2.1 Nautical charts in the digital era

The information that helps the navigator can be divided into two general categories, the (relatively) unchanged in time or static and the time-changing or dynamic. The first category is related to seamarks and landmarks, which are visible from the bridge of the vessel and indicate a danger or a passage or route that the vessel may follow within a channel. The second category concerns objects in motion close to the vessel's course, with data dynamically generated by navigation-supporting devices such as the automated identification systems (AISs) and the radar. Dynamic information can be either north-facing or boat-oriented. Printed nautical charts display static information and are always oriented northwards, while electronic charts with radar and AIS information usually show the course of the boat at the top of the screen and include both types of information. This difference between orientations presupposes methods of perception, which the navigator must realise in order to put the information in the field of view of the bridge, to make a decision and act based on the course and speed of the vessel (Morgere et al., Reference Morgere, Diguet and Laurent2014).

The first navigation systems using some form of electronic chart appeared in the market in the early 1980s. The main feature of these systems, also called electronic chart systems (ECS), was the graphical display of the position provided by an electronic positioning system on a computer screen combined with the simultaneous display of some form of nautical chart of the area. With this system, the mariner had constant visual control of the vessels’ position. The operation of the ECS is supported by a database, which consists of the cartographic objects and nautical information that can be displayed. ENCs have significant advantages and capabilities over printed charts, such as:

• • logical processing of cartographic objects to provide alerts in case of danger,

• • change the scale of the displayed area,

• • selective display of chart elements,

• • display of additional information (e.g. sailing directions).

The first commercial digital charts were produced without official specifications and did not contain all the details of nautical charts issued by the hydrographic offices of the various states. Although ECSs made it simpler for seafarers to plan and execute a journey, they were not exempt from conventional voyage procedures, owing to lack of cartographic data. Over time, the need to establish electronic charts that would meet all the required specifications and standards for a safe voyage became apparent and in November 1995 after a thorough examination of the issue by the International Maritime Organisation (IMO), the ENCs were recognised as an official navigation aid and a replacement of paper charts (IMO, 2006).

2.2 Nautical charts compilation challenges

The primary function of a nautical chart is to aid navigation, and the visual aspect may be secondary, as opposed to topographic maps. Another important difference is accuracy. Most of the manmade objects on land can be measured in detail and verified relatively easily. Conversely, sea is a highly dynamic environment that is constantly changing. Reference points are more difficult to identify, as the depths and the shape of a coastline can change with the variation of tide. Underwater features cannot be easily examined and remote sensing is often the only way to detect them in shallow-depth areas.

Remote sensing techniques, in turn, are subject to measurement errors, as the medium through which the signal travels is not uniform, since salinity, temperature and density are constantly changing (NOAA, 2020). Shipwrecks and underwater dunes can be displaced by strong currents, and floating aids with strong weather conditions can rotate around their anchor. It is difficult to create a static map that takes into account these dynamics. At the same time, the navigatorhas no choice but to trust the information on the chart, as most of the depicted objects are under water and beyond visual verification.

The ENC is not just a digital version of a printed chart it introduces a new navigation culture with features that differ greatly from those of the printed charts era. The ENC loaded to the Electronic Chart Display and Information System (ECDIS) has become the legal equivalent of the printed chart as approved by the IMO, and the requirements for more specific information have led to the publication of a new generation of cartographic products (IMO, 2006). In addition, bathymetry charts created from processed hydrographic data or from multi-beam data enable the portrayal of the seabed with blue colour shades and depth contours (IHO, 2019). Similarly, lateral-scan sonar mosaic charts published in the form of diagrams or atlases depict large geomorphological structures. These charts are no longer merely about safety of navigation, as they also enable the knowledge of the environment required for underwater navigation and oceanographic research, as well as industrial uses such as cable laying, hydrocarbon mining and subsoil exploitation (IHO, 2011).

2.3 Hydrographic standards

The progress of nautical cartography, with the wide production of ENCs, gave a great impetus to the development of e-navigation. IMO recommends that ENCs be used on all merchant vessels. Specifically, the Marine Life Safety Convention (SOLAS Chapter V Regulation 19/2.1.4) states: ‘All ships, irrespective of size, shall have nautical charts and nautical publications to plan and display the ship's route for the intended voyage and to plot and monitor positions throughout the voyage.’

The Universal Hydrographic Data Model S-100 is the latest generation of marine geographic information data modelling standard developed by the IHO. The S-100 addresses the limitations of the S-57 standard (IHO, 2000; Alexander et al., Reference Alexander, Brown, Brown and Pharaoh2007) and provides a unified theoretical and application framework for modelling, coding, visualisation, exchange of nautical geographic information, and production and distribution of data products and services. The S-100 version 1.0.0 was released in January 2010, aiming to support a much wider range of hydrographic data sources, products and customers. It is compliant with international standards, especially the ISO 19100 series of geographical information standards thus allowing for the easier integration of hydrographic data and services in geospatial applications (Lovrinčević and Kljajić, Reference Lovrinčević and Kljajić2014). Although S-100 has been updated four times since then, there are still shortcomings that must be improved. Specifications for S-100-based data products and services are still under development (IHO, 2020d).

Until 2015, IHO working groups and researchers concentrated primarily on the theoretical development of the S-100, including essential ideas such as its composition, comparison with the existing S-57 standard and the introduction of new product specifications. Since then, emphasis has been focused on S-100 applied research, such as the creation of software tools, new products and navigation services, data visualisation and correlation with the IMO's e-navigation and maritime services (Weintrit and Zalewski, Reference Weintrit and Zalewski2020). This signifies that the S-100 is steadily progressing from theory to practice and from conceptual consideration and design to execution, optimisation and innovation (Duan et al., Reference Duan, Wan and Luo2021).

2.4 Depth data uncertainty

According to S-44, the IHO standard for hydrographic surveys, the navigation of surface vessels requires accurate knowledge of depths. Where underkeel clearance is a concern, bathymetric coverage must be 100 percent, feature detection must be sufficient and depth uncertainties must be tightly monitored and understood (IHO, 2020b). It is well known that measurements have a degree of uncertainty caused mainly by two factors, the limitations of the measuring instrument -usually referred as systematic error- and the skills of the hydrographers making the measurements, referred as random error (JCGM, 2008). The IHO Working Group on Crowdsourced Bathymetry (CSBWG) has developed a publication, which defines the concept, equipment, methodology, format and uncertainty of data collection, processing and storage (Pavic et al., Reference Pavic, MiŁkovic, Kasum and Alujevic2020).

According to CSBWG, metadata provide information that helps data users to determine the quality of the bathymetry data related to depth and position information (IHO, 2020a). The Quality of Bathymetric Data in S-101, or the CATZOC (CATegory Zone Of Confidence) in S-57, is an IHO classification of the level of accuracy that aims to provide qualitative indications on the uncertainties attached to the bathymetric information underlying the ENCs. The primary intention is for the users to assess how confident one should be with respect to the representation of obstacles to navigation on nautical charts. In that regard, HOs mainly rely on elements of uncertainty on the vertical and horizontal positions of the soundings, the sampling density and temporal variation of the seafloor that may have occurred since the data acquisition, or the information captured as metadata during the hydrographic surveys (EMODnet, 2021; Kastrisios and Ware, Reference Kastrisios and Ware2021).

2.5 Research methodology

This research approach stems from the authors’ long-term involvement with subjects related to nautical cartography, particularly the standardisation of data products and services. The main driver is to study and document whether open reliable sources and open software can be considered as an applicable methodology for the composition of web-based nautical charts in compliance with the S-101 IHO product specification. During the first phase of the research, a wide variety of sources were reviewed, including open-source software repositories, for example open-source software from GitHub, IHO hydrographic standards and specifications, OGC geospatial data processing standards as well as related academic literature. Given the breadth of the research subject, it is important to note that the research work should not be considered as a niche scientific experiment by applying a proposed methodology that yields specific results, but rather as a research that addresses a number of interdependent topics ranging from hydrographic processes to state of the art cartographic applications.

Based on these factors, the research is divided into the following stages:

• • detailed description of the research topic,

• • review of research literature (primarily applicable IHO Standards);

• • identification of important committees and working groups of the relevant international organisations (IHO, IMO, OGC, W3C) and the documentation adopted;

• • observation of activities and involvement as an observer in relevant working group meetings, mostly those of IHO, such as TSMAD / S-100WG, S-101PT, DQWG, NIPWG, MSDIWG, WENDWG, and those of OGC, such as the API SWGs and the Marine DWG;

• • extensive review of the reports of the aforementioned working groups and related committees (such as the IHO's HSSC);

• • review of relevant open source software projects and libraries, for example OpenCPN, SMAC-M (mapserver ENC-based wms), display ENC in QGIS and other,

• • experimentation with open source technologies, for example GeoPackage, gdal, mapbox gl, and Tegola;

• • design of OpenNCS system components and chart compilation processes;

• • development of system testbeds for charts visualisation, process enhancements and experimentation with various navigation services (route planning, optimal route finding); and

• • addressing interoperability issues and discussing improvement areas for further research and development.

3.1 System overview

Figure 1 depicts a high-level architectural design of the logical units of the OpenNCS system developed in the context of the research.

Figure 1. OpenNCS high-level architecture

The system essentially consists of five logical units:

• • the data sources, which include the reference digital seabed model, coastline and marine markings among other objects;

• • the import engine, which includes tools and software code (scripts) for converting data into a common format and importing data into the system database;

• • the main part of the system, which consists of the database and the services for providing the nautical charts to the users;

• • the process automation, that consists of geoprocessing tools and scripts that automate the data import, the compilation process and the system services; and

• • the users part, which includes the devices of the users that can be a web browser on a personal computer or tablet as well as smart mobile devices.

3.2 Compilation process

The compilation process of the web-based nautical charts that is adopted in the OpenNCS consists of four main phases: data acquisition, bathymetry features compilation, chart features packaging and chart visualisation, as shown in Figure 2.

Figure 2. Web-based nautical charts compilation and generation phases

Data acquisition is the process of identifying and automating the import of the open hydrospatial data that will be used for compiling the objects of the chart. The main sources selected are the EMODnet as well as the OpenStreetMap coastline and seamarks. The selection of charted soundings and the generation of depth contours are the tasks that receive the main attention of automation efforts. Both contribute to retaining and displaying the morphological characteristics and differentiating elements of the seabed on the charts (Skopeliti et al., Reference Skopeliti, Tsoulos and Pe'eri2021). Chart features packaging includes the tasks for the preparation of the database layers for the various objects and the symbols library, as well as the lights sectors and labels. For chart visualisation, web-mapping libraries and techniques are used to compile vector tiles, in both static and dynamic modes depending on the chart scale. Relevant test beds were prepared to facilitate development and quality assurance activities. The next sections discuss the four process steps in detail.

4.1 EMODnet bathymetry

The EMODnet bathymetry portal provides open and free access to bathymetry of European seas and is being developed within the scope of the EMODnet, established by the European Commission. EMODnet bathymetry intends to improve access to bathymetric data maintained by a variety of organisations, including hydrographic offices, academic institutions and enterprises. The portal (www.emodnet-bathymetry.eu) provides access to the EMODnet DTM with a grid resolution of 1/16 arc minute × 1/16 arc minute (approximately 115 m width and 60 to 90 m height, depending on the latitude). It contains approximately 12⋅3 billion grid nodes organised in 64 tiles, where 3 tiles where used for the Mediterranean testbed of the research (see relevant section below). The DTM is available for viewing, downloading and sharing through OGC web services (Schmitt et al., Reference Schmitt, Schaap, Spoelstra, Slabon, Wintersteller and Hartmann2020).

The EMODnet DTM has been created using the best available bathymetry from a variety of sources using various methods. The data coverage is still incomplete and some data were collected several years ago, however as new data and other composite DTM datasets are made available, the gaps will be filled and the grid will be updated. The information contained in the grid is believed to be trustworthy. However, its accuracy and completeness cannot be guaranteed. Four indicators have been defined and described in the EMODnet high-resolution seabed mapping (HRSM) document ‘Completing metadata elements for the generation of the quality index for the EMODnet DTM’, known as ‘Quality Indices’:

• • Horizontal accuracy (QI Horizontal)

• • Vertical accuracy (QI Vertical)

• • Purpose of the survey (QI Purpose)

• • Age of the survey (QI Age)

Improvements in the latest EMODnet DTM include the quality index provided at the grid node level (EMODnet, 2021). However, it is clearly mentioned by the EMODnet producers that whilst every effort has been made to ensure its reliability within the limits of present knowledge, no responsibility can be accepted by those involved in its compilation for any consequential loss or damage arising from its use. The EMODnet DTM for European seas is not to be used for navigation or for any other purpose relating to safety at sea.Footnote ³

4.2 The General Bathymetric Chart of the Oceans (GEBCO)

The General Bathymetric Chart of the Oceans (GEBCO) intends to provide the most authoritative, publicly available bathymetry datasets for the world's oceans. GEBCO's current gridded bathymetric dataset, the GEBCO_2021 Grid, is a global terrain model for ocean and land, providing elevation data, in meters, on a 15 arc-second interval grid, a resolution that is useful for the compilation of charts for sea areas where bathymetric data without the accuracy required are not available. It is complemented by a type identifier (TID) grid which provides information on the types of source data that the grid is based. The GEBCO_2021 Grid has been developed through the Nippon Foundation-GEBCO Seabed 2030 Project. It is a collaborative project with the Nippon Foundation of Japan and intends to compile all available bathymetric data in order to create the definitive map of the entire ocean bottom by 2030 and make it available to everybody (Mayer et al., Reference Mayer, Jakobsson, Allen, Dorschel, Falconer, Ferrini, Lamarche, Snaith and Weatherall2018; GEBCO Bathymetric Compilation Group, 2021). GEBCO is essentially a deep-ocean product and does not include detailed bathymetry for shallow waters. Even to the present day, most areas of the world's oceans have not been fully surveyed and, for the most part, bathymetric mapping is an interpretation based on random tracklines of data from many different sources. The quality and coverage of data from these sources is highly variable and similarly to EMODnet's disclaimer; the GEBCO's data sets are not to be used for navigation or for any purpose relating to safety at sea.Footnote ⁴

4.3 Voluntary geographical information (OpenStreetMap)

The beginning of the 21st century brought the rise of crowdsourced geographic information-collection projects. A number of major technological developments, such as online maps, the proliferation of mobile positioning devices and open access to satellite imagery, allowed citizens to become key players in production and exchange of geospatial information (Foody et al., Reference Foody, See, Fritz, Mooney, Olteanu-Raimond, Fonte and Antoniou2017). This process is known as voluntary geographic information (VGI) (Goodchild, Reference Goodchild2007), although a number of similar terms have been coined in the literature (See et al., Reference See, Mooney, Foody, Bastin, Comber, Estima, Fritz, Kerle, Jiang, Laakso, Liu, Milčinski, Nikšič, Painho, Pődör, Olteanu-Raimond and Rutzinger2016). The number of such initiatives is constantly increasing thanks to emerging technologies, and covers a large number of applications, disciplines, communities and tools with a huge volume of literature already produced. VGI has become an important source of geoinformation and describes collaborative and voluntary spatial data collection (Goetz and Zipf, Reference Goetz and Zipf2013). Users can join voluntary communities and share their data with other members of the community free of charge. The data are based on personal measurements or personal knowledge, as well as processing from available aerial images that can be made available freely by various entities (e.g. European Copernicus Sentinel). Initially, data collection was about 2D data, but lately more and more users are contributing 3D data, which include height and depth information. Using such data, one can create digital terrain and seabed models.

Among such initiatives, the most popular infrastructure is OpenStreetMap (OSM), a vector database of multiple geospatial sources with global coverage and available under the Open Database (ODbL) License (Mooney and Minghini, Reference Mooney and Minghini2017). OSM has attracted data contributions from millions of users thanks to the simplicity of its data model, which is based on just three types of objects accompanied by a list of any number of key-value pairs. This makes it extremely easy to search, access and use OSM data using common geographical information system (GIS) software as well as to create applications. There is significant research on the quality of VGIs, and in particular OSM (Senaratne et al., Reference Senaratne, Mobasheri, Ali, Capineri and Haklay2017), which confirms that despite the nonprofessional nature of the sources and the possible lack of quality metadata, VGI data can be used as a valid alternative, or complement government and private data sources. An OSM tag schema called SeamarkFootnote ⁵ is based on elements defined in IHO's S-101 feature catalogue. Seamark tags have risen significantly during the last decade, with more than 500 K tags recorded as of January 2022. Moreover, important seamarks for the safety of navigation such as rocks, are recorded with very high rate, as depicted in Figure 3.

Figure 3. Chronology of rock seamark type. Source OSM taginfo service

5.1 Sounding selection

Water depths (soundings) are numerically displayed on nautical charts and sounding selection is ‘the process of choosing individual water depths from a hydrographic survey in the compilation of a chart’ (IHO, 1994). For OpenNCS, a well-defined methodology, as proposed by Skopeliti et al. (Reference Skopeliti, Stamou, Tsoulos and Pe'eri2020), is used for soundings selection. Based on this methodology, a rhomboidal fishnet is used as a reference structure (Figure 4), where the cell size depends on the compilation scale.

Figure 4. Results from sounding selection methodology

The input dataset is a high-resolution DTM, and the output is a dataset with soundings appropriate for portrayal on the various ENC scales. The sounding selection process is implemented using python geoprocessing libraries. The generalisation procedure adheres to the ‘ladder principle’, which is important for the depiction of soundings on nautical charts, that is every sounding displayed on a smaller-scale chart should also be shown on larger-scale charts. The method successfully produced generalised soundings for Overview, General, Coastal & Approach ENC scales, as shown in Figure 4.

5.2 Depth contours generation

Depth contours generation, according to S-32 (IHO, 1994) is ‘the process of establishing lines representing equal values of a quantity on a map or chart’. Depth contours generation has been identified as the most challenging task (Manzano, Reference Manzano2021). Another challenge is the generalisation and smoothing of depth contours (e.g. removing or merging small rings, fixing contours intersections, or modifications according to soundings out of depth contours) to achieve an outcome that meets the safety and morphology constraints stated by IHO S-4 (IHO, 2021). Depth contours generation is performed using Geospatial Data Abstraction Library (GDAL). The gdal_contour program generates vector contour file (Figure 5) from the input raster terrain model (DTM). Depth contours generalisation is performed using an Open-Source Geographic Information System (QGIS) plugin called Geo Simplification that contains a processing script named Reduce Bend, a geospatial simplification and generalisation tool for lines and polygons.

Figure 5. Depth contours generation process in QGIS. (a) Contours as produced through DTM processing. (b) Elimination of very small rings. (c) Line simplification

Reduce Bend is an improved implementation of the Wang and Müller (Reference Wang and Müller1998) algorithm, often known as ‘Bend Simplify’Footnote ⁶ or ‘Wang Algorithm’. It can simplify lines but also polygons, while preserving topology within and between features of the same layer. The algorithm for each line analyses its curves and decides which one must be simplified, trying to emulate what a cartographer would do manually to simplify a line. Simplification is a line generalisation operator for removing vertices along a line while trying to keep the maximum number of details within the line at the level of the screen's scale. Reduce Bend falls more in the category of line generalisation tools, the process of removing meaningless (unwanted) details of a line usually for scaling down, where the well-known Douglas and Peucker (Reference Douglas and Peucker1973) algorithm is another good example of line simplification tool. Both algorithms can be complementary, since Reduce Bend does not remove unnecessary vertices in the case of very high densities of vertices and it is good practice to use Simplify before Reduce Bend in the case of very dense geometries.

6.1 OSM seamark tags

The OSM Seamark tagsFootnote ⁷ are based on elements defined in the S-57 & S-101 feature catalogues. For human readability, the six-letter mnemonics used in the S-57 standard have been replaced in S-101 by words or underscore-separated English phrases. The format of seamark tags is seamark: <object>:<attribute>=<value> where <object> is the object key, <attribute> is the attribute key and <value> is the attribute value, where the seamark:type= <object> is the parent object. Any given node or way can be tagged with more than one S-57/S-101 objects – for example, a buoy with a light and top mark is tagged with three objects and their attributes. These objects are structured as a parent–child relationship (hierarchical model) with one parent object and zero or more child objects. The parent object is indicated with a tag in this scheme form, with a value corresponding to a valid object key. An example of the tagging scheme – a buoy with top mark and light – is presented in Table 2.

Table 2. Seamark object example

6.2 Lights labels and sectors

In the case of lights, there are two possible objects (light_major & light_minor) in OSM, both of them are treated similarly, differentiated only by the range of visibility across the ENC scales. Special scripts were developed to prepare the light labels for the characteristics of their circle of visibility as well as the curves for those lights that have sectors. The former circle of visibility is defined (IHO, 1994) as the ‘circle surrounding an aid to navigation and in which the aid is visible’ while the latter as ‘a sector of the circle of visibility of a navigational light in which a specific color light is exhibited’. Such sectors are designated by their limiting bearings as observed at some point other than the light. Red sectors are often located so that they warn of danger to vessels’.

6.3 Portrayal

Along with the S-57 (the S-101 predecessor), IHO developed the S-52 standard for presenting ENC content on an Electronic Chart Display and Information System (ECDIS) used for navigation that complies with IMO regulations. The purpose of S-52 is to contribute to the secure operation of ECDIS systems by providing a base as well as complementary display levels for ENC data, symbol patterns, colours and their standard correlation with objects and their features, the appropriate compatibility with printed chart symbols as standard on IHO specifications, ensuring that the screen is clear and there is no uncertainty about the meaning of colours and symbols on the screen. In addition, the introduction of an acceptable pattern for the presentation on ECDIS becomes familiar to mariners and can be recognised immediately without creating confusion. The IHO Geospatial Information Registry includes the so-called portrayal register, which contains chart symbols including their SVG and XML origin. The elements of the portrayal catalogues relate the elements of the feature catalogues to their graphical representation. Figure 7 depicts the S-101 symbols file that was compiled and used in OpenNCS.

Figure 7. S-101 portrayal symbols used in OpenNCS

7.1 Web vector-based cartography

Recent advancements in web map applications have had a significant impact on how cartographic backgrounds are presented. Raster tiles are currently considered a standard method, whereas vector tiles are gaining popularity. Raster tiles are based on the production of an original data set that incorporates custom symbols and style, and all tiles are made in accordance with a standard scheme (Netek et al., Reference Netek, Masopust, Pavlicek and Pechanec2020). With vector tiles, the principle of dividing the maps/charts into squares, is still used, but instead of images vector objects are loaded. Only vector geometry is stored on the server, while symbolisation, rendering and zoom levels are performed by the client device. This method simplifies the change of symbols or topology, as the objects on a map/chart can be manipulated using libraries and scripts of the JavaScript programming language, their symbols can be changed and more features can be portrayed. It is important to note that when developing internet maps, the capabilities and limitations of the online environment must be considered. OpenNCS operates on the principles of internet cartography applications and has the capability to display beyond the six scales for navigational purposes provided in S-101, with smaller and larger zoom levels (see Table 3).

Table 3. Mapping of zoom levels, chart scales and navigational purposes

7.2 ENC & open data testbeds

Two OpenNCS testbeds were set up to facilitate the chart visualisation phase of the web nautical charts’ generation process. The first testbed displays vector tiles compiled from NOAA's ENC data while the second displays vector tiles compiled from open hydrospatial data sources. Figure 8 depicts the two OpenNCS test-beds tested, with the NOAA (ENC) data shown on the left, and with EMODnet & OSM data shown on the right. This figure provides evidence that the outcome of the methodology described gives similar results for both free ENC and open hydrospatial data. Each testbed includes tools to zoom, select scales and choose individual objects from vertical menus on the left side of the screen, for testing and troubleshooting the inclusion of objects, as shown in Figure 9.

Figure 8. OpenNCS testbeds with NOAA (ENC) data and EMODnet & openStreetMap (OSM) data

Figure 9. OpenNCS visualisation testbed