastal and ocean context, where management-relevant data have traditionally been difficult to generate (Rumson et al. 2017). More than any other geographic space, coasts are the place where the greatest confluence of societal activities occurs. With approximately 356 000 km 2 of global coastline (Central Intelligence Agency 2016) benefiting from the convergence of some of the most productive and dynamic natural systems, coasts provide a large range of all four categories of ecosystem services: supporting, provisioning, regulating, and cultural services (Rodriguez et al. 2005;Neumann et al. 2017). Along the coast, services such as coastal hazard protection (Spalding et al. 2014), food and livelihood security (Visbeck et al. 2014), carbon sequestration (Chmura et al. 2003), and recreation opportunities (Spalding et al. 2017) are all essential for human well-being at every scale (e.g., see Wheeler et al. 2012 for well-being benefits associated with populations living by the coast in the UK). However, much of the world's coast represents a ribbon of exposure to natural hazards with a history of catastrophic impacts. Climate warming and sea-level rise (SLR) are already negatively affecting natural ecosystems and human communities (IPCC 2014). In addition, increased urbanisation at the coasts has both costs and benefits to society, directly in terms of the risks and utility derived from where people live and work as well as indirectly through distortions to cultural heritage, use patterns, and food chains (Pelling and Blackburn 2014).
Sustainable coastal management, despite increasingly more expressive and binding legislation, increased scientific endeavor, and investment, still poses numerous implementation problems around the world (Ali et al. 2018). Proposed solutions point to a need for more integrated approaches underpinned by participative and multidisciplinary processes, which support the sustainable use of coastal resources (Henocque and Denis 2001). Emphasis is placed on coordinated approaches to better facilitate collaborative working. While concepts such as integrated coastal management and integrated coastal area management are now mature (Birch and Reyes 2018), and much has been learned from implementation practices; certain recurrent challenges are likely to feature in the management and planning for coastal zones. Examples such as working with incomplete data for the region in question, insufficient understanding of the scale at which coastal systems operate, and undertaking and aggregating simultaneous monitoring of different activities are well described in the existing literature (e.g., Cash and Moser 2000;Cornu et al. 2014). However, sustainable development cannot be attained while disasters and unsustainable practices continue to undermine economic growth and social progress (United Nations level change (e.g., Dwarakish et al. 2009;Bosch et al. 2014); and urban development and coastal wetland loss (Wu et al. 2018;Rojas et al. 2019). In addition, endeavours such as Marine Spatial Planning (Ouellette and Getine 2016) and sustainable urban development at the coast (Corbane et al. 2008;Ban et al. 2015) actively incorporate multiple data sources, including EO data, to generate potential solutions to increasingly complex issues.
This paper aims to present four innovative aquatic EO applications whose relevance to the sustainable management and planning of coastal waters is currently being explored, and, using these examples, discuss the potential role EO can play in the delivery of international conventions, such as the 2030 Agenda, CBD, and Sendai Framework. By identifying potential data gaps, opportunities, and challenges, this paper begins to answer the key question of how EO can most effectively contribute to crafting options, approaches, and solutions towards sustainability (COASTS) for coastal regions of the world. Even though this paper focuses on four water applications as key examples, the authors acknowledge the existence of, and ongoing developments on, several other water, land, and climate or atmosphere EO applications that hold potential for the delivery of international conventions, but it is not within the scope of the paper to address those in detail.

Examples of earth observation applications with relevance to coasts
Exploitation of Big EO Data enables global mapping of natural resources along the coast, ranging from fisheries, to water quality and bathymetry, as well as natural hazards, including storm surges, coastal inundation, and SLR. Four innovative advanced EO applications are presented below as examples to showcase methodological advances in different disciplines of coastal water research and oceanography. Application 1. Optical water types for coastal water quality monitoring The monitoring of coastal water quality is a key requirement of national and international regulations, as well as water quality monitoring policies and programmes, such as the European Union's (EU) Water Framework Directive, USA's National Aquatic Resource Surveys -National Coastal Condition Assessment, and Australia's State of the Environment on Coastal Waters. However, ensuring that this can be done adequately and affordably remains a global challenge. In oceanic waters, where the optical complexity of the water is relatively low, the Ocean Colour suite (OCx) of algorithms (O'Reilly et al. 1998; ESA Climate Change Initiative Ocean Colour 2016; NASA Ocean Color Web 2019) are operationally used to produce global maps of chlorophyll-a estimates. The advantage of the OCx algorithm series (x denoting the number of wavebands used in the equation) over other ocean colour algorithms is that they work across a wide range of chlorophyll-a concentrations, and OC4 in particular provides a very high signal-to-noise ratio, and, thus, much more accurate chlorophyll-a retrievals than a variety of other empirical and semi-analytical algorithms (O'Reilly et al. 1998(O'Reilly et al. , 2000. In coastal waters, several studies have shown considerable potential for the application of EO-based methods for deriving water quality estimates (see reviews by Gholizadeh et al. (2016) and Odermatt et al. (2012)) over long temporal and spatial scales (i.e., regional, continental, and, ultimately, global), but the reliable application of these methods across time and space is complicated by the diversity of water types, sensor configuration, and inherent limitations of the approaches used (e.g., atmospheric effects, adjacency effect, sun glint, sea bottom reflectance, empirical algorithm restrictions) (Brewin et al. 2015;Mouw et al. 2015;Zheng and DiGiacomo 2017). In addition, water quality retrieval in optically complex coastal waters is often determined by a combination of spatially and temporally variable properties, such as phytoplankton, suspended material, and coloured dissolved organic matter, all of which affect water colour and transparency to varying degrees. To accurately map coastal water, any operational algorithm(s) would need to account for all these properties rather than focussing on each one of them individually. For this to be achieved, prior knowledge on water typology is required to assess the partial contribution of each property to the overall water quality in the water body under investigation. Unfortunately, this knowledge is currently unavailable for most coastal regions of the world and, as a result, no single operational algorithm has been developed yet for coastal waters.
A relatively new type of classification, based on the optical properties of coastal waters, provides novel avenues for progress, and its application has recently been demonstrated with the production of the global and long-term ESA Climate Change Initiative (CCI) Ocean Colour product (www.esa-oceancolour-cci.org/). The approach clusters coastal waters into optical water types (OWTs) (Moore et al. 2001), based solely on field-, air-or space-borne spectral data (Moore et al. 2014). It has been applied in both coastal and inland water studies with very promising results (Moore et al. 2014;Trochta et al. 2015;Eleveld et al. 2017;Spyrakos et al. 2018), and is now used by ESA CCI OC v3.1 in a blended approach to map chlorophyll-a concentrations globally (ESA Climate Change Initiative Ocean Colour 2016). The advantage of OWTs is that they enable grouping of complex coastal waters based on remotely sensed optical characteristics of coastal waters that are inherently dependent upon ecology without requiring information on ecology itself (Moore et al. 2009). Even though the exact and potentially complex relationship between OWTs and coastal water ecology has not yet been investigated in depth, previous studies have demonstrated the existence of relationships between spatiotemporal distributions of OWTs and physical and biogeochemical processes and ecological diversity indices (Moore et al. 2012;Mélin and Vantrepotte 2015;Trochta et al. 2015), signifying the potential benefit from inclusion of the approach in coastal ecological monitoring and water resource management.
The innovative OWT approach not only has the potential to enable stakeholders to map water quality, such as eutrophication, and primary productivity (by using chlorophyll-a concentrations as a proxy) in coastal waters without a priori or in situ information on coastal water body ecology, and, thus, exploit global remote sensing archives for previously unmonitored sites, but to also provide users with potentially more accurate retrievals of water quality, tailored to the dynamic optical complexity of the water body in question. And while the OWT approach is becoming a powerful tool for data analysts and product developers in improving the accuracy of their EO-based water quality products, the type of information delivered to coastal water managers remains unchanged (i.e., products such as chlorophyll-a maps, or water quality indicators, such as trophic indices, are still delivered, but are more reliable). However, as tends to be the case in applications based on optical EO data, the success of the OWT approach can be limited by the effectiveness of the atmospheric correction (AC) applied to the data. There are various AC models used over coastal waters (cf. Mograne et al. 2019), such as Polymer (Steinmetz et al. 2011) and Case 2 Regional Coast Colour (C2RCC) (Doerffer and Schiller 2007), but there is no single model that works well in all cases, so improved understanding of the limitations posed by ineffective AC is required before the OWT approach can be reliably applied at global scales.

Application 2. Species niche habitat distribution mapping
Fisheries are crucial for food and to maintain livelihoods across the globe (FAO 2018); many coastal communities are dependent on fishing or fishing-related industries for income and food security, with fishing being a key part of their cultural identity. Using satellite-derived data to deliver effective fisheries management (e.g., approaches advocated by the international community, such as Ecosystem Approach to Fisheries Management (Chassot et al. 2011;Staples et al. 2014)) ensures access to daily, global, systematic, and high-resolution datasets for incorporating habitat considerations into marine fish population dynamics (e.g., see Druon (2010) for use of remote sensing to support conservation of the Atlantic bluefin tuna), and in particular fish species that support economies of coastal communities. Whilst satellite imagery cannot be used to identify individual fish species, it can provide critical information on the environmental pressures that shape a species population's survival success. For example, nursery habitats, such as free-floating Sargassum mats in the tropical Atlantic and Caribbean, which are considered as both a benefit (e.g., by providing a key fish nursery habitat (Wells and Rooker 2004)) and a hindrance (when periodic mass landings of the free-floating macroalgae strain local economies (Webster and Linton 2013;). These mats can now be detected, characterized, and forecast using optical remote sensing imagery (e.g., Maréchal et al. 2017;Wang and Hu 2017;. Hu et al. (2015) overview how remote sensing analysts can use the red-edge in satellite spectra (i.e., the comparison of high reflectance in the near-infrared region to the low reflectance in the red region of the electromagnetic spectrum, characteristic of vegetation) to remotely detect Sargassum presence or absence on the sea surface. However, Sargassum detection and quantification is often hindered by its spectral similarity to other floating organisms, materials, and often by the inadequate spatial resolution of remote sensing products. Hu et al. (2015) collected and recorded a collection of Sargassum spectra, noting that a distinctive reflectance, caused by the presence and levels of chlorophyll-c pigmentation in Sargassum tissue, can be used to effectively differentiate Sargassum from all other materials, provided there is a minimum of 20%-30% coverage within a satellite pixel. Wang and Hu (2017) describe how these satellite spectra records have been integrated into a Sargassum mapping algorithm. A stepwise process is used to eliminate all data that are of either poor quality or have very little chance of containing Sargassum. This is followed by conducting spectral un-mixing on Sargassum pixels using the spectra reported in 2015 as endmembers. By determining the fractional coverage of each pixel, distribution and area coverage maps can be produced. This process forms the basis of an automated Sargassum detection and biomass estimation process, underpinning the satellite-based Sargassum Watch System (Wang and Hu 2017). Through use of MODIS data archives, it also provisions researchers with a source of hindcast data with which to build predictive models for bloom forecasting (Wang and Hu 2017), with reported accuracies of over 80% in eastern Caribbean regions. There are limitations regarding the reliance on optical imagery, the need to adapt to cloud cover, and the lack of extensive in situ comparisons to accompany products with quality information. However, despite these limitations, this research narrative demonstrates how EO data can feed into advisory services and coastal fisheries management activities, whilst providing fisheries managers with quantifiable estimates of fish nursery habitat extents (e.g., . It is not only habitats that can be examined using EO. Satellite-derived data on habitat parameters can also feed into species models to understand the impacts of changes on their populations. An example of how niche characteristics, which impact a commercial species population over the course of their life cycle, can be examined and modelled is demonstrated in work done by Garrido et al. (2017). The authors clarified the influence that environmental factors measurable by satellite remote sensing had on the lifecycle of the European sardine Sardina pilchardus. Using satellite-derived sea surface temperature (SST) and estimates of photosynthetic activity based on chlorophyll-a measurements, the authors found that, in general, high recruitment years are associated with high chlorophyll-a and low SST levels. Such work exemplifies how actionable information can be garnered from satellite remote sensing, enabling those tasked with responsibilities for fisheries management and regulation to take appropriate management actions, and conduct more sustainable and integrated economic activities.

Application 3. Complementary multi-platform coastal bathymetry
Spatial data on coastal areas are key to the management of coastal and marine environments, covering aspects such as boundary and jurisdiction delineation; establishment of baseline, navigation, and seafloor change detection; issuing of consents and licences; and maintaining an inventory of coastal assets. Coastal bathymetry is an important source of spatial data used mainly for safe shipping and navigation (e.g., Jagalingam et al. 2015), marine surveying and mapping (e.g., Huang et al. 2017), but also classification of habitats and modelling of species distribution over a variety of sea floor terrains (e.g., Kostylev et al. 2003;Lundblad et al. 2006;Wilson et al. 2007). Traditionally, active remote sensing methods, such as ship-borne multi-beam echo sounder (MBES) and airborne light detection and ranging (LiDAR), have achieved accurate bathymetry results (Calder and Mayer 2003;Van Son et al. 2009;Kennedy et al. 2014). However, these conventional methods are very expensive, time-consuming, often require special permissions to work in sensitive coastal areas, and, in the case of MBES, inapplicable to shallow waters (e.g., Gao 2009;Monteys et al. 2015;Bannari and Kadhem 2017).
The complementary use of satellite EO and UAV surveying may provide a more efficient and cost-effective method of bathymetric data collection in coastal areas (Lejot et al. 2007;Klemas 2015b). On one hand, multispectral and hyperspectral EO data are already used as a valuable tool in deriving nearshore bathymetry (e.g., Hamylton et al. 2015). Being a nonintrusive technique, it allows for data collection, and the consequent production of bathymetry and seafloor maps, without causing any disturbance to local wildlife and ecosystems. On the other hand, UAVs provide the opportunity to fill data gaps and act as a validation technique to existing EO data products. UAVs, in particular, helicopter platforms, offer increased mobility, hover capabilities, and lower operating costs compared to conventional high-spatial-resolution data capture methods (Klemas 2015b). Collection of spectral data concurrent with satellite pass-over times allows for cross-calibration and validation of processed satellite data. This complementary duel approach has the potential to provide many benefits: low cost, operational flexibility and versatility, ground truth data, reduction of the influence of cloud cover, and surveying of typically inaccessible coastal waters and coastal terrain (Long et al. 2016). However, satellite-and UAV-derived bathymetry data are still susceptible to limitations based on AC inefficiencies and coastal complexities that hinder the optical bottom signal detected at the water surface (Gao 2009;Pacheco et al. 2015). Distortion of the spectral signature due to turbid waters and (or) biota coverage, such as seagrass, proves a challenging restraint to the application of satellite-derived bathymetry across broad coastal areas (Gao 2009;Caballero et al. 2019). Research into the impacts of such coastal complexities on satellite-derived bathymetry and accompanying depth limitations is required to determine confidence levels and identify optimum algorithms (Halls and Costin 2016;Caballero et al. 2019). Integrating high-resolution optical remote sensing with habitat classification data to ascertain the relationship between water reflectance, water depth, and seafloor habitat coverage is used to demonstrate the production of complementary bathymetry and habitat data from satellite data (Lyons et al. 2011;Traganos et al. 2018), with successful applications in coral reef bathymetry (Huang et al. 2017) and seagrass mapping (Hossain et al. 2015). Bathymetric mapping of valuable substrate, such as coral reefs, is a useful tool for monitoring vulnerable areas or areas that undergo topographic change (Vanderstraete et al. 2003) and ESA's Sentinel-2 satellite captures nearly 90% of the world's coral reefs, providing vast potential for high-resolution monitoring of benthic change (Hedley et al. 2018).
The cost-effective repeatability of coastal bathymetry products through EO and UAV multispectral data collection offers vast improvements to coastal management challenges and ecosystem monitoring, and offers invaluable up-to-date insights into how shorelines respond to climate change events, globally. An ever-increasing fleet of optical EO satellites, and advancements in UAV technology, can only improve the collection, processing, and production of accurate and reliable coastal bathymetry products.
Application 4. Coastal inundation mapping and prediction, and storm surge risk assessment Changing climates and increased variability present many threats and risks to both human and biophysical systems at multiple scales across the globe; this means that integrating short-term disaster risk reduction measures with long-term climate change adaptation is an ever-increasing priority. One of the challenges with integrated risk management approaches remains identifying climate variables and developing robust, reliable, longterm climate observations linked with extreme events (EEA 2017). In response to this challenge, the Global Climate Observing System has identified 54 essential climate variables (WMO 2016) several of which can be derived from satellite EO data (Hollmann et al. 2013).
Sea level has been identified as an essential climate variable. SLR has historically been measured using coastal tide gauges; however, the development of satellite altimeters has enabled the extrapolation of measurements further offshore providing a wider, consistent spatial context at global scales as well as a greater understanding of casual factors (Ablain et al. 2015). Long-term trends in SLR can now be derived from over 50 years of combined in situ and remote sensing data (Cabanes et al. 2001). The applicability of remote sensing in mapping SLR at global scales has also been demonstrated for a 20-year period  by the ESA CCI Sea Level product (www.esa-sealevel-cci.org/). Furthermore, work by Nicholls and Cazanave (2010) demonstrated the capability to supplement long-term trend estimates with quantifications of regional variability. This ability to translate global trends down to sea basins and regional shorelines allows responsive adaptive management across both temporal and spatial scales.
Even though satellite altimetry is a mature technology for studying open oceans, nearcoastal applications are impeded because of mixed pixels and contaminated return signals caused by hard surfaces, such as nearshore rocky outcrops and coastal infrastructure; this results in unreliable sea level estimates close to the coast (Vignudelli et al. 2005). However, altimetry specialists have begun extracting usable information from data previously considered to be of poor quality. For instance, total water level envelope estimates (i.e., the sea surface level taking into account tides, waves, effects of a storm surge, and (usually) precipitation and river flows) can now be extracted up to a distance of 3 km from the shoreline, a distance that is decreasing as sensor capabilities and processing algorithms improve (Vignudelli et al. 2011). Despite these advancements, the establishment of targeted nearshore (i.e., 3-50 km from the coastline) and offshore (further than 50 km from the coastline) sea level monitoring sites allowing for direct satellite-to-field measurement comparisons would be beneficial to gauge the accuracy of satellite-derived sea level data.
In addition to SLR, one of the most promising applications of coastal altimetry is in the study of storm surges (Cipollini et al. 2014). Impacts from storm surges and associated coastal flooding events are anticipated to increase in severity (e.g., for the Atlantic, see Bernier and Thompson (2006) and Tebaldi et al. (2012)) and so reliable modelling of storm surges can support both preparation and mitigation activities by improving understanding of how waves propagate and impact on coastal infrastructure. Improvements in this understanding could bring enormous societal benefits, particularly to populated low-elevation coastal zones. Research by Madsen et al. (2015) demonstrated how, using a stationary blending method, coastal satellite altimetry data can be related with corresponding tide gauge measurements, allowing generation of sea level maps and reducing storm surge forecast error. Liu et al. (2018) developed a risk and vulnerability assessment approach combining coastal hydrological observations, numerical storm surge modelling, and multi-source remote sensing data to map coastal infrastructure and land use. This approach was then used to assess overall risk of storm surges. Another example includes the use of UAVs to estimate the extent of coastal flooding and assess damages in critical areas demonstrating innovative EO capacities in the support of coastal hazard risk management activities (Popescu et al. 2017).

Supporting the delivery of international conventions through EO
The 2015 adoption of these landmark UN agreements, in combination with the CBD (1992), the COP21's Paris Climate Conference (2015), and Habitat III's New Urban Agenda (2017) created a significant opportunity to build coherence across overlapping social and ecological policy spaces. The lack of integration across strategies, policies, and implementation has long been perceived as one of the main barriers to more sustainable development (International Council for Science 2017). Insufficient accounting for trade-offs and synergies at multiple scales has resulted in incoherence, adverse and unexpected impacts, and ultimately in diverging outcomes and trends across global objectives for sustainable development (Neumann et al. 2017). However, when taken in combination, this new suit of frameworks covers a more comprehensive resilience agenda spanning development, humanitarian, climate, and disaster risk reduction areas.
This reality implies a need for greater cohesion at the implementation level. The previous section of this paper presented four novel and rapidly advancing EO applications with strong relevance to various aspects of coastal sustainability and management. These four example applications also highlight the potential for EO data and derived information to supplement and support the simultaneous delivery of multiple indicators, targets, and goals of different international conventions, such as 2030 Agenda, the Sendai Framework, and the CBD ( Table 1). The inherent interconnectedness of many of the goals and targets of international conventions demonstrate the potential for added value when applying technologies, such as EO, to support implementation. As shown in Table 1, the four EO applications described in this study can indirectly and directly support 19 UN SDG indicators pertaining to SD Goals 1, 2, 6, 11, 12, 13, and 14. This is in agreement with a similar synthesis performed by the international initiative "Earth Observations for the Sustainable Development Goals (EO4SDG)" on the most likely UN SDG targets and indicators to which EO can contribute (GEO 2016). Many of the same EO-based data and information products can also be used to inform and support five Sendai Targets (12 Sendai indicators) and two CDB Goals (six CBD targets) (Table 1).
It should be noted that EO data and information can contribute indirectly (i.e., support) or directly (i.e., measurement) to the above-mentioned indicators, targets, and goals. For example, indirect contribution can be demonstrated using SDG2 Target 2.4, Indicator 2.4.1 "Proportion of agricultural area under productive and sustainable agriculture". Agricultural runoff is a major cause of eutrophication and deterioration of water quality in coastal areas (Ongley 1996). Water quality measurements provide an indicator of the nature and health of an ecosystem and degraded water quality can lead to a reduction in marine biodiversity (Jackson 2008) as well as dramatically impact quality and quantity of ecosystem services (Barbier et al. 2011) leading to knock-on effects on human health and well-being (Myers et al. 2013). Quantifying the amount of land under agricultural use in coastal zones is therefore a key baseline statistic that can be used as a proxy for nonpoint-source pollution, and EO-based land cover and use mapping can play an important role in this. By supplementing that knowledge with information on coastal water quality and trophic status along coastlines (see section entitled "Application 1. Optical water types for coastal water quality monitoring"), targeted action toward better management of agricultural runoff and establishment of measures for sustainable agriculture can be taken to to the people at the national and local levels G-6 -Percentage of population exposed to or at risk from disasters protected through pre-emptive evacuation following early warning improve overall ecosystem health and tackle biodiversity loss. Therefore, the task of enhancing EO-based land cover and use data with accurate OWT-based water quality retrievals can help support the implementation of both SDG Indicator 2.4.1 and CDB Target B.8. In fact, in most cases presented in Table 1, the role of EO is supportive and supplementary to overall implementation. There are also cases where EO measurements can directly contribute to an indicator, such as in the case of SDG6 Target 6.3 Indicator 6.3.2 "Proportion of bodies of water with good ambient water quality". EO has been used for decades as a tool for local, regional, and global coastal water monitoring. The most recent technological and methodological advances (e.g., see section entitled "Application 1. Optical water types for coastal water quality monitoring") have shown great promise for coastal water quality retrievals (e.g., Moore et al. 2014) and various initiatives are now engaging local and national stakeholders and decisionmakers to build trust and capacity in EO-based water monitoring (e.g., Paterson et al. 2018; UN Environment 2018) and explore the implementation of the SDGs using EO at the regional level (e.g., Plag and the Workshop Participants 2018).
While there are numerous benefits of using EO technology as an operational support tool for sustainability planning, implementation, and monitoring, there is an acceptance that EO data are not a one-size-fits-all solution to increased coastal sustainability and improved management. The incorporation of EO data can complement existing spatiotemporal data collections as well as allow for inaccessible areas to be more efficiently monitored and managed, but these data must be used in conjunction with in situ social and biophysical datasets.
Recognising challenges, and opportunities, with EO data use Whilst international conventions are signed at the global level, actions and successes take place at national and sub-national scales. However, limited capacity and capabilities, both human and technical, drive the disparity in current EO use across regions (north versus south and west versus east) and restrict much of the potential uses of EO applications at national and sub-national scales, especially in the global south (Vanhove et al. 2017) (i.e., countries seen as low and middle income in Asia, Africa, Latin America, and the Caribbean by the World Bank). In addition, the translation of the relevance of technical applications such as EO to decision support knowledge, along with language surrounding multiple monitoring and evaluation mechanisms, such as the indicators and targets described in Table 1, is often a difficult link to make. Also, past experiences of a variety of state actors, working for various state agencies, have resulted in a continued express of concern around data sources, data processing, and uses of data (EDPS 2015); this has resulted in mistrust and political rejection of methodologies, technologies, and data products (von Maurich and Golkar 2018), highlighting the added issue of limited institutional acceptance of EO data and information.
Beyond issues stemming from lack of infrastructure and capacity, product accuracy, and reliability, lack of validation data, data continuity, and cost have been identified as main reasons as to why management decisions do not typically rely on EO-derived water quality products, according to a recent survey that was conducted within the USA Environmental Protection Agency (Schaeffer et al. 2013). Product accuracy and reliability and lack of suitable validation data are issues that the EO community has begun to address in recent years. Through interaction with users and stakeholders, and collaborations with in situ data holders, EO products have become more accurate and application-driven as, for example, in the case of the ESA CCI programme (http://cci.esa.int/) and projects such as EU H2020 CyanoAlert (www.cyanoalert.com/), EU H2020 CoastObs (https://coastobs.eu/), and ESA Thematic Exploitation Platforms (TEPs; https://tep.eo.esa.int/). Large and open access marine and terrestrial data repositories of in situ measurements, for example, SeaDataNet (www.seadatanet.org/), EMODnet (www.emodnet.eu/), Argo (www.argo.ucsd.edu/ index.html), Global Ocean Observing System (GOOS; www.goosocean.org/), and International Soil Moisture Network (ISMN; https://ismn.geo.tuwien.ac.at/) serve as sources of validation data for water and land remote sensing. Examples of large-scale validation efforts include ESA MERMAID (http://mermaid.acri.fr/home/home.php); a centralised database of merged in situ optical measurements and concurrent ENVISAT MERIS acquisitions for ocean research, UK NERC LIMNADES (www.limnades.org/home.psp); a centralised database of worldwide bio-optical measurements and match-up data for remote sensing, and EU H2020 MONOCLE (www.monocle-h2020.eu/Home); a project that is currently developing in situ observation solutions for EO-based coastal and inland water quality. Finally, the Group on Earth Observations "system of systems" approach (GEOSS; www.earthobservations.org/geoss.php) combines information from different sources and promotes common technical standards so that different datasets can be combined in a coherent manner towards advancement of EO techniques and applications.
Sensor discontinuity has in the past affected the remote sensing community. In the field of ocean colour and water monitoring, the most recent example was when ENVISAT MERIS unexpectedly failed in April 2012 creating a 4 year data gap in medium spatial resolution, multi-spectral EO-based water monitoring until the launch of its successor Sentinel-3A OLCI in February 2016. However, the consensus on this is changing and space agencies have started developing satellite sensors that offer continuity to previous missions, with one of the best examples of this effort being the NASA-USGS (US Geological Survey) Landsat-series. At the same time, methodological advances enable researchers to better harmonise datasets from different instruments and, thus, generate merged long-term datasets, such as the global multi-decadal products of the ESA CCI programme (see, e.g., the Soil Moisture (Dorigo et al. 2017) and Ocean Colour (Mélin et al. 2017) products). Furthermore, these methodological advances are coupled with a recent increase in free data provision by large initiatives, such as the European Copernicus programme (http://copernicus.eu/), providing longterm spatially explicit time series of free applications-focused data.
Challenges that exist around barriers to effective data access, sharing, management, storage, and usage are numerous and complex. Many institutional and financial barriers to data access exist, including costs for data already collected and processed, often with public funding, and license conditions on the use of the data and publication of results (Beniston et al. 2012). In some cases, for example, data collected may be highly political and beyond the influence of the scientific community (Giuliani et al. 2017; and in other cases there may be geographical or temporal sparseness of data reducing utility (Espey et al. 2015; Plag and the Workshop Participants 2018). In the case of the USGS Landsat archive, recent considerations around changing the open data access policy to cost-sharing models (LAG 2019) could mean that the current free access to the longest environmental satellite archive may be restrained by one's budget in the future and, thus, restricting data use and stifling innovation and business activity (NGAC 2012). In addition, limited investment to centralise or secure different types of data, and (or) storing the same type of data in various formats with little compatibility can create cascading issues across potential or actual users. While there are examples of efforts under way in Europe to remove the obstacles and costs related to data access for the purposes of research and policy-making, in particular, the European INSPIRE Directive of 2007 (http://inspire.jrc.ec.europa.eu/), the trajectory of progress is still slow and additional efforts are still needed by most EU Member States towards identification of spatial datasets, provision of services for access to spatial data and alignment of spatial data with regards to the common data models under INSPIRE (Cetl et al. 2017).
In recent years, EO organisations and funding bodies have begun to address some of the abovementioned issues and challenges by attempting to utilise cloud computing as a means of allowing access to EO data and processing capabilities under intuitive cloud processing platforms (e.g., EU H2020 Co-ReSyF and ESA TEPs). This emerging trend of thematically dedicated EO cloud platforms aims to provide virtual environments which not only facilitate the use of EO data for thematic research (e.g., coastal, but also urban, forestry, geohazards, and food security), but also broaden the uptake of EO data use in scientific communities by providing improved processing capabilities, easier data access, and more accessible infrastructure. A key step in the initiation of said platforms is stakeholder engagement within the scientific community. Building services based on stakeholder needs and recommendations not only ensures relevant and sustainable platform development but it also serves to engage, educate, and animate the wider community with regard to the capabilities of EO.
Finally, one of the most common limitations of optical (see the sections entitled Application 1. Optical water types for coastal water quality monitoring; Application 2. Species niche habitat distribution mapping; and Application 3. Complementary multi-platform coastal bathymetry) and thermal (e.g., used for SSTsee the section entitled Application 2. Species niche habitat distribution mapping) remote sensing applications is the effectiveness of the AC applied to the EO data. In turbid coastal waters, standard AC models often exhibit large inaccuracies (Fan et al. 2017) and previous studies have shown that not one single AC works in all cases (e.g., Larnicol et al. 2018;M. Zhang et al. 2018). However, AC models, such as the Polymer and C2RCC have shown great potential in optically complex waters (Mograne et al. 2019), such as those found in coastal areas, and it may be the case that different ACs are suited for different regions, coastal water bodies or applications. For example, the image correction for atmospheric effects (iCOR) algorithm was developed for both land and water pixels to reduce discontinuities in reflectance caused by separate application of water-and landspecific AC within one scene (De Keukelaere et al. 2018), and could therefore be a suitable option for transitional areas, such as the coastal zone.

Discussion and conclusions
There is a growing realisation and acceptance that sustainable development cannot be attained while disasters and unsustainable practices continue to undermine economic growth and social progress at many scales (United Nations 2015); this remains most pertinent in the coastal zones of the world. This reality has led to an ever-increasing demand from decision makers and planners alike for salient, near-real-time information and decision-support tools that can provide a sound evidence base to help enable the realisation of both the sustainability and the disaster risk reduction agendas in tandem. Earth observation holds much unrealised potential in data and knowledge provision in the field of coastal management. Data inaccuracy, cost, and discontinuity have in the past prohibited long-term and reliable retrievals of EO data, but this is changing. With the recent open data policy adopted by large space agencies around the world (e.g., NASA and ESA), users can now freely download the full archives of satellite sensors, including the 40 year long NASA-USGS Landsat archive and the new Copernicus Sentinel-series datasets. In addition, a variety of EO products with known and improved accuracies now exist (see, e.g., the ESA Climate Change Initiative programme) and the remote sensing community, supported by global initiatives, such the Group on Earth Observations, has been tackling issues such as product validation and AC to further improve product accuracy and uptake. With user engagement initiatives and related funding opportunities also increasing in number in the last few years, there has been an increase in contributions towards building national capacity around the world and overcoming obstacles, such as lack of infrastructure and training. Examples include the ESA-funded "Utilising Earth Observation to support Blue Growth and Risk Management in the Caribbean", which coupled capacity building with user engagement and consultation  to produce a stakeholder-led set of actions and a roadmap  for enhancing EO support for targeted coastal issues and opportunities.
This paper clearly demonstrates that EO can be used to provide a more synoptic view of coastal water systems that can support and enhance sustainable management. While this paper has focused on only four innovative applications, there exist many more satellite EO-based applications that should also be considered and integrated into efforts towards the implementation of sustainable management practices. As it was not the intention of this study to list all suitable, or potentially suitable, EO applications covering water, land, and air in the coastal zone, we used four coastal applications as key examples to demonstrate the applicability of EO for coastal sustainable management and discuss associated challenges. There is also a strong case for the increased use of archived EO datasets to analyse long-term trends now that improved technical cloud processing-based applications are becoming more user-friendly, and data are becoming more accessible through shifting data sharing practices. The paper also demonstrates the value of EO applications in supporting the implementation of multiple policy instruments and conventions that are currently driving the global transition to sustainable development, both as stand-alone sources of information and as complementary data sources to more traditional in situ collections. This, in turn, emphasises the importance of transdisciplinary collaborations to realise the full potential to transform existing management practices and access as-yet undefined solution spaces within coastal management.