TOPICAL COMMUNICATIONS The UK needs an open data portal dedicated to coastal flood and erosion hazard risk and resilience 1

: In the UK, coastal flooding and erosion are two of the primary climate-related hazards to communities, businesses, and infrastructure. To better address the ramifications of those hazards, now and into the future, the UK needs to transform its scattered, frag- mented coastal data resources into a systematic, integrated portal for quality-assured, publicly accessible open data. Such a portal would support analyses of coastal risk and resilience by hosting, in addition to data layers for coastal flooding and erosion, a diverse array of spatial datasets for building footprints, infrastructure networks, land use, popula-tion, and various socio-economic measures and indicators derived from survey and census data. The portal would facilitate novel combinations of spatial data layers to yield scientifi-cally, societally, and economically beneficial insights into UK coastal systems. a national-scale, quantitative, analytical map of risk from coastal flood and erosion hazard in England using existing open datasets. We found that this could not be done to our collective satisfaction — nor to the satisfaction of nearly forty well-informed stakeholders at a national workshop that we hosted. Difficulties stemmed from the availability, accessibility, and quality of the necessary datasets: gaps in the spatial data that precluded a national syn-thesis; proprietary and thus inaccessible data sets; inconsistent levels of spatial and temporal resolution; incompatible analytical methodologies between related datasets; and information that had simply never been gathered.

a national-scale, quantitative, analytical map of risk from coastal flood and erosion hazard in England using existing open datasets. We found that this could not be done to our collective satisfactionnor to the satisfaction of nearly forty well-informed stakeholders at a national workshop that we hosted. Difficulties stemmed from the availability, accessibility, and quality of the necessary datasets: gaps in the spatial data that precluded a national synthesis; proprietary and thus inaccessible data sets; inconsistent levels of spatial and temporal resolution; incompatible analytical methodologies between related datasets; and information that had simply never been gathered.
Analyses of risk and resilience to coastal hazard like the kind we attempted matter because, in the UK, flooding and coastal change are leading climate-related hazards to communities, businesses, and infrastructure (CCC 2018). Managing the impacts of flooding and coastal change carries a heavy financial burden (Penning-Rowsell 2015; Uberoi and Priestley 2017;EA 2018). Reports to UK Government on flood risk (Uberoi and Priestley 2017) highlight the need for more maintenance spending on flood protection, efficiency savings to offset costs of new defences, and "value for money" analysis of local flood protection. The UK Department for Environment, Food & Rural Affairs (Defra) recently announced a project titled "Updating guidance on shoreline management plans: UK Coastal Database", motivated by the fact that "to date there is no record of the total loss of homes, land or infrastructure on the coast", and there exists no clear, systematic way to estimate what future losses might occur under different climate scenarios (Defra 2020). The Environment Agency has a statutory duty, per the Flood and Water Management Act of 2010, to develop and deliver a National Flood and Coastal Erosion Risk Management (FCERM) Strategy for England, which was recently revised (EA 2020a). In November 2020, the Environment Agency and Defra announced a £200 million Flood and Coastal Resilience Innovation Programme in England, which will fund competitively selected projects to run into 2027; FCERM was allocated a total budget of £5.2B from 2021 to 2027 (EA/Defra 2020).
Our national analysis confirmed that England lacks the comprehensive, qualitycontrolled, compatible, and collated open datasets of coastal hazard, exposure, and defences required to assess spatial patterns of risk and resilience (Box 1). Analysis of those patterns support data-driven, forward-looking decisions for sustainable management of current and future coastal systems. We emphasise open data, meaning data that anyone can access, use, and share, typically with attribution; "open access" typically connotes content, particularly publications, that might otherwise be restricted by a paywall; "open source" refers to source code under an open licence (OKF 2021). There are proprietary databases and data products maintained by the insurance industry, engineering consultancies, and private geospatial companies. There are also relevant datasets maintained by government agencies but not necessarily publicly available. In some cases, awareness of certain datasets (and their provenance) depends on the institutional knowledge of a handful of individuals nearing retirement. Many datasets that are available lack the completion and standardisation needed to systematically assess coastal risk or resilience (Box 2). We found potentially relatable datasets that were not standardised or coordinated scattered across a fragmented network of organisations with responsibility for coastal protection and defences infrastructure. Some datasets exist for one nation of the UK (e.g., England or Scotland) but not the others, forcing certain comparisons to end abruptly at political rather than geographical boundaries. There are also plentiful "raw" data sources availablehistorical maps, ortho-rectified aerial imagery, lidar, bathymetry, and morethat are not yet processed into standardised data products (e.g., benchmarked shoreline position) ready for data users. This is not a plea for more datathe Big Data revolution and rapid expansion of remote-sensing capabilities are already ensuring that more data are becoming available. Rather, this is a call for quality-assured, open data, as a catalyst not only for innovation in analytical and fundamental scientific insight, but also for the delivery of coastal risk and resilience strategy and planning. The UK has an opportunity to take better care of the diverse coastal spatial datasets it already has developed, and to build the data-management infrastructure for new generations of spatial data productsincluding those from remotesensing technologies that are yet to be operationalised. A portal with open data dedicated to the component systems from which coastal flood and erosion risk emerge (i.e., not only characteristics of the coastal hazards, but also the assets and populations exposed and vulnerable to those hazards) needs to be regarded as an essential and achievable national resource and priority.

Examples of issues encountered with spatial datasets in England
The spatial scale of our attempt to evaluate coastal flooding and erosion hazard risk was effectively set by the most complete spatial coverage of coastal defences that we could source. A dataset of English coastal defences, both engineered and natural, is available through the Channel Coastal Observatory (CCO 2020b), and is based on the 1997 Coastal Protection Survey of England (Coastal Protection Survey Dataset 1997) and aerial photography. Aside from extending only to England, the dataset is valuable but incomplete: for example, the dataset only includes open coastline and does not follow estuarine shorelines, despite the widespread presence of defences there; no beach-nourishment works are Box 1. Defining coastal risk.
There are many ways to map risk from geohazards, but most combine probabilistic representations of physical hazard, exposure of assets or people, and (or) vulnerability.
Natural coastal phenomena such as storm-driven flooding and erosion are hazards when they threaten damage to human communities or environmental resources that people value (Lavell et al. 2012). Exposure may refer to people, infrastructure, and socio-economic and environmental assets that are subject to potential damage or loss in the event of a hazard occurrence. Vulnerability attempts to characterise ways in which exposed people and assets may be adversely affected by a hazard, especially where hazard impacts may have differential effects across a demographic mosaic (Cutter and Emrich 2006;Lavell et al. 2012;NRC 2014).
Natural coastal systems, such as beaches and marshes, can buffer some of the flood and storm impacts on exposed populations and assets, but their protective capacitieswhich also have limitsare often compromised by development pressures. As a result, on many developed coastlines around the world, engineered hazard protection plays an important role: infrastructure like "hard" seawalls or "soft" beach nourishment can buffer exposed populations and assets from all but very large-magnitude hazard events. Flood defences are built to a design standard of, for example, a 1:100 year flood event (0.01 likelihood of occurring each year). Seawalls might have an expected lifespan on the order of a century, whereas beach nourishment requires sustained, cyclical renourishment every few years.
Hazard protection alters the probabilistic distribution of hazard events (Werner and McNamara 2007) and may also encourage additional development behind itan unintended feedback variously termed "the levee effect" or the "safe-development paradox" (White 1945;Tobin 1995;Burby 2006;Di Baldassarre et al. 2015, 2018Armstrong et al. 2016;Armstrong and Lazarus 2019). Spatial connectivity may add further complexity, if the failure of defences in one location results in damage at another, as can occur in many low-lying floodplains (Wang et al. 2019).
Quantifying coastal risk in some coastal regions, such as estuaries or large bays, may be especially challenging because flood hazard can arise from oceanographic (storm surges plus tides and (or) waves), fluvial (increased river discharge) and (or) pluvial (direct surface runoff) sources. Most existing flood risk assessments consider these main drivers of flooding separately, despite their intrinsic correlation. Depending on local geographic characteristics (which influence lag times between flooding drivers), "compound flood events" can result in disproportionately extreme impacts (Wahl et al. 2018;Zscheischler et al. 2018). Compound events remain underexamined and excluded from disaster-management plansan omission that fundamentally and seriously biases existing flood risk assessments.
included; nor does the dataset include records of defence installation, maintenance, functional condition, or repairs. We note that there is a national statutory requirement to maintain a registry of inland flood defences, but not coastal defences. Despite the ubiquity of beach-nourishment projects around the country, the UK lacks any comprehensive record of their application, cost, volume, or spatial extent. The review of European beachnourishment practices by Hanson et al. (2002) is nearly two decades old, and unlike the US dataset maintained by the Program for the Study of Developed Shorelines (PSDS 2020), its underlying dataset is not publicly available.
Given the extent to which readily erodible shorelines in England and the wider UK are constrained by coastal-defence infrastructure, information on hard and soft defences, and their management, is vital. In addition to the Coastal Protection Survey of England from 1997 (Coastal Protection Survey Dataset 1997), there is the National Flood and Coastal Defence Database, now included within the Environment Agency's new Asset Information Management System, but this only includes assets under the auspices of the Environment Agency in England, omitting defences under other jurisdictions. The National Receptors Dataset likewise provides some information on assets and property at risk, but access is limited by a restricted licence (EA 2020b). The problem extends to other UK nations. Reporting for Scotland's recent comprehensive national assessment of coastal change (Dynamic Coast 2020) notes that data availability for coastal defences around the Scottish coast is Box 2. Completion and standardisation of dataset attributes.
Geospatial coastal datasets tend to be structured and managed differently by different local authorities and other agencies, making the collation and integration of data at the devolved-administration or UK-wide scale a challenging process. A basic dataset attribute that would aid integration is spatial coverage and spatial registration to a common basal reference. At present, coastal datasets do not necessarily include both the open coast and estuaries, for example. (Users can always exclude what they do not want, but they cannot include data that do not exist.) For datasets that track (or could track) changes over time, versions and metadata that are not recorded consistentlythat is, according to a standardised protocolultimately hinder efforts to investigate evolving, spatially correlated relationships among hazard, exposure, and vulnerability.
We suggest that key coastal data to support UK coastal assessments might include the following: • coastal physical characteristics, morphology and physiography and material • coastal erosion/accretion datasetsgeospatial data and attribute data by erosion and accretion mechanism, reclamation • coastal-defence data, record of defences over time, by type, condition, and maintenance actions, defended area • natural defence types, structure, standards, and condition • event records, by type (e.g., landslip, erosion, flooding), severity, and impact • coastal setback actions/managed realignment/natural breaches, locations, extent and mechanisms • assets/infrastructure defended, including buried infrastructure • records of losses, by actions and costs incurred in response to erosion and flood • monitoring types, responsibilities, and costs • historic properties/development histories, by type • location of planning policies for protection vs development, land cover, habitat, and land-use histories • coastal community structure and historic records of demographic change and disadvantage (sensitivity and adaptive capacity) "nationally patchy and has not yet been assimilated into a single and standardised dataset" .
To address coastal-erosion hazard at a spatial scale that matched the coastal defences dataset for England, we ultimately used a Landsat-derived global dataset of shorelinechange trends (Luijendijk et al. 2018) because it was the only resource that offered complete, standardised coverage of shoreline change at a spatial scale greater than sub-national regions. England-wide data ostensibly exist from the FutureCoast project (FutureCoast 2002), but these are not in a readily accessible format and are approaching two decades of dormancy. The Environment Agency National Coastal Erosion Risk Map (EA 2020c) comprises binned projections of future change based on past erosion rates, and thus as a data product is some steps removed from the data that underpin it.
To capture broad categories of flood likelihoods on coastal floodplains in the presence of current flood defences, we used the Environment Agency "Risk of Flooding by Rivers and Sea" dataset (EA 2020b). However, because that dataset does not include specific information about flood source (i.e., river or sea), we overlayed the "Flood Map for Planning (Rivers and Sea)" (EA 2020d) to define areas of coastal floodplain susceptible to flooding from coastal, tidal and (or) fluvial events. Notably, the polygons that comprise these two datasets -"Risk of Flooding by Rivers and Sea" and "Flood Map for Planning (Rivers and Sea)"differ in their spatial extents because the former considers the influence of extant flood defences and the latter does not.
These examples illustrate just some of the data-assimilation issues we encounteredeven having limited our analysis to England.

The data-management legacy of Shoreline Management Plans
Much of the impetus for a data-driven understanding of national coastal flood and erosion risk, is to gain an integrated vantage of regional Shoreline Management Plans (SMPs). Shoreline Management Plans are non-statutory, large-scale, long-term strategic plans that aim at reducing the risks of coastal flooding and erosion on population, infrastructures, and natural environments (Cooper et al. 2002). The first generation of SMPs were developed in the 1990swith contributions from a few of the authors hereand segmented the coastline of England and Wales into 11 littoral cells and 46 sub-cells according to general patterns of alongshore sediment transport (Motyka and Brampton 1993;Leafe et al. 1998;Cooper et al. 2002;Nicholls et al. 2013). The process of establishing the SMPs prompted recommendations for an improved evidence base of coastal change, which ultimately led to the creation of the National Network of Coastal Monitoring Programmes of England (CCO 2020a). Revised between 2006 and2011, 22 SMPs, subdivided into nearly 2000 Policy Units, presently cover the coastline of England and Wales. Shoreline Management Plans have also been applied to some reaches of Scotland's coast (Dynamic Coast 2020).
Data compilation and analysis for previous rounds of coastal assessments in England and Wales, particularly in the late 1990s, were outsourced to consultants, but those datasets were largely lost or remain proprietary information, rather than being made publicly available. Different SMPs employed different consultants, introducing methodological disparities and differences in quality control (Potts 1999). Regional studies have used different methods of shoreline-change analysis, for example, without standardising to a common data framework, complicating the essential process of stitching regional datasets into a freely accessible, searchable national inventory. The recent Infrastructure UK review (EA 2014) recognised the need for better asset data, to be supported by the Creating Asset Management Capacity (CAMC) programme, including improved records for defences, such as berm-crest levels and standard-of-protection. Five years later, recognition of that need has not yet translated into accessible, publicly available data products or a platform for themthough user communities of coastal data remain hopeful.
For now, separate databases for different jurisdictions, the lack of integrated datasets from local to national scales, inconsistent data protocols, and the patchiness of public availability present significant hurdles to any transparent and reproducible analysis of UK coastal flood and erosion risk using open data. Availability of baseline coastal data has been highlighted by the UK Geospatial Commission as a national spatial data infrastructure need (Geospatial Commission 2019; Admiralty Marine Data Portal 2020). Further work is planned through the UK Hydrographic Office (UKHO) to support the greater understanding of the British coastline via the Coastal Zone Mapping Project (UKHO 2020). This initiative is currently specifying best-practice and collating an understanding of needs and auditing current data "so that integration, discoverability and access to this data can be improved" (UKHO 2020). In addition, the UKHO has developed an automated mapping of the present coastline from Sentinel 2 satellite data, which will provide an updated framework for coastal mapping and open data. National agencies and regional groups are developing their own platforms of standardised, open data for coastal environments and coastal change, such as the Regional Flood and Coastal Committees (RFCC) Decision Support Tool, which provides web-based applications for the East Anglia RFCC region (RFCC 2020), and the data resources from the National Coastal Change Assessment in Scotland (Dynamic Coast 2020), which were created as an evidence base for strategic management .

From risk to resiliencea portal imagined
Using open data with common standards to develop a more holistic, multi-dimensional perspective of coastal risk can reinforce policy instruments of coastal management in a world where sea-level rise and climate change are recognised as a growing threat to livelihoods and lives (Rumson and Hallet 2018). Beyond risk, there is growing interest in measuring and enhancing resilience to coastal hazards (Rosati et al. 2015;Masselink and Lazarus 2019;Townend et al. 2021). If risk represents systemic exposure to disruption by a hazard, then resilience extends to how a system anticipates and recovers from disruption. While there are a set of established metrics for risk, metrics for resilience are still taking shape (Masselink and Lazarus 2019). The data portal proposed here will greatly facilitate the development of such metrics, which are multi-dimensional, requiring stakeholder valuation and multi-criteria analysis (e.g., Townend et al. 2021).
In the UK, some coastal data acquisition, processing, and analysis is undertaken and archived by the National Network of Regional Coastal Monitoring Programmes of England and the British Geological Survey. The Environment Agencyand its equivalents in the devolved national administrationsalso maintains their own geomatics teams, in charge of surveying, remote sensing, and data analysis. Independent research teams funded by national research councils also generate new coastal geospatial datasets, including repeated high-resolution imagery, topographic and bathymetric scans, and surveys of coastal ecological biodiversity. Where public money is spent on data-generating projects via national funding bodies, a framework for the provision and management of open data could ensure national standards across datasets, rapidly integrate new datasets into the national catalogue, generate simple but valuable products from these data (e.g., shorelines from orthophotos and structure-from-motion terrains).
To integrate these and other coastal data sources, both archival and new, the National Network of Regional Coastal Monitoring Programmes of England is an obvious hostalthough quality control, standardisation, and geospatial analysis (e.g., systematic shoreline delineation) are resource-intensive activities. But to support analyses of coastal risk and potentially coastal resiliencenot just coastal hazardany such portal will need to integrate a wide array of spatial datasets for building footprints, infrastructure networks, land use, heritage sites, ecosystem services, population, and various socio-economic measures and indicators derived from survey and census data. The portal could ensure that different datasets could be readily and reliably downloaded and integrated to facilitate novel analyses of spatial relationships of interest to a given user.
One example of new, value-added data resources that such a portal could provide would be layers of housing footprints, infrastructure, transportation networks, and coastal defences digitised from detailed (1:2500) historical maps, of which the UK has a rich catalogue. Such a resource would enable quantitative assessments of how patterns of coastal risk have evolved in space and time. These patterns could be linked to datasets derived from census data, such as indices of social disadvantage at the coast (UK Parliament HL 2019), and to historic hazard events, such as data archived by SurgeWatch (Haigh et al. 2017). The data portal could also include repeated empirical and modelled assessments of natural defencesbeaches, tidal wetlandsthat may be impacted by human activities, given that changes in the states and behaviours of those natural systems can affect, and be affected by, engineered interventions. By including coastal physical topography, management units such as mapped floodplains, and administrative units such as post codes and local authorities, users would be free to define the coastal zone according to their specific focusby some fixed shoreline, or a threshold elevation, or official delineationand pursue anything from local case studies to regional comparisons to a national assessment. Moreover, users could select from different data levels (e.g., raw imagery, post-processed/simplified layers, value-added analytics), spatial scales, and temporal series, depending on their analytical needs.

Realising a resource
One existing model of a standardised, searchable, freely accessible platform for coastal datasetsnational-scale coverage of sea-level rise impacts and short-and long-term shoreline change, along with hurricane strikes and geomorphic forecasts of storm-driven changeis the USGS Coastal Change Hazards Portal (USGS 2020a), which is further reinforced by the USGS EarthExplorer (USGS 2020b). Other examples include coastal portals for Scotland (Dynamic Coast 2020), Belgium (Flanders Marine Institute 2020), and the Netherlands (Rijkswaterstaat 2020). The European Topic Centre on Inland, Coastal and Marine waters, an international consortium working with the European Environment Agency, has likewise highlighted a vision for the assimilation of coastal datasets (ETC-ICM 2020). Our concept of an open-data portal aligns with and encourages the ambitions articulated in a recent strategy document by the Environment Agency for a revamped National Flood Risk Assessment tool that would use an open-data framework to provide "a single picture of flood and coastal risk" (EA 2020a). That a government agency, or agency partner, serve as host is not an essential requirement here; a willing and able third-sector organisation could conceivably support an open-access portal. But having already funded the acquisition of many of these coastal datasets, the UK government has a clear vested interest in guaranteeing their fruition as a public good (Nagaraj et al. 2020). We also leave conceptual room for the possibility that user involvement in a coastal open-data portal could evolve to enable active data production by users, such as a quality-assured, open-source platform on which users might add, update, and download data layers within an integrative framework (e.g., OpenStreetMap: https://www.openstreetmap.org/about).
With an open-data portal for risk and resilience to coastal flood and erosion hazard, exciting new analytical outcomes and insights could emerge from users creating novel combinations and analyses of spatial datasets, facilitated by robust data management. Unanticipated scientific insights into UK coastal systems may better support existing monitoring and assessment initiatives, and motivate novel research programmes. The portal for which we advocate would not only comprise a public good unto itself, but also enable societal and economic benefits of innovation and discovery from analysis of those data (Zhu et al. 2019;Nagaraj et al. 2020;Tassa 2020)precisely because they are open to all.