Section 12 Charting and Navigation Cartographie Marine Et Navigation

Contents / Matières Session / Séance Authors / Auteurs 12 Index

Section 12

Charting and Navigation Cartographie marine et navigation

Electronic Navigational Charting around Australia: Finally, we get to the cartographic production bit! Ronald A. Furness

The Cartographic Generalization of Soundings on Chart by Artificial Neural Network Techniques Jia Yao Wang, Zhen Tian

Performance Measurement of Combined Versus Separate Radar and Electronic Chart Displays Don C. Donderi, Sharon McFadden

The Challenges of Production of ENC Cells and Paper Charts from one Common Database Tiina Tuurnala, Ismo Laitakari

An expert system approach for the design and composition of nautical charts Lysandros Tsoulos, Konstantinos Stefanakis

Croatian State Boundary at the Adriatic Sea Ivka Tunjic, Miljenko Lapaine

Production of Thematic Nautical Charts and Handbooks for the Sea Area of the Eastern Adriatic Coast Slavko Horvat, Zeljko Zeleznjak, Tea Duplancic

The use of global mathematical models in the cartography of marine sandbanks Tom Vande Wiele

Making practical and effective electronic aeronautical charts Sonia Rivest, Rupert Brooks, Bob Johnson

Charting and Navigation / Cartographie marine et navigation Contents / Matières Session / Séance Authors / Auteurs 12 Index

Environmental Mapping of Russia’s Seas Using GIS I. Suetova and L. Ushakova

Airborne remote sensing for water quality mapping on the coastal zone of Abruzzo (Italy) Claudio Conese, Marco Benvenuti, Paola Grande

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Session / Séance 02-B Electronic Navigational Charting around Australia: Finally, we get to the cartographic production bit!

Ronald A. Furness Australian Hydrographic Office 8 Station Street, WOLLONGONG 2500, Australia [email protected]

Abstract The paper (and presentation) will address the cartographic challenges presently being grappled with and met by the Australian Hydrographic Office’s (AHO) cartographers as they work to provide mariners with the first authorised and thus, government backed, electronic navigational chart (ENC) database of the inner shipping route through Australia’s Great Barrier Reef. The title reflects a certain personal level of frustration with the time it has taken the world’s hydrographic community to get down to the process of compiling official ENCs. It will place the Australian experiences in context and discuss some of the cartographic production issues that have arisen.

Background

The last decade or so has been a thrilling period in the development of nautical charting around the globe. This truly international activity has seen the development of electronic charts to the point where they will soon, routinely, deliver a range of capabilities suited for use on a variety of ingenious programmed devices and applications. Electronic charts promise both improvements in productivity and increased safety margins to the navigation task of vessels at sea. However, progress has not been without its challenges! Government agencies and private commercial concerns around the world have cooperated, competed, fought, argued, created and generally excelled to bring electronic charting to fruition. The 1990s have been a decade of concerted activity – some of it cooperative and some competitive. This has resulted in international standards for electronic charting which promise to deliver to mariners authorised electronic charts of the highest standard. This is supported by the delivery to the market of complementary systems which provide dynamic and real-time navigational capability to the ship’s master, navigator, regulator, safety authority and ship’s owners and crews simultaneously. Meanwhile, many commercial companies have pushed ahead to market, often brilliantly designed, navigational applications for viewing and integrating electronic charts with onboard ship sensors, such as the Global Positioning System (GPS). The result is that the market will be soon awash with products of various detail, quality, capability and theme. The database design for official electronic charts has been nothing short of brilliant in this author’s view – and cartographers have been to the fore on the decade long, tortuous development road. The design of object- oriented models of reality – sheer genius! Cartographic communication theory has withstood the acid test in perception laboratories as screen-based presentation and symbolisation issues relating to the presentation of charting graphics have been all but resolved. Yet we marine cartographers are only just in the position to start to be able to grapple with production cartography issues. “That’s tomorrow’s problem!” we have been saying, but tomorrow has already passed.

Charting and Navigation / Cartographie marine et navigation 12 Index

The promises of increases in safety margins are so attractive and necessary that the market for electronic charts and systems is increasing apace. Government agencies and commercial organisations are now tussling to define their roles and their positions in a market hungry for electronic chart technology, yet some are often oblivious or unwilling to acknowledge any potential shortcomings. Realisation of the potential increases in safety margins demands the highest quality and integrity of data in the electronic digital chart. In the view of this author, they will only be obtained through reappraisal of the basic tenets of cartographic professionalism. Compilation of electronic charts must ultimately rest on a full cartographic assessment of the best fundamental data available. A simple transference of existing, single product focussed cartography into the object oriented data models which underpin electronic navigational charts is inadequate, notwithstanding the fact that the exigencies of business dictate “acceptable” short-term alterna- tives, such as scanned or vectorised paper charts. The whole process is presently testing the professional skills of the cartographers of the Australian Hydrographic Office as it orients its future chart production capability towards delivery of ENCs to the mariner.

The Australian Context

Consider if you will a relatively remote continent, approximately the size of the United States and with a continental shelf of similar area. This continent, Australia, has a population which is approximately 18 millions. In order for Australia to meet its international charting obligations it needs to be clever and to exploit technology to the hilt. Those technologies about to be deliv- ered to its new state-of-the-art, purpose built, hydrographic survey ships and Laser Airborne Depth Sounder have been world leaders and still Australia struggles, like many other major hydrographic nations, to meet its obligations to Diagram 1 modernise its nautical charts and deliver ENC. Diagram 1 indicates that much of Australia’s maritime area is not adequately surveyed to modern standards. The dark areas are considered adequate. The faintly shaded area represents the continental shelf. A brief perusal of this diagram will alert the reader to the fact that, for the foreseeable future, the Australian marine cartographer will be forced to meld together disparate data sets and this will impact on methodologies for the immediate future. It is clearly some time off before Australia will have a complete ENC based totally on full bottom coverage digital data.

A Brief History of Charting in Australia

The Australian Hydrographic Office (AHO) has been a traditional charting agency until relatively recent times, delivering to mariners authorised paper nautical charts. In the mid-1980s Australia embarked on a program of modernisation which continues today as it transits from a single product deliverer (paper charts) towards becoming a modern hydrographic information supplier. The earlier computer aided techniques used in the AHO were limited and limiting in that they sought to assist chart production by replicating paper chart production methodologies. The main limiting factor in the 1970s and 1980s was technology of course; screens,

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index plotters, digital data collection, positioning; but nevertheless, most of the main ideas were taking shape at that time for developing ultimate ENCs. The AHO converted all of its charts to a raster scanned medium and routinely delivers charts in this form to mariners within a product regime that has mastered the updating requirements of the charts. The Australian Seafarer® service was introduced in 1997. Seafarer® is a fully electronic chart service providing digital repro- ductions of the official Australian paper charts in a raster format. It is suitable for use in a wide range of maritime applications, from fully integrated bridge systems to stand-alone PC based Electronic Charting Sys- tems (ENC). An integral part of the product is an update service, which provides the latest new editions and Notices to Mariners updates on a CD that automatically updates the charts. The product has received a number of prestigious awards. However, the market clearly wants electronic navigation charts of vector format for use in the now available ECDIS systems. The main limiting factor on all hydrographic offices, but especially the AHO, is the lack of available digital data. The AHO has embarked on a major capital investment which will see its data in digital form within about three years (known as Project SEA 1430) but in the interim must seek to find ways of bringing nationally authorised ENCs to the market sooner than that. Since 1998 it has been clear that, for the foreseeable future, ECDIS systems will be multi fuelled. ECDIS is the acronym that means Electronic Chart Display and Information System. The authorised national Electronic Navigational Chart is that chart which is specified to “fuel” the data requirements of ECDIS. The term “multi fuelling” refers to the use in an ECDIS of various authorised charts in different formats. An ECDIS capable of utilising, for example, authorised ENCs and authorised raster charts as well as other publications could be said to be multi-fuelled. Many countries and private companies, in an effort to deliver ENC products to mariners and having regards to the paucity of available digital data, have looked to digitising their charts in a way that vectorises them and which meet the format specifications of the International Hydrographic Organisation for ENC (S 52, S57). Such an approach is useful in the shorter term and has been adopted in Australia. Willis [1998] has pointed out that the reasons such data sets of and around major ports were produced in Australia include: ¨ ECDIS testing and demonstration purposes; ¨ development of standard operating procedures for digital data capture; ¨ proof of concept for conversions from Autochart digital files (Autochart and ChartStation are applications used by the Australian Hydrographic Office under licence from Hydrographic Sciences Australia Pty. Ltd.); ¨ proof of concept for conversions from S-57 to other formats such as the Digital Nautical Chart (electronic chart to military specification), raster and paper chart; ¨ support of requirements raised within the Department of Defence and by maritime and port authorities in Australia, and ¨ training of cartographic staff on the newer generation ChartStation application.

Geographical considerations for an Australian ENC in the Great Barrier Reef

Most readers will have heard of the Great Barrier Reef (GBR). Many of you will dream of visiting the GBR and chances are, if you get to see it you will do so by boat or ship. The GBR is the most navigationally complex region for Australia. The beauty of the Reef is its potential downfall since, although it is hazardous to shipping, shipping is more hazardous to it! The main navigational passage through the GBR is known as The Inner Route. Diagram 2 illustrates the location of this route with some of the main routes through the Reef to the outer routes and deeper water. The Reef is a magnet for researchers and tourists. Being so environmentally sensitive it attracts world attention and is managed appropriately. The International Maritime Organisation has

Charting and Navigation / Cartographie marine et navigation 12 Index declared the entire GBR a Particularly Sen- sitive Area. The GBR receives special attention from everybody and so also it is the case with the national charting authority, the AHO. Compulsory pilotage was introduced in Oc- tober 1991 for ships transiting the GBR In- ner Route and Hydrographers Passage and applies to all vessels of 70 metres or more in length as well as all loaded oil tankers, chemical tankers and gas carriers of any length. Willis [1998] has stated that in response to international calls for hydrographic offices to expedite the release of official ENC data sets compliant with S-57 Edition 3.0, and national calls to improve the state of electronic charting in the Inner Route of the Great Barrier Reef, the AHO undertook to produce ENCs of the Torres Strait and Inner Route. At the time of writing (early March 1999) the first cells from Weipa in the Gulf of Carpentaria to the western approaches to Torres Strait are ready for trialling and demonstration. It is anticipated that they will be so used by AHO staff and ships’ crew on a large commercial vessel, River Boyne, from March 1999. Pro- duction effort progresses cells through Torres Strait and down to the Queensland port of Diagram 2 Gladstone. This undertaking has imposed a number of significant challenges and questions that needed robust responses [Willis, 1998]: ¨ There was a need to develop a comprehensive data capture specification for the ENC. ¨ Was vectorisation of the existing paper chart the best way to present an ENC? ¨ Could we create an ENC of a long narrow strip of waterway, such as one might capture a river, without converting all the paper charts covering that strip? The Inner Route of the GBR is about 1000 kms long by an average 10 – 20 kms wide. ¨ What compromise could be made in limiting the content outside the Inner Route? ¨ Should more information be included within the Inner Route from source data? The specifications adopted by Australia within the IHO guidelines for the Inner Route ENC are: ¨ Depth contours at 1 metre intervals in critical areas derived from original source data ¨ Soundings initially only in critical areas ¨ Smaller scale charts to provide a backdrop ¨ High resolution in critical areas ¨ Coverage of the Inner Route strip ¨ Adherence to IHO Specifications and Standard Display requirements

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Experiences

Paper charts The processes for the production of paper charts are well known but it serves my purposes to recapitulate some main aspects of those processes. Hydrographic surveys are carried out to measure the actuality of depth over time. As a result, the chart making cartographer is faced with the task of melding multiple, disparate data sets of differing age, fidelity, accuracy and quality. These hydrographic data sets have to be interpreted at every stage (having already been interpreted by surveyors and instruments in the first place) and again melded with other disparate data sets: topography, navigation aids, Notices to Mariners, textual data, boundaries information, pilot information, hydrographic notes from all manner of sources and politico-geographical information of the widest sense. The task facing the chart-compiling cartographer then, when faced with a blank sheet and a keen mind, is to abstract, extract, interpret, generalise and construct ultimately a cartographic representation of earthly reality. This representation follows specification yet seeks to serve every conceivable mariner who might seek to go down to that particular part of the sea in a boat or ship. The cartographer’s painstakingly created palimpsest is eventually affixed in a form that bears printed reproduction, or these days, computer reproduction and becomes a document that must be maintained forever that it exists as an authorised document. The marine cartographer has been working as an interpreter, as a combina- tive agent. The result of her labours a cartographic representation of reality: a thematically biased chart fixed by its very medium. Fixed in its presentation and fixed in form and use. Why have I laboured this point? I am endeavouring to set up in the reader’s mind the paradigm in which, until recently, marine cartographers working in the area of nautical charts have operated. I will now attempt to explore the issues raised by the new paradigm for ENC production before drawing my conclusion.

Electronic Navigation Charts The AHO experiences I have outlined earlier describe what might be seen as a transitioning from traditional cartographic processes to a more modern process where the cartographer is returning to the original data and is attempting to interpret the data in a way that suits a more objectified presentation. At the same time, the imperatives of technology are impacting on the ability of those in the field to collect more and more new and precise data. Laser Airborne depth sounders, digital swathe devices and data management systems are already pushing the limits of the most powerful graphics machines and smart graphics seem the only way humans can impose some quality processes on the mass of data collected. So in a sense, the interpretative skills of the hydrographic surveyor have changed. Some might say they have been eroded as the data collection process becomes more objectified. There’s some merit in this argument and it might be inevitable, but this author disagrees that complete objective data is presently deliverable to the chart making process. Hydrographic Offices, including the AHO in particular, have been aware for some time of the need for them to position themselves to be able to generate ENC, raster and paper charts products from the single moderated source data set. They are painfully aware that the data set must be maintained, as must be the products. So marine cartographers must focus on new ways to present chart information. Paper products have tolerated regional differences in presentation. The dream of a single worldwide ENC coverage and the rigid demands of computer systems required to run these ENC databases mean that myriads of minor differences are intolerable. No longer is the chart compiler able to have some determining role as to how some “feature” might be depicted, or even how far it might be displaced in position for aesthetic cartographic presentation, but she must be faithful to the “object”, describing it and its relationships precisely. So we are starting to see a shift in cartographic thinking here from the subjective to the objective. Subjectively defined graphic has given way to precisely defined and comprehensively described real world object.

Charting and Navigation / Cartographie marine et navigation 12 Index

But what of theoretical objects such as dredged channels? Clearly I do not mean that dredged channels exist only in theory but that their traditional cartographic presentation as features depicts them in a theoretical way. Sharp, precise channels with absolutely level bottoms. Modern data collecting devices soon disabuse one from the notion that such preciseness exists in the real world. What is traditionally depicted on charts for dredged channels typically is a gazetted limit with a defined (guaranteed) minimum depth and not the precise actual channel. So where does this leave the marine cartographer grappling to define reality more precisely? Well, clearly still with cartographic objects in a real world. But the example serves to lead into the point about the changing view of data and the synergy that modern computers unleash for others such as environmentalists, researchers, legislatures, law enforcers, searchers and rescuers, coastal zone managers and so on. This list is by no means exhaustive but illustrates a largely untapped demand for hydrographers’ data. It also serves to make the point that the marine cartographers’ responsibilities are shifting away from “just” the mariner. As expert interpreters of the hydrographic and navigational data, cartographers are inexorably repositioning themselves and their professional role and obligation to society. If you think I am drawing a long bow then consider the scramble for interpretive advice arising from the requirement for nations to define their various offshore zones! Any cartographer who has worked for a major map or chart-producing agency will be aware of the push for minimising effort for derived mapping. The push continues and is almost there conceptually in the object oriented marine world. However, the mind processes of the cartographic as she defines objects are at once spanning all possible uses of that object for a myriad of possible uses. So the mind is moving here from the particular (THIS scale, THIS use, THIS presentation) to the general (ALL possible scales, ALL possible uses, ALL possible presentations). The mariner, at sea, on a cramped bridge, with terrible visibility, with immense responsibility, unstoppable momentum and almost certain dependency on electronics, is interested ultimately in a fundamental binary equation – GO or NO GO! The mariner is now able to relate precisely and absolutely to the real world in real time. A realisation of this must be uppermost in the mind of the cartographer interpret- ing the data in the comfort of an office, particularly when making “decrees” as to issues such as usage, quality, relative accuracy. This is quite a shift in the thinking for most chart makers and requires that they thus develop new paradigms. The mariner is not able, as we were entreated by Cervantes’ Don Quixote, to “journey all over the universe in a map, without the expense and fatigue of travelling, without suffering the inconveniences of heat, cold, hunger and thirst”. Chart makers must come out from the comfort of their offices and start to experience some of the “heat, cold, hunger and thirst” of the real world as they make their charts. This is starting to happen in Australia. One interesting side issue for cartographers is the increase of ephemeral mapping or charting with a concomi- tant decrease in graphic quality. So what! The world now looks at polygon colours and reports rather than lines and text and presentation of both need to conform with emerging standards if they are going to synergise with what already has emerged rather than reinvent wheels: chart standards, SGML, HTML, GIF, JPEG, internet etc. No longer is it be productive to argue for hours as to whether a piece of text has rounded ends, is exactly 6.0000000 points in size and 0.0001235 mm thick. Modern cartographers have all but dispensed with that mind-set and many an eye glass has found its way to the historical artefact shrines of cartography (which ironically still adorn many an entrance to the most progressive mapping and charting enterprises). No longer can cartographers expect their images to remain extant forever though this author admits to a certain feeling of immortality to discover that the first chart he ever compiled remains the current Admiralty chart for the particular port. Another interesting issue for the cartographers engaged in nautical chart production is the increasing use of data based information such as lights, buoys, boundaries, text, photographs, attribute management, tomes and on. The actual compilation process is requiring compilers to focus more on the actual bathymetry and the construction of bathymetric polygons. When the aim is to include isobaths at metre or sub-metre interval, then

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index the subsequent polygonisation is, to say the least, challenging. Overkill? Hardly when increasingly mariners are demanding greater and greater detail and accuracy in order to rationalise their loads to minimum underkeel clearances in critically sensitive areas such as the GBR. The challenges for quality assurance and control are immense and are being faced as opportunities by the day! I have already alluded to the expanding horizons for chart data interpreters and, while it might seem trite to do so, I point out that the present IHO specifications have been designed to a large extent from a cartographic perspective. To that extent they are limiting. However, future (decade plus) improvements will ultimately see precise interaction between dynamic data (ship’s draught, tidal models, ice edges, weather, for example) and the databased chart information and tolerances between the GO / NO GO margin will shrink. Thus cartographers need to be far reaching in their present day mind sets. We face all the time the dilemma between the need for certainty in our specifications and processes and the need for innovation and flexibility. Notions of multi-fuelling with various data sources, in itself a rational and transitional option, forces compromise and stifles creativity and innovation. When marine cartographers deliver a paper chart the medium is very much the message. Rigid and generalised in presentation, the presentation could be said to conform to the principle that “one size fits all”! The advent of digital data separates the medium from the message in that the message to the mariner must come via an electronic chart system, or ECDIS. This has spawned a number of Faustian relationships between HOs and manufacturers with no clear lines of demarcation in respective roles. This is clearly part of the development conundrum, but as we progress there are signs that the market place is better realising proper roles for authorised government agencies and the commercial imperatives of system manufacture.

Conclusions

So where does this all leave me given that it’s time to begin to draw my thoughts together? It leaves me, I believe, arguing a case for a recognition that the cartographic paradigm for marine cartography is shifting from a more subjective and perhaps narrower professional role to one that is more general, or wider, and more objective in its professionalism. I believe that this modern role is encapsulated in the notional general description “mapping scientist” of which traditional cartography remains an important part. Such folk will be required to be more eclectic in their thinking and will be able to project forward as they work to the multiple uses their work will be put. I believe that I have shown that such a role is better descriptive of the essential paradigm shift in production processes that cartographers must undertake if they are to remain at the forefront of delivering geographical interpretations of the real world to modern users of such information. So in the final instance we have got to the cartographic production bit but need to recognise that the paradigm has changed already. We need to change as professionals to be in line! As one wit has said, “we are at the bleeding edge”! If that’s the case then the experiences are presently being written in blood, but finally, they are being written and we are rising to meet the challengesS!

References

Willis R J, [1998] Towards “AUS ENC ONE”, Proceedings, Institution of Surveyors 39th Australian Surveyors Con- gress, Launceston.

Charting and Navigation / Cartographie marine et navigation 12 Index

Session / Séance 02-D The Cartographic Generalization of Soundings on Chart by Artificial Neural Network Techniques

Jia Yao Wang Dept. of Cartography & Geoinformatics, Zhengzhou Institute of Surveying & Mapping 66,Mid-Longhai Rd. Zhengzhou, China, 450052 E-mail: [email protected]

Zhen Tian Dept. of Marine Surveying and Mapping, Dalian Naval Academy, Dalian, China, 116018 Tel: +86-411-2678355 E-mail: [email protected]

Abstract As a main part of nautical chart cartographic generalization, the generalization of soundings is also one of the bottlenecks in the way of automatic chart generalization. In this article, the design of neural network generalizing soundings was discussed, along with the analysis on the network’s overall structure, the working methods, the network factors, learning. To solve the problem of dealing with both the spatial and attributive factors at the same time in the selection of soundings, a new set of operations is designed based on a practical problem-solving model called “Hierarchical Information Structure”. The experimental result is shown upon a protocol system.

1 Generalization of Soundings on Nautical Chart

Soundings, sometimes called the notation of water depth, are the numeral symbols showing the depth of seawater from depth datum to the bottom of sea, scattering in the sea area on nautical charts. For both the purposes of making it practical and good-looking, soundings on nautical charts are required to form a net of approximately diamond-shaped, which means that the generalization of soundings deals synchronously with both spatial (the scattering pattern) and attributive(depth, and more) characteristics of the chart features. On current Chinese nautical charts, the seabed terrain is shown mainly by soundings, with the assistance of depth contours and bottom natures. So the correct and reasonable generalization (selection) of soundings is essential to make a good chart, scientific and helpful to the mariners

1.1 The Requirements of Sounding Selection 1. Soundings which are at the top of a raised under-water area (shoals, rocks, etc.) have the highest priority to be selected. Generally select the shallower prior to the deeper ones to guarantee the safety of navigation; 2. Then choose the soundings which show the waterways; 3. Then choose the soundings which represent the outline of sea bottom the best; 4. Choose the other soundings to fulfill the chart; 5. The soundings should spread in such a way that they form a net of rhomboidal shape.

Charting and Navigation / Cartographie marine et navigation 12 Index

1.2 The Difficulties The soundings selected by the first three rules are usually scattered on the chart irregularly. When in the manual generalization, Cartographers have to repeat the selection procedure for several times to attain satisfactory result that best represents the underwater relief while soundings scatter in the required pattern. Not only the knowledge and steps of thinking for this job is highly complex, difficult to explain, thus hard to computerize in ordinary ways, the processing of both depth values of soundings and scattering pattern of them interactively at same time also cause great problems. A possible way to deal with this situation is to examine the inner motion of thinking of human beings to solve similar problems, trying to take into consideration as many factors as possible, and then learn from the natural problem-solving method of manual work.

2 The design of Artificial Neural Network for Sounding Selection

2.1 Basic Idea Hierarchical Information Structure (HIS)[3] is a pattern to solve the problems with partly-known knowledge of it. The idea of problem- solving by HIS can be simply explained as the following. (1) if the knowledge is not enough to solve the whole problem, try to divide the problem into different parts; (2) search for the answer to each part with what we have known now, the answer(s) will add some new knowledge; (3) try to attack more parts of the problem by the new extent of knowledge; (4) loop till all finished. Not very precise, but it does work. From the main idea of HIS and theory of neural network, we design a method of sounding selection. Two significant points distinguish our method from the usual way of sounding selection. First, we divide the whole scattering area of soundings into small sub-area (sounding patch), the selection of soundings is gradually finished, from the selection of sounding patches to the selection of individual soundings. Second, in each selected sounding patch a sounding is left on the chart and the others are omitted. So the selection or deletion will be determined not only upon the sounding itself, but also upon the surrounding terrain features.

2.2 Structure and Strategy Many models and structures of neural network have been proposed, but none suits the generalization problem very well. Based upon the processing ideas above, we designed a neural network for our special purpose. The network is basically a combination of two backpropagation networks (fig. 1). The first part deals with the processing of sounding patches, and the upper one do the final selection Fig.1 Structure of Neural Network for works. Data are put forward layer by layer (serial connection), while Sounding Selection some special info (such like “dangerous sounding”) can yet be passed directly to the upper layer (parallel connection).

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

2.3 Factors and Learning The factors are those items of info about the characteristics of the soundings. The network makes selection decisions by working on the factors. More factors employed, the decision may get more reasonable, but on the other hand the network may become less stable. In the experiment we selected 16 factors for the process of soundings in each patch, and 17 factors for the evaluating of sounding patches. Some of them include: Type (italic, upright, dangerous, etc.); Value of depth; Minimum distance to the fore-selected soundings; Minimum distance to shore and inter-tidal area; Minimum distance to obstacles; Whether surrounded isolatedly by depth contours; Depth difference to the surrounding depth contour; General info of chart; General info of sea area; etc.

Factor set for sounding patch, which is somehow “of area type”, is slightly different to that for individual sounding. For example, those info about the area feature “in patch area”, such like the mini-terrain type (peak, range, valley, basin, or slope, etc.), sounding density, greatest depth difference, etc., are concerned. Weight training of the network is by backpropagation algorithm. It adjust weights by[1]

WW()kk=+⋅ ()−−11α ∆ () W () k where 1 n p ∆(W) = ∇E(W) =−∑ ϕ (,tW )(() d t ytW (, )) n p t =1 which can guarantee a local optimal solution.

Result and Study

The experiment is based upon the discussions above. As the preparing work, three sheets of navigation chart is chosen to set supervised learning samples, which summed up to approximately 500 training areas. Also, some 100 designed training areas are used to complete the types of patch feature. Figure 2 shows a sea area in South China Sea, generalized by the protocol network. Fig.(a) is the source and (b) is the result. Main requirements of sounding selection have been satisfied in the result figure, yet it is hard to say the system is powerful enough. Further research is been planning (including the combination with expert system, knowledge-based system and fuzzy mathematics, etc.) and we welcome cooperation and suggestions.

Charting and Navigation / Cartographie marine et navigation 12 Index

Fig.2 Result of experiment

Reference

1. Fu, L.M. (1994), Neural Networks in Computer Intelligence. McGraw-Hill, Inc., New York. 2. Tian, Z., Wang, J.Y., Liang, K.L. (1997), Design of Neural Network for Automated Selection of Soundings in Nautical Chart Making, Proceeding of 18th ICC, Stockholm. 3. Tian, Z. (1997), Research of the Automatic Cartographic Generalization Based on Artificial Neural Network, Ph.D. thesis, Zhengzhou Institute of Surveying and Mapping, Zhengzhou, China. 4. Ziborov, V.V., etc.(1989), Generalization of Depth in the Modeling of Bottom Relief, Mapping Science and Remote Sensing, New York

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Session / Séance 02-C Performance Measurement of Combined Versus Separate Radar and Electronic Chart Displays

Don C. Donderi Human Factors North Inc and McGill University [email protected]

Sharon McFadden Defence and Civil Institute of Environmental Medicine [email protected]

Abstract There was no published evidence to say whether combining the information available in marine radar with the information available in electronic charts on the same display screen would help or hinder the use of this information in marine navigation. We carried out a laboratory experiment in which information had to be obtained from either a radar plot or from its associated electronic chart, or had to be combined from both the plot and the chart in order to correctly respond to true-false statements like: “There is a ship return to the north of Ownship” (radar question), or “Radar shows a ship return approaching St. Mary’s Island” (radar and chart question), or “There is a large building at the end of Smith’s pier” (chart question). Eighteen inexperienced (college undergraduates, average age 21) and nine experienced (deck officers, average age 35) observers were given written instructions and practice in using marine radar and electronic chart displays. They then answered true-false questions about the radar and chart displays . Half of the questions were presented with the radar and the chart displays on separate adjacent monitors, and half were presented with the radar image overlaid over the electronic chart information on the same display. Observers responded as rapidly as they could consistent with being confident that their answers were correct. The pattern of results was the same for both officers and students. They were equally accurate (71 percent correct overall), and equally accurate on the overlay and the separate displays. Officers were slower to answer than students, but officers and students alike were faster on the overlay displays than on the separate displays. Questions requiring radar information alone were answered fastest; questions requiring chart information were answered next fastest, and questions requiring the integration of chart and radar information were answered slowest, and this was true whether the information was displayed on separate or overlaid displays. A measure of efficiency (percent correct/viewing time) was at a maximum for chart-radar pairs that had an intermediate level of subjectively judged dissimilarity between the radar and the chart displays. We conclude that information from overlay displays is evaluated just as accurately as, and faster than, information from separate radar and chart displays regardless of what kind of information: chart, radar or a combination of both, is needed to evaluate the statements about the displays.

Charting and Navigation / Cartographie marine et navigation 12 Index

Introduction

Marine navigation and piloting procedures and display standards are in flux. From the 1950s to the present, navigation and piloting has been carried out with paper charts, land-based electronic positioning systems like Loran, and celestial fixes, aided when possible by radar ranges and bearings. Courses were planned and plotted on paper charts. The charts were corrected by hand as changes to navigation aids, soundings or topography were collated and distributed to users by the national hydrographic services (Maloney, 1978). An automated display of ships’ position information was added to paper charts when Loran-driven plotting pens were set to mark a ships’ track on paper charts. A similar position display was based on an inertial navigation system that mechanically projected spot of light, locating the ship on a paper chart (Millar & Hansford, 1983). Radar technology also advanced to the point where marine radar systems can identify moving targets, extrapo- late their courses, and predict and display a calculated closest point of approach. However, research with these advanced radar plotting aids (ARPA) shows that large individual differences in navigation styles, encompassing a wide range of risk-taking tolerance, are still found among navigating officers even when using ARPA (Habberly et al, 1984), and “radar-assisted collisions” continued to occur. Until recently no single positioning system was accurate enough to provide the necessary confidence to locate a ships’ position on a large-scale chart. Now, a combination of computer-assisted dead reckoning, Loran, local microwave positioning systems, and differential GPS, (a world-wide satellite network supplemented by ground stations), make it possible to produce a single navigtion display image that locates the ship on an electronic chart with an error measured in meters, and superimposes properly oriented and scaled radar returns over the ship and chart image (Rolfe, 1996, Baziw, 1996). The integrated display image also contains a character-based summary of relevant navigation information including speed and heading, progress along a predetermined track, deviation from the track, and the status of the positioning information used to locate the ship on the track and display. These combined displays are called electronic chart display and information systems (ECDIS). They are currently installed on many Canadian lake carriers and are increasingly being installed on ferries and ocean-going ships. The task of the marine navigator or pilot can be either ameliorated or complicated by the layout of the command and control equipment on the bridge (Schuffel, 1985). Driven by human factors/ergonomics research and analysis, ships’ bridge designs have evolved over the past twenty years to allow the safe and efficient operation of large ships by minimum navigating crews, consisting at times of only one officer (Herman, 1977; Istace, 1977; Hall & Anderson, 1980). Most of the bridge design studies preceded the development of ECDIS displays. Many design and operational questions remain to be answered about the deployment and use of ECDIS displays, including: Where should the display be put? What information should appear on it? In what format should the information appear? How should the informatin display be controlled? How should ECDIS be integrated with other navigation aids? How should navigators be trained to use ECDIS displays? In the absence of any empirical or analytical studies, these questions are answered by applying local expertise or local opinion whenever a decision has to be made. The last word in bridge layout and equipment installation is almost always that of the owner’s repre- sentative, usually a senior captain, who supervises the fitting-out at the dockyard. But the optimum location for, the optimum format for, the optimum use of, the optimum training on, and the optimum integration of ECDIS displays into the piloting and navigation process have been neither systematically analyzed nor empiri- cally investigated, at least in the open literature. The study reported here begins to investigate one capability the marine navigator gains from ECDIS: the ability to superimpose a properly oriented and scaled version of the current radar return onto the electronic chart, with the ship’s position indicated on the chart corresponding to the origin of the radar returns. The radar

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

is overlaid in a “transparent” mode so that radar returns and the chart information beneath the returns are simulataneously visible. The combination of a reliable ECDIS display with reliable superimposed radar might be thought to give the navigator all the information needed. The chart display indicates the location of fixed hazards to navigation, the soundings, the recommended traffic lanes, etc, while the radar displays the transient hazards to navigation including other ships, icebergs, floating debris, etc, as well as the location and identity of radar-responding lights, buoys and ranges. But radar returns are ambiguous and may be incomplete, electronic chart information may be out-of date, the positioning sensors may be damaged, and local aids to navigation like radar beacons, buoys, etc may have just been damaged or destroyed. Experienced navigating officers know that chart information and radar returns are no substitute for continuous human visual verification, both at sea and in pilotage waters. So at the onset, ECDIS displays, even more than ARPA radar displays, offer the navigating officer the tempta- tion of sticking his or her head into the display hood and keeping it there, to the detriment of common sense and sound navigation practice. This hazard, like “radar-assisted collisions” can be avoided through proper training. The fact that a good source of information may be abused does not mean that it must be abused, and the superimposition of radar and ECDIS displays may very well have real advantages that outweigh these potential hazards. The focus of our studies was on the use of radar information and chart information either superimposed on the same display screen, or separately on two adjacent screens. Our studies were carried out in a laboratory as opposed to a marine or marine simulator environment. Therefore there was no potential distraction from, or conflict with, other sources of information about either the ship or the navigating environment. The reason we were interested in radar-chart superimposition is that in many circumstances display complexity is known to have a deleterious effect on both the accuracy and speed with which information can be extracted from visual displays. Superimposing radar information on electronic chart information increases the complexity of the resulting display, by comparison to the two simpler original displays (radar alone and chart alone). On the other hand, it takes less head movement and fewer eye fixations to visually scan observe the smaller area of one display screen than it does to scan the larger, disjunct areas of two screens; therefore there may be a counterevailing advantage in displaying both radar and chart information on the same screen. The relative advantage of one versus two screens may be also be influenced by the complexity of the superimposed displays, so the answer: one screen or two? may very well depend on the particular chart and radar returns displayed. The goal of this study was to determine under laboratory conditions, what differences were observable in extracting useful navigating information from one superimposed chart and radar display, as opposed to two separate, side-by-side displays, one of which displays radar information and the other, the matching chart information. We presented statements about the displays which the participants had to evaluate as either “true” or “false”, as rapidly and accurately as they could. The participant had to read the statement, decide what information was needed to evaluate it, and then look for that information on the display or displays. We carried out the study using university undergraduates who were given written information about marine radar and electronic chart imagery, and who were then allowed to practice, before beginning the experimental tasks. We also recruited participants among experienced marine navigators including Canadian Navy officers and commercial captains and deck officers.

Charting and Navigation / Cartographie marine et navigation 12 Index

Experimental Method

Displays Fifty-two electronic chart images were obtained from a Canadian marine simulator installation. The charts covered coastal areas of Newfoundland and Norway, and Halifax, N.S. and New York harbors. The color rendering of these charts was corrected to IMO standards by the research team. The charts ranged from the subjectively very simple to the subjectively very complex. (These subjective descriptions were quantified in another study, not reported here). Each chart image had an associated radar image. The radar images were generated by a simulator control computer from a position corresponding to the “ownship” symbol on the chart. An overlay image was generated for each chart-radar pair. The separate chart and radar images were carefully superimposed, and then scaled and rotated so that radar landmarks corresponded to the appropriate chart symbols, and the radar coastline matched as closely as possible the chart coastline. Then each superimposed pair was saved for use as a chart-radar overlay display.

Presentation The task required that the participant observe either one or two display screens. The physical arrangement consisted of two identical 17 in SVGA display monitors (AcerVue 76e) placed side-by-side, and a single 101- key computer keyboard placed beneath the right-hand display. The keyboards and displays were placed on a desk at 72cm from the floor, and the particpant sat in front of the displays on a secretarial chair. The experiment was carried out under ambient illumination of 300 lux, measured at the keyboard, provided by overhead fluorescent fixtures which did not produce glare on either display screen.

Experimental Task Statements Six different statements were written for each chart-radar display pair. Two of the statements could be confirmed or disconfirmed by information that could be obtained exclusively from the radar display without reference to the chart display, two depended on information that could be obtained exclusively from the chart dislay without reference to the radar display, and two required that information be combined from the chart and radar displays. Displays The entire display presentation sequence was controlled by computer but was paced by the participant. The sequence began with a message “Press any key to see the next question” in white characters at the top of one of the two blank (black) display screens. The screen displaying the messages was consistent for each participant but randomly varied between participants. When the participant pressed a key, the statement chosen for that trial appeared and a question duration timer started. The participant read the statement and when he or she was ready, pressed any key to present the display and start a display duration timer. The “separate” displays on both screens were always presented with the radar on the left screen and the chart on the right screen, but the overlay displays were presented on either the left screen or the right screen, consistently for each participant but randomly between participants. When the participant had decided whether the statement was true or false, he or she pressed the “T” [rue] or “F” [alse] keys on the keyboard to record the response, stop the timers and erase the statement and the displays. These keys were marked with colored stickers so they were quite con- spicuous. Then the “Press any key to see the next question” message appeared at the top of the blank message screen, in preparation for the next trial. Instructions All participants received written instructions at the beginning of the experiment. For the university undergraduates, the instructions briefly explained the nature of marine radar, the nature of electronic chart

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index display systems, and the purpose of the study (see Appendix 1). A glossary of chart symbols was continuously available during the experiment. The navigating officers received shorter instructions that simply explained the purpose of the experiment. However, each officer was thoroughly debriefed after participating to elicit com- ments about both the experimental study and EDCDIS displays in general. Randomization The fifty-two image sets were randomly assigned for each participant to three separate tasks: true-false practice, true-false test and map-matching (Map-matching results are not discussed in this paper). Ten images were assigned to the true-false practice sessions, thirty-two to the true-false task, and ten to the map-matching task. On the true-false practice and test tasks, there were six display presentations of each image set: one presentation for each of the six statements written for that set. Thus there were a total of sixty (10 x 6) presentations in the true-false practice task and 192 (32 x 6) presentations in the true-false test. The presentations were randomized within participants with respect to which image sets appeared in which part of the study (practice, true-false test or map-matching). They were also randomized with respect to which presentations appeared in either the overlay or separate display conditions. Finally, the order of presentation for the displays accompanying the six questions for each image set was also randomized. In other words, the true-false experiment was executed using a within-subject design in which everything was randomized across the classication variables of display type (separate or overlay) and question type (radar, chart, or radar+chart).

Participants Eighteen McGill University undergraduates participated. Five were men. The average age was 22 and the range was from 20 to 27. Nine experienced navigators participated, all of whom were men. They ranged in age from 25 to 55. Three of them were Canadian Navy officers, ranging in rank from lieutenant-commander to lieutenant. Two were on active duty and one was a reservist. Six were commercial masters or deck officers, ranging in seniority from the captain of a lake carrier to a former second officer now enrolled in a management program at McGill. Seven of the navigators were currently seagoing.

Data Recording and Analysis The data from a practice task that consisted of the first ten samples (sixty trials) of the true-false task, were not analyzed. Data collected from the remaining true-false tasks included the time required to read each statement, the display time for each display, and the answer returned (True or False). This made it possible to assign another value (Right or Wrong) to the answer associated with each question presented with each display. The classification variables were the type of display (separate or overlaid) and the type of question (radar, chart, or radar+chart). Variables associated with each display included the subjectively obtained complexity ratings for the radar, chart and overlay presentations, and the subjectively judged dissimilarity of each radar- chart pair. The general linear model of analysis of variance was used to analyze the output data from the true-false task, because as a result of the randomization of display condition (separate versus overlay), the variations in the distribution of answers (True and False) across displays, and differences in the number of correct responses (Right or Wrong), the number of data entries per analytical cell varied across participants. In all analyses, observers were treated as a random variable.

Charting and Navigation / Cartographie marine et navigation 12 Index

Results

The effects reported here are those that were consistent between officers and students. Some analysis-of- variance interactions were “significant” without being “important’, and those are left for future discussion and elaboration.

Correct Responses Officers and students were equally accurate at the task: officers with a proportion of 0.706 correct responses, and students with a proportion of 0.707. These proportions are not significantly different. Both officers and students were more accurate with statements for which the correct answer was “False” (Table 1).

Table 1. Proportion Correct: Officers and Students on “True” and “False” Statements

Participant Group “True” correct answers “False” correct answers Officers 0.6395 0.7728 Students 0.6568 0.7132

There were consistent difference for both officers and students in the proportion of correct responses across the kind of information (radar, chart, or radar + chart) required to evaluate a statement, and the type of statement (correct “True” or correct “False”) (Table 2). Notice that the officers were consistently more accurate than the students on the “False” statements, while the students were more accurate on the “True” statements.

Table 2. Proportion Correct: Officers and Students on “True” or “False” Chart, Radar or Chart+Radar Statements

Information Radar Chart Radar+Chart Type True False True False True False Officers 0.594 0.781 0.649 0.803 0.676 0.734 Students 0.671 0.764 0.663 0.754 0.702 0.687

Display Duration Officers took longer on the average (15.43 sec) to respond to the statements than did students (12.34 sec). This difference was highly significant. Officers and students took longer to respond when they were wrong (15.87 sec) than when they were right (13.10 sec); again highly significant. And both officers and students took the least time to respond to radar questions, more time to respond to chart questions, and the most time to respond to radar+chart questions (Table 3).

Table 3. Display Duration: Officers and Students on Three Types of Question Participants Radar questions Chart questions Radar+Chart questions Officers 13.88 16.11 16.32 Students 10.89 12.54 13.60

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Efficiency Our measure of efficiency for every condition was the proportion of correct responses divided by the image display duration; or, in other words, accuracy achieved divided by the time taken to achieve it. On this standard, the students, whose responses were much faster, were more efficient (0.067) than the officers (0.057), a highly significant differenct. This measure aksi differed significantly for both display screen (Overlay versus Separate) and Statement Type (Radar, Chart, or Radar+Chart), for both officers and students, as shown in Tables 4 and 5

Table 4. Efficiency Measures for Separate and Overlay Screens Participants Separate Screens Overlay Screen Officers .070 .063 Students .061 .053

Table 5. Efficiency Measures for Radar, Chart and Radar+Chart Questions

Participants Radar questions Chart questions Radar+Chart questions Officers .063 .059 .050 Students .075 .067 .058

Conclusions

There was no performance penalty attached to overlaying radar displays on electronic chart displays. Regard- less of whether the true-false statement that was evaluated concerned the radar display alone, the chart alone, or required a synthesis of information from those two sources to evaluate correctly, the evaluation was carried out as accurately on the overlay as on the separate display, and it was carried out faster on the overlay display by both officers and students. Because the accuracy of the responses to separate and overlay displays was the same, the efficiency of the overlay displays was also higher. As a first approximation, then, we can say — based on a laboratory study under static display conditions — that superimposing a marine radar display over an electronic chart display does not make the information in either display harder to evaluate. There are practical limitations to our conclusions, some of them pointed out in our interesting conversations with the marine navigators who participated. All of these chart and radar images were simulating long-range, small scale charts and radar displays. There was some clutter caused by the Ownship return, but very little of the sea clutter that the mariners told us would likely interfere on the radar display with a short-range setting, and so make the chart information “under” the transparent radar display harder to read. A static display is by nature ambiguous with respect to radar information, because much is learned by observing the changes in the radar display over time. Dynamic displays, and short-range use, might give different answers to the question we studied here. In the meantime, it is clear that one property of ECDIS displays — the ability to superimpose radar and electronic chart imagery — may be a real advantage to the marine navigator.

Charting and Navigation / Cartographie marine et navigation 12 Index

References

Baziw, E. (1996). Field Trials of ECPINS Vessel Positioning Algorithm. IEEE Position, Location and Navigation Sym- posium, Altanta, GA, 121-129. Hall, J. W. & Anderson, M. D. (1980).The U.S. Coast Guard Multi-Mission Cutter: Command, Display and Control (COMDAC). Naval Engineer’s Journal, 92(35), 59-69. Habberley, J. S., Shaddick, C. A. & Taylor, D. H. (1984). A behavioral study of the collision avoidance task in bridge watchkeeping. Report to the Marine Directorate, Department of Transport, U.K. Herrman, R. (1977). Two studies for optimizing operating bridges and their application in inland and sea-navigation. Human Factors in the Design and Operation of Ships: Proceedings of the First International Conference on Human Factors in the Design and Operation of Ships, Gothenberg, Sweden,, 58-68, Istace, H. (1977). An experimental evaluation of a “one-man control” bridge .layout. Human Factors in the Design and Operation of Ships: Proceedings of the First International Conference on Human Factors in the Design and Operation of Ships, Gothenberg, Sweden 167-173. Maloney, E. S. (1978) Duttons’s Navigation and Piloting, 13th edition. Annapolis, Md., Naval Institute Press. Millar, I. C. & Hansford, R. F. (1983). The ‘Manav’ integrated navigation system. Journal of Navigation, 36(1), 81-92. Rolfe, G. A. (1996). Radar image overlay on an ECDIS system - an overview. IEEE Plans, Position Location and Navigation Symposium, Piscatawnay, NJ, 130-136 Schuffel, H. (1985). Determining effects of bridge design on ship control. Society of Naval Architects and Marine Engineers, STAR Symposium, Norfolk, Va, 275-283.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Session / Séance 02-A The Challenges of Production of ENC Cells and Paper Charts from one Common Database

Tiina Tuurnala Finnish Maritime Administration Department of Hydrography and Waterways, Porkkalankatu 5, 00181 Helsinki, Finland Phone: +358 204 48 4426 Fax: +358 204 48 4620 e-mail: [email protected]

Ismo Laitakari Finnish Maritime Administration Department of Hydrography and Waterways, Porkkalankatu 5, 00181 Helsinki, Finland Phone: +358 204 48 4407 Fax: +358 204 48 4620 e-mail: [email protected]

Abstract Previously, the main purpose of nautical chart databases was the production of paper charts. Advances in navigation technology, e.g. satellite positioning systems, have set new demands on data accuracy, reliability and the format of data. The Hydrographic Offices are required to produce more and more accurate charts and especially electronic navigational chart (ENC) data. This rapidly increased need for electronic chart data has led many offices to a situation where there are two separate production lines for two products, ENC cells and paper charts. However, it is vital for the safety of navigation that the content of data is exactly the same in both products and the products are not in conflict with one another. It is also a waste of time to do the same updates twice in two different databases. Research and development in many Hydrographic Offices is now concentrated on the problem how to combine electronic and paper chart production? The Finnish Maritime Administration is developing a new data management and chart production system. The development project is called the HIS (Hydrographic Information System) Project. The project is divided into two phases, the first phase containing development of a data management system and an ENC production line and the second phase development of a new paper chart production line which uses the new data management system. At the time of writing, the data management system, project phase 1, is about to be ready for testing. At the Finnish Maritime Administration the question of the day is how to implement the new chart production line, which confirms that different products (ENC and paper chart) are not in conflict with one another. This paper starts with a general description of the present state of production of nautical charts and ENC data. The paper then outlines the problems and challenges which must be solved before it is possible to produce these different kinds of products, ENC cells and paper charts, from the same database. This is done by comparing the characteristics of the products and by introducing some solutions how to handle these differences.

Charting and Navigation / Cartographie marine et navigation 12 Index

1. Introduction

1.1 Background Navigation practices have changed dramatically over the last 10 years due to the modern global positioning technology. Navigation had traditionally been based on visual observations and radar measurements, which give an indication of current position on paper chart. This method is called ’relative navigation’. Nowadays navigation is to a great deal based on so called absolute methods. These satellite driven services, such as GPS and Differential GPS and Russian GLONASS are already widely available. A European option, GNSS, will be operational in due course. Specific requirements of nautical chart production compared to topographic map production: • Accuracy of the data affects directly to the safety of navigation • The requirements for the correctness of the data are very high • Different nautical products must be consistent • There must be arranged a rapid updating service New technology requirements for the nautical charts: • the positional accuracy of the chart data must meet the increased accuracy of the positioning systems (e.g. DGPS 1 – 5 m). • in order to fully benefit from the dynamics of the modern positioning methods, there must be available digital chart products parallel to the traditional paper charts. These would allow real time tracking of ship’s own position on chart display system. • digital data contains valuable information of chart features and the navigator does not necessarily have to search for that information from printed nautical publications. • there must be arranged on-line updating services for the digital products The focus has been on pure technical development and therefore it has become evident that the effects of the new infrastructure to the content, production processes, distribution services and use of nautical chart products has re- mained vague.

1.2 The urgency of the ENC production Shipping companies have, for some time now, requested electronic chart systems, which would meet the requirements of both the navigator and International Maritime Organisation’s SOLAS convention as well as proper chart data for these. Relevant international standards for ECDIS (Elec- tronic Chart Display and Information System) and ENC (Electronic Navi- gational Chart) were issued fairly late 1995 and 1996 by the respective bodies and this has caused difficulties in fulfilling the commitments towards Figure 1 Information sources of the Electronic Chart Display and the end users. Information System

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

The division of work has however been quite clear: The end user requires services, regulatory bodies announce standards, private system manufacturers develop ECDIS and other charts systems and national hydrographic offices put efforts into ENC production. Most of the hydrographic offices have chosen the fastest solution to the ENC production challenges. They vectorise current paper charts as such into ENC format. Only a little extra information is added apart from the information content of the paper chart. In Finland we are in a situation, where approximately 50% of the national sea area is already covered by vector data (internal Fingis format) whereas the rest of the water areas are covered by manually maintained charts. Almost all of the Finnish navigational aids (beacons etc.) are however in a common Oracle database and the information is available in digital form for many purposes, including chart production. We are currently taking into production use a new system called HIS (Hydrographic Information System), which: • uses vectorised chart data (Fingis) and navaids information (Oracle) as main source data • allows editing, quality assurance and storing to a seamless HIS database • enables updating, multiple scales and history management as well as other traditional database management functions • allows to produce ENC data and its feature based updates according to pre-defined product definitions The benefits of this arrangement include the ability to exploit in the new ENC products the hard work already done in processing Fingis data for the current printed chart production purposes. The key weaknesses are two- fold: The first deficiency is related to the big differences of the Fingis and ENC data structures, which cause a lot of interactive work when converting Fingis files into the HIS system. The other is the need for duplicated maintenance and updating of chart data for both the printed chart production (making all accrued changes into Fingis files before printing) and ENC production (continuous updating of the HIS database).

1.3 Finnish Maritime Administration aims to unify ENC and printed chart production processes There are many good reasons to unify the production of ENC and printed charts: • common production line increases the level of the consistency between the products. That has positive effects to the quality, which increases the safety of navigation • a single, combined process flow instead of a duplicated one will make data management more efficient and reduce the costs • it will be easier to develop the content of the products when the production line is well integrated. This will have positive effect to the provision of the services to the mariners We have started at the Finnish Maritime Administration a project for replacing the well served FINGIS (Finn- ish Geographic Information system) system with a new printed chart production line, which would not only bring improvements to the production of printed charts but also facilitate closer integration with the HIS system and the ENC production. During the system specification feasibility study we have met several problems which have to be solved before system implementation. This paper outlines three remarkable ‘challenges’ which must be solved before it is possible to produce ENC cells and paper charts from the same database. These ‘challenges’ are as follows: 1. Cartographic presentation 2. Multiscale data management 3. The relationship between updating of printed charts and ENC cells

Charting and Navigation / Cartographie marine et navigation 12 Index

2. Cartographic Presentation

2.1 Differences in printed charts and electronic navigational charts Printed charts present all important information as chart objects with appropriate symbology and descriptive cartographic information texts and symbols. The volume of information is limited due to the size of the chart as well as the readability aspects of it. One of the most important preparation work of the data to be published on the printed chart is cartographic generalisation and cartographic editing of the data. This includes e.g. displacement, aggregation, selection, rotation and text width, font and placement [Jatkola, M. and T. Tuurnala, 1998]. ENC data is presented on the CRT display of the ECDIS. The symbology of the chart objects is determined by the object class and its attribute values as well as very dedicated conditional symbology rules, which take into consideration the current navigational situation and the nature of the surrounding data. It is easy to add and remove information on the ECDIS display, the user can select which object classes he wants to be visualised on the display (except ‘display base’ object classes) [IMO, 1996]. A large part of the attribute information does not affect to the symbology, and the mariner can have an access to the information by a simple spatial query function. Some of the features do however have such important attribute information that is displayed on the screen as text without any separate query, if the mariners choose that display option. ENC data also uses so called meta object classes in order to provide the mariner with information about e.g. the quality of the chart or the direction of boyage system applied locally. Should tidal information or traffic restrictions exist, these can be presented as raster images or text files, which are stored as ‘annexes’ to the ENC file and opened by attribute query to the appropriate ENC objects on the ECDIS display.

2.2 Challenges The level of the presentation detail in ENC information is not limited by the size of the display and its readability aspects in a same manner as in paper charts since the user can select which information is presented at each moment. The real core of the problem is the question of how to unify the production processes and data management of the two navigational chart products i.e. ENC and printed charts.

2.3 One possible solution Our principal hypothesis is that the major part of the symbology of printed charts can be directly determined from the information stored into ENC object classes and attribute values. Only a minor part, i.e. pure cartographic information, will need additional object classes or attribute fields apart from the stand- ardised ENC data content.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

An example of a need for additional attributation in order to provide good symbolisation in printed charts: Figures 2 and 3 present lateral marks in ECDIS symbology, where zooming of chart display helps in keeping the view clear. Figure 4 presents the method in printed charts, i.e. rotating the spar buoys in order to allow all the information to fit into available space. This rotation of buoys could be handled with one additional attribute compared to pure ENC object model. Another good example is the displacement of depth figure. In ENC the depth figure has to be exactly in the correct position, but sometimes this is not possible on paper chart because of limited space. This can also be implemented by additional attributes (displacement, dx, dy).

3 Multiscale Data Management

3.1 Research Problem Multiscale data management is one of today’s key research topics at the field of cartography science. Multiple representation and generalisation has been regarded as one of the most difficult tasks of the cartographer. In multiple representation databases the same objects/features are presented in separate layers that provide various spatial resolutions and degrees of details. There are several alternative solutions to handle the problem of multiscale data management but however many problems at this area are still unsolved. Here the problem of multiscale data management is approached from the point of view of nautical charting. The multiscale data management as part of the problem of integrating the production and data management of two different kind of products, ENC and paper charts, makes the problem of multiscale management even more challenging. Traditionally, digital nautical chart data has been prepared to be published at a certain unambiguously specified scale on printed chart. The purpose of databases has been the production of paper charts. This way the cartographic generalisation and visualisation of the data can be modified to serve the needs of one certain scale paper chart. However, this traditional method is not sufficient anymore, because of a completely new kind of product, electronic navigational chart. Types of paper charts published by the Finnish Maritime Administration are presented in the Table 2. ENC has set new demands on presentation and visualisation of the data. The data is planned to be visualised on the screen in different scales defined by the user (zoom-in, zoom-out). ENC data is also produced in scale related navigational purposes. These are defined by the S-57 standard, see Table 1.

Table 1. Definitions of navigational purposes of ENC [IHO, 1996; TSMADWG, 1998] Navigational purpose Definition Overview Oceanic crossing and route planning for large and/or long distance crossing – Worldwide coverage. General Commonly used as coast is approached from the open ocean or for sailing along the coast, frequently outer sight of the coast, between distant major ports. Coastal Navigation along the coastline, with well offshore courses, eventually intricate and frequently in sight of the coast. Approach Near shore navigation to gain access to major ports through channels or marked areas leading to a port, and navigation through intricate or congested coastal waters. Harbour Navigation in harbours or smaller waterways and for anchorage. Berthing Sufficient data to allow a vessel to berth.

Charting and Navigation / Cartographie marine et navigation 12 Index

The standard does not define the resolution (except the point density of linear features) or the degree of generalisation for each navigational purpose. The scale of “source data”, from which the certain navigational purpose ENC data is produced, is decided by the Hydrographic Office. In addition to navigational purposes, the presentation of ENC data can be managed by the scale attribute (SCAMIN), which defines the smallest scale where the object is visualised.

Table 2. Scales of paper charts in the FMA at the moment. [FMA, 1997]

Type of chart Scale Definition General chart 1:100 000 – 1:500 000 Intended for high sea navigation and voyage planning. Coastal chart 1:50 000 Intended for navigation in the archipelago and on the coast. Special chart 1:5 000 – 1:25 000 Intended to facilitate harbour traffic.

It has been noticed that the generalisation level of a paper chart at a certain scale is not sufficient to be visualised on the screen. E.g. aids to navigation, rocks and buildings cause problems (see figures 5 and 6).

3.2 Possible Solution to be studied Our primary aim is to produce both ENC cells and paper charts from one common database. This means that the database have to be implemented so that production of two different kinds of products in different scales is possible. Figure 7 presents a multiscale model where production of ENC and printed chart has been taken into consideration. The basic idea is that the data stored in the database scale layer is generalised for certain scale printed chart. To improve the readability of ENC the attribute ‘minimum scale’ has been used (SCAMIN). This means that the object is visible when using the attribute value or larger scale on a certain naviga- Figure 5. An example of ENC data on tion purpose. the screen in scale 1:50 000. The compilation scale of each navigational purpose (e.g. Coastal: 1:50 001) has been determined by experimental research. The basis of our testing was to use as detailed data as possible. In other words each navigation purpose data is used on as small a scale as is readable. This smallest readable scale defines the compilation scale of the next navigation purpose. E.g. the data on navigation purpose Approach is still readable in scale 1:50 000, but not in smaller scale. This defines the compilation scale of navigation purpose Coastal (1:50 001). The use of database scale layers for each navigational purpose is decided in order to correspond the definitions of navigational purposes (see Table 1). The test data was Finnish Maritime Administration’s unofficial ENC data. Figure 6. A copy of 1:50 000 scale paper chart from the same area.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Figure 7. Multiscale management model for ENC and paper chart production.

4 Updating Services

4.1 Distribution service for electronic navigational charts The co-operative body of the European hydrographic offices, the Northern Europe RENC (Regional Enc Co- ordinating Centre) has introduced a commercial ENC distribution and updating service for mariners. This service covers official ENC products from most of the European hydrographic offices and they are sold under a common brand name PRIMAR (see figure 8).

4.2 The relationship between traditional updating of printed charts and modern ENC updating messages This PRIMAR service has been faced with the question of how the publication schedule of NtMs (Notices to Mariners –booklet, which is mailed to mariners in order to give him instructions to make needed chart correc- tions) and the production schedule of ENC update messages would have to be synchronised. The fundamental

Charting and Navigation / Cartographie marine et navigation 12 Index question for the Hydrographic Offices is what their legal liability is regarding the requirement to provide digital ENC update messages at the same time as their paper equivalents (NtMs for printed charts). This question is related to the status of digital chart products like ENC versus a traditional printed chart product. Are these considered as two sovereign products, which are also updated inde- pendently, or are they basically the same nautical product in two different media, but with a common status in Figure 8. The concept for ENC distribution and updating service terms of updating frequency and schedule toward the end user? The opinions among the European hydrographic offices differ and the grounds for the deviations can be traced back to the national practices and circumstances. Some hydrographic offices produce their ENCs directly from printed charts and use the same chart correction cycle for both products. Other offices on the other hand compile their ENCs from many different digital sources and they end up with a navigational product, which differs from the corresponding paper charts in many respects. The updating cycle may also differ. The use of this wider source information may cause situations, where a need for an ENC updating message may be necessary even though the change would not affect to the paper chart in any way.

4.3 Discussion Closer harmonisation of the ENC and printed chart data content and production processes would ease the problem of two updating services. If all the incoming correction information was treated in only one process leading to one update to one common database, there would be less need for publishing any ENC update messages or NtM notices in a separate cycle. This approach, if applied, would cause a change to the current updating practices of the printed charts. Updating of the data for printed charts has traditionally followed printing schedules, but now, if two production lines were unified, a real-time updating of the data would be more feasible. In Finland our aim is a well-synchronised updating service, since we believe that it would be advantageous for the mariners. We do not however see it necessary to come to any specific international decisions on the legal liability aspects of the relationship of the two updating services. Each Hydrographic Office should apply the most suitable practice according to the national legislation and their chart production processes.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

5. Conclusions

In this paper the ‘key challenges’ of production of ENC and paper chart from one common database was presented. Cartographic presentation and multiscale management are the most challenging aspects to be considered when unifying the production. Here the problems are approached by comparing the characteristics of the products and introducing some solutions, but however, more studies are needed. It is evident that there are many other aspects, which have to be taken into consideration, especially concerning the implementation of the database (technical aspects) and the updating process of multiscale database. It also is obvious that the production process (workflow) and the updating services can not be the same as before. Our intention is to continue the research and development and publish more detailed experiences later.

References:

Finnish Maritime Administration (FMA), 1997. Finnish Nautical Charts, Hydrography and Waterways Department Helsinki, 20 p. International Maritime Organization (IMO), 1996. Performance Standards for Electronic Chart Display and Informa- tion System (ECDIS). Resolution A.817(19). Jatkola, M. and T. Tuurnala, 1998. Multiscale Data Management as a Part of New Chart Production Process. Master’s Thesis, Department of Surveying, Helsinki University of Technology, 106 p. (in Finnish) Kilpeläinen, T, 1997. Multiple Representation and Generalization of Geo-Databases for Topographic Maps. Doctoral Thesis, Finnish Geodetic Institute, Kirkkonummi 1997, 229 p. International Hydrographic Organization (IHO), 1996. S-57 – IHO Transfer Standard for Digital Hydrographic Data, Edition 3.0 Transfer Standard Maintenance and Application Development Working Group (TSMADWG), 1998. Proposed Clarifi- cations to S-57 Appendix B1, Annex A – Use of the Object Catalogue for ENC. Proposals from SHOM – France Clarifications, TSMADWG meeting, Monaco 19-23 October 1998.

Charting and Navigation / Cartographie marine et navigation 12 Index

Session / Séance 12-B An expert system approach for the design and composition of nautical charts

Lysandros Tsoulos Cartography Laboratory, Department of Rural and Surveying Engineering, National Technical University of Athens, 9, H. Polytechneiou St., 15780 Zografos, Athens, Greece. Tel: (30+1)7722730, Fax: (30+1)7722734 email: [email protected]

Konstantinos Stefanakis Cartography Laboratory, Department of Rural and Surveying Engineering, National Technical University of Athens, 9, H. Polytechneiou St., 15780 Zografos, Athens, Greece. Tel: (30+1)7722639, Fax: (30+1)7722734 email: [email protected]

Abstract The current digital cartographic systems do not incorporate tools to support an ‘automated’ map/chart design and composition procedure. This is due to the fact that cartographic design is a complex and rather subjective process, mainly based on the cartographic knowledge, which can not be easily described algorithmically. On the other hand traditional cartographic knowledge is perishing and is being substituted by map/chart specifications which do not always cover the diversity of cases appearing in map/chart design and composition. Thus the digital cartographic systems – although constitute valuable tools for the production of maps/charts – do not lead to efficient and economic solutions. The utilization of the expert systems technology to substitute - in a certain degree – the human factor and to ‘absorb’ the knowledge required for the design and composition of maps and charts, is a very promising solution. This paper elaborates - at a conceptual level - a ‘hybrid’ system utilizing the technologies of Geographic Information Systems and Expert Systems towards the production of Nautical Charts.

Introduction

The idea of the utilization of expert systems technology in a cartographic design and production environment, is not a new one. A number of serious attempts have been made [Forrest, 1995; Freeman and Ahn, 1984; Tsoulos and Stefanakis, 1997] which were successful in solving particular cartographic problems like cartographic design of small scale maps, name placement, sounding selection etc. It is evident that a ‘holistic’ approach to the problem is required, which will possibly address map design and composition process, as a whole. Such an approach, beyond the software components of geographic information system and expert system, requires the full set of specifications used for the design of the specific cartographic product, which will be used for the development of the knowledge base of the system. Nautical chart, is one of the cartographic

Charting and Navigation / Cartographie marine et navigation 12 Index

products covered by detailed specifications agreed and adopted by the international hydrographic community. The above mentioned standardization level, along with the relatively simple design of the nautical chart, constitute an ideal paradigm for the development of an expert system leading to its production. In the framework of this project, an expert system (Elements Environment) interfaced with a Geographic Information System (Arc/Info) is being used. Elements Environment incorporates through its knowledge base, the design and composition methodology and handles the wide variety of entities appearing in nautical charts. Rules capture the knowledge necessary to solve particular domain problems (i.e. resolution of graphic conflicts) and they represent among other things relations, heuristics and procedural knowledge. Rules are sym- metric so they can be processed in either a forward or backward direction. Elements Environment provides with a number of representational structures. There are objects and classes to describe the cartographic entities. There are properties which are characteristics of objects and classes and slots which store information about specific objects and classes. Meta-slots describe how the slots behave. Properties and values can be inherited from a class or object to another class or object. Certain meta-slots can be inherited from a class or object to another object. In conjunction with rules, the expert system supports methods and message passing. Methods can be triggered explicitly after receiving a message from a rule or other method, or they can be triggered automatically following a determination made by the system. Methods can also inherited down the object hierarchy. Elements Environment is an agenda-based system. The agenda is a dynamic mechanism. It is the engine of the system that provides the central transformation between the perception of events and the actions the system will take. It is modeled after the notion of attention. At any time, the complexity of the real world can be reduced to a limited set of parameters and possible decisions. In turn they will affect the world and perhaps the very next events or actions that were planned. Agenda-based programming incorporates the notions of conflict-resolution which is a decision between different possible inference paths and nonmonotonic reasoning. The agenda incorporates forward and backward mechanisms. The Geographic Information System (GIS) manages the geographic entities and provides for the required graphic tools and the interface with the user of the system The system utilizes the entities stored in the cartographic database which has been organized according to the International Hydrographic Organization Transfer Standard for Digital Hydrographic Data [I.H.O., 1996]. There are various architectures that permit the integration of a rule-based system into GIS [Smith and Jiang, 1991]: · To enhance a GIS and the relevant database with rule-based system capabilities, such as knowledge acquisition and representation techniques. · To employ ‘loose coupling’, in which an application is written using an ES shell. · To employ ‘tight coupling’ to facilitate communication between the rule-based system and the GIS. · To build a fully integrated system. Although the ideal solution would be a fully integrated system, this is not realistic for the time being. We therefore focus on a ‘loosely coupled’ system not only because of the significance of this architecture in GIS, but due to the functionality of such a system which includes the following: · The provision of the services of an Expert System · The triggering of external actions in response to Expert System evaluation of rules · The provision of state constraints, including referential and semantic integrity constraints. When the efficiency of this environment is proved, this will justify the design and implementation of a fully integrated system.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

The chart production process

The production of a nautical chart with the utilization of the ‘system’ is implemented through the following phases: Area Definition, Selection of Chart Information, Transformations, Identification of Portrayal Method, Graphic Conflict Resolution, Portrayal of Symbols and Texts, Generation of Supplementary Chart Information (e.g. tittle, tables, notes), Production. The degree of involvement of the expert system and of geographic information system, varies from phase to phase, due to the nature of the processes inherent to each phase. We can generally distinguish the phases and the relevant actions of chart design and composition process to those based on ‘knowledge’ and those based on ‘algorithms’. The first category includes the phases of Selec- tion, Identification of Portrayal Method and Graphic Conflict Resolution. These processes are being resolved in the Expert System environment. In the following paragraphs the phases of Method of Portrayal Selection (Design) and Graphic Conflict Resolution (Composition) are elaborated.

Conceptual framework

The portrayal of cartographic entities on the chart is dictated by their degree of importance, their relationship with the surrounding entities and the size of the respective symbols. It is evident that the problem is a generalization problem. A ‘holistic’ approach to this problem would entail the simultaneous participation of all entities to the generalization process at first place. This would lead to a very complex and uncontrollable process. If this approach were substituted by a layered one, the layers and the entities assigned to them would be members of a hierarchical structure implying that entities which are members of the higher levels dictate the portrayal of entities belonging to the lower ones. The same logic can be applied for the entities belonging to the same layer. The above mentioned approach must be embedded to a digital generalization model suitable for the integration of expert systems technology. The Brassel and Weibel, model [Brassel and Weibel, 1988] consisting of structure recognition, process recognition, process modeling and process execution – with minor modifications - is considered to be the most appropriate one, due to it’s inherent characteristics which serve efficiently the required processes. The modifications refer to the operational steps (process execution) where the system presented here, distinguishes the generalization at the layer level with the generalization at the object level.

Knowledge base architecture

The expert system applies two main representational paradigms: objects and rules. The system designer describes the ‘world’ (the nautical chart with its entities) in terms of physical symbols (objects), generalizations of physical objects (classes), parts of physical symbols (sub-objects) and attributes of physical objects (properties). The knowledge in the domain is coded in the form of rules which constitute the ‘building’ components of the knowledge base. The application logic and the procedural information of the system, is described by rules and operate on objects, classes and slots. The knowledge base of the expert system for the design and composition of nautical charts, contains the following categories of rules: · Selection rules serve the selection from the database of those entities required for the chart production. The selection rules generate the representations of chart entities to the object-orientated structure of the system · Design rules give cartographic ‘substance’ to the entities of the representation

Charting and Navigation / Cartographie marine et navigation 12 Index

· Composition rules make the appropriate changes to the portrayal of the chart entities in order to resolve the undesired events, according to the agreed chart specifications · Procedural rules control of the overall process and guide the system through the various phases The expert system enables the modularization of the knowledge base, by breaking it up into several knowledge bases. This feature is being utilized in this application, in order to provide efficiency and assist control during the development of the system. The selection, design, composition and procedural rules, are organized into separate knowledge bases which are loaded and unloaded accordingly. The central mechanism which is responsible for the control of these procedures is composed of procedural rules. The individual knowledge bases are also provided by central mechanisms. These mechanisms control the processes executed within their environment. These mechanisms can also control the behavior of the agenda (strategy). Rules within the same knowledge base can be grouped. The central mechanism can call sets of rules instead of individual rules in a specific sequence.

Method of portrayal

The identification of portrayal method for the chart entities, is generally dictated by a number of factors relevant to the chart (e.g. chart scale, chart category, congestion of chart information) as well as factors inherent to the entities themselves (e.g. entity characteristics and attributes). Chart specifications describe in detail the method of portrayal for all entities which may be portrayed on a nautical chart at any scale. For instance airports on medium scale charts are depicted by their outline, while in small scale by a point symbol. Some entities are portrayed with ‘simple’ symbols while others with more complex ones. Soundings are portrayed with numbers representing their values, while lights are portrayed with the use of point symbols along with text describing their characteristics. ‘Knowledge’ encompassing the diversity of methods of portrayal, as well as the way they are implemented on the chart, is also encoded in the form of rules. The design process in the expert system environment, is initiated as soon as the cartographic entities which will be used for the composition of nautical chart are loaded and organized in the object-oriented model of the system. The relational structure table-record-item implemented by the database management system and the geographic information system, is mapped into the object- oriented structure class-object-property of the expert system. Chart entities constitute objects belonging to classes. The conditions under which any entity is linked to the appropriate method of portrayal, is expressed in the form of rules (design rules). The set of these rules contain conditions which evaluate parameters like chart scale and the particular attributes of the objects which may influence their portrayal. The Figure 1. The method of portrayal selection for wrecks.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

evaluation of these conditions results to the method of portrayal. This is implemented by the attachment of the particular entity objects to classes (design classes), where the various portrayal methods have been organized. The objects/entities inherit the methods of portrayal from the parent class. Figure 1 describes this process for wrecks. The system in order to support the graphic conflict resolution, implements specific representations for the chart entities. Point, linear, areal and text entities are represented as follows (see Figure 2): · Point entities are represented by their minimum boundary rectangles (MBRs) · Linear entities are represented by the edges of the Constraint Triangular Irregular Network (TIN) which is computed using their centerlines. The edges of the linear entities have specific direction and attributes, such as the buffer distance · Areal entities are represented by the triangles of the Constraint TIN which is based on their outlines. Each areal entity is composed by the triangles which are located into it · Text entities are represented by polygons according to their alignment. Texts aligned as straight lines (e.g. light characteristics) are substituted by their MBRs. Texts aligned along curves (e.g. toponyms) are sub- Figure 2. Representations of the chart entities stituted by their bounding polygons An object attached to a specific design class in order to inherit the method of its portrayal, inherits the parameters which will enable its ‘abstract’ representation. In the case of point entities, objects inherit the relative coordinates in respect to the bottom-left and top-right corner of the respective MBR. The bounding polygons are computed for the text entities, while buffer distances are inherited for the individual linear entities.

Graphic conflict detection and resolution

Once the entities of the area to be charted are selected, assigned to the predefined layers and the way of their portrayal is identified, the phase of conflict detection and resolution is executed. In general, cartographic entities require more space than their actual dimensions dictate. Maps/charts also portray ‘abstract’ phenomena like names (e.g. text descriptions, toponyms), isolines (e.g. contour, depth contours), heights or soundings which are not tangible and do not have real dimensions. These entities provide an additional source of graphic constraints which are generally resolved through omission, simplification, exaggeration, combination and displacement, as well as combinations of them. The interactions between point, linear, areal and textual entities may generate graphic conflicts. Entities are represented within the expert system environment as polygons (MBR is considered as a special case of polygon), edges and triangles of a constraint TIN. In order to detect graphic conflicts, the system searches for all possible combinations among these representations. The graphic conflicts which are expected to be detected and resolved are: · Polygon vs. Polygon · Polygon vs. Edge · Polygon vs. Triangle

Charting and Navigation / Cartographie marine et navigation 12 Index

· Edge vs. Edge · Edge vs. Triangle · Triangle vs. Triangle The detection of graphic conflicts will be carried out through the evaluation of the condition expressions in the composition rules. This will trigger the appropriate actions for their resolution. The graphic conflict resolution methods are organized into classes (conflict classes). The individual conflicts produce conflict-objects having as slots the characteristics of the entities involved. The conflict-object is attached by the rule that has been detected to the relevant conflict class, where from it inherits the appropri- Figure 3. Graphic conflict resolution process ate methods for its resolution (see Figure 3). The above mentioned methodology for the resolution of graphic conflicts must eliminate the possibility of generation of new graphic conflicts. The cartographic products are generally characterized by the interrela- tions of the graphic entities. This creates serious difficulties to the design of a linear cartographic composition process which is indispensable for the development of an ‘automated’ cartographic system. The design of a linear composition process is being considered here. The relatively ‘simple’ design of nautical charts, enables the drawing of guidelines which the system follows in order to resolve the graphic conflicts and create progres- sively the final image of the chart. The land and sea parts of the nautical chart are processed separately. These two basic layers (or more precisely sets of layers) are adjoined along the coastline. The resolution of graphic conflicts among topographic/ hydrographic entities is being executed in the respective areas. This approach minimizes the synthesis problems of the individual sets of layers. Chart entities are organized in layers according to their individual characteristics. Homogeneous entities and entities which must be processed simultaneously, share common layers. The applied procedure composes first the layers which include entities belonging to the higher levels of hierarchy and moves to the lower ones. For instance, the system first processes the layers containing navigational aids or dangers and subsequently the layer of soundings which is more flexible in the sense that if a sounding generates a conflict, it may be omitted and an adjacent one may be portrayed instead. The sea part of nautical charts is generally organized into the following layers which are processed in accordance with the sequence they appear: · Coastline, depth contours · Navigation dangers, navigation aids, ‘areas’ (e.g. restricted areas, anchorage areas, pipe areas) · Depth Soundings · Nature of seabed In order one layer to be processed is ‘added’ to the already processed layer or layers. The scales of the original database and of the chart under construction, must be relevant. Small differences in scale do not lead to really problematic situations and minimize the effects of generalization processes to topology.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Conclusions

The efficient utilization of geographic information systems and expert systems technologies, provides promising solutions to a number of cartographic problems. The procedures of design and composition of maps/charts can be ‘automated’ to a considerable degree within the environment of a hybrid system. Expert systems provide flexible mechanisms to detect and resolve conflicts, according to the designers needs. The implementation of such a system, requires a flexible and integrated application logic which must be embedded into it. Tailoring this logic to a specific map type - the nautical chart – is considered essential due to the fact that rules effective for the design and composition of one map type, may not be effective for another. The knowledge base must be continuously enhanced with new/refined rules without influencing the internal structure and operation of the system. The numerous situations which emerge, during the design and composition process, can be handled only if the system is well designed and ‘complete’. This implies thorough analysis of the specific map/chart characteristics and identification of all types of constraints (structural, graphic, application, procedural) inherent to the specific map/chart product. Specifications for nautical charts along with its high level of standardization, constitute a sound background for the design and implementation of the system. Without ignoring the inherent difficulties and taking into account the results and experiences gained so far, we believe that such a system is not far from its materialization.

References

Brassel, K.E., and Weibel, R. (1988). A review and conceptual framework of automated map generalization. Interna- tional Journal of Geographical Information Systems, 2(3), 229-244. Forrest, D. (1995). Don’t break the rules or helping non-cartographers to design maps: An application for cartographic expert systems. Proceedings, 17th International Cartographic Conference, Barcelona, Spain, 570-579. Freeman, H., and Ahn, J. (1984). AUTONAP - an expert system for automatic name placement. Proceedings, 1st Inter- national Symposium on Spatial Data Handling, 544-569. I.H.O. (1996). IHO transfer standard for digital hydrographic data. Edition 3.0, International Hydrographic Bureau, Monaco. Smith, T.R., and Jiang Y. (1991). Knowledge-based approaches in GIS. In D.J. Maguire, M. Goodchild, and D.W. Rhind (Eds.). Geographical Information Systems. Principles and applications. John Wiley & Sons, New York. Tsoulos, L., and Stefanakis, K. (1997). Sounding selection for nautical charts: An expert system approach. Proceedings, 18th International Cartographic Conference, Stockholm, Sweden, 2021-2029.

Charting and Navigation / Cartographie marine et navigation 12 Index

Session / Séance 27-C Croatian State Boundary at the Adriatic Sea

Ivka Tunjic University of Zagreb, Faculty of Geodesy, Kaciceva 26, 10000 Zagreb, Croatia E-mail: [email protected]

Miljenko Lapaine University of Zagreb, Faculty of Geodesy, Kaciceva 26, 10000 Zagreb, Croatia E-mail: [email protected]

Abstract The increased possibilities of geographic information systems in spatial data management require repeated auditing of basic presentations and measurements in spatial analyses. The paper explains the geometric way of determining the outer boundary of the territorial waters. The applicability of Gauß-Krüger projection has been tested in solving the given problem in the plane. The practical application of the suggested methods has been illustrated in the example of the outer boundary of Croatian territorial waters.

1. Introduction

The majority of answers referring to the maritime boundary contains a statement that the “straight lines” will be used for connecting the set of points which are given by geographic coordinates. This is one of the omis- sions that can lead to various interpretations. One should namely distinguish between the geodesics, great circle and loxodrome [Thamsborg, 1974]. The “distance” is for the seaman the arc length of loxodrome, but for a surveyor it is most often the length of the geodesics. Since the maritime chart is based on the map projection, the nature of the straight line depends on the geometrical properties of the projection. Baezley [1982] has noticed that the Mercator projection is not very convenient for the determination of distances, but he does not say which projection it might be. Carrera [1987] deals with the method of delimitation with equidistant boundaries between coastal states. Mayer et al. [1992] suggest the approximate method of determining the length of the geodesics by applying them to the maritime boundaries. From the cartographic and geodetic point of view, the most contents about the problem of the maritime boundary determination can be found in the book of Maling [1989]. In this paper the method for the determination of the line at the given distance from the given line at sea is suggested. One can say, that it is a modified and automated version of Shalowitz method [1962]. The research has included the estimation of the deviation of geodesics and loxodrome from the straight line in the particular example of the Croatian maritime boundary determination in the Adriatic Sea. In practice, when drawing the outer boundary of the territorial sea, one of the two following methods is applied, or their combination (see Figure 1): a) Parallel line method (tracé parallèle in French). A line determining the outer limit of the territorial sea is drawn parallel to the baseline. This method can be applied when the coastline is relatively straight.

Charting and Navigation / Cartographie marine et navigation 12 Index b) Method of envelopes of arc of circles (méthode de la courbe tangente in French). The arcs of circles are drawn with the centres at the most forward positions of land and with the radius that corresponds to the territorial sea breadth of the coastal state. Then, the outer parts of the arcs of circles, between the points of inter- sections with the neighboring arcs, represent the outer limit of the territorial sea.

At the end of the introductory part of this article, we would like to give some remarks. When talking about a straight line or a circle, then it should be clear that such objects generally do not exist on a sphere or ellipsoid. On the other hand, if one thinks of the representation in a plane, then the name and parameters defining the used map projection should be stated. Namely, it is well known that the properties of map projections regarding the distortions of lengths, angles and area are quite different. Regarding the methods of envelopes of arcs, shown at Figure 1A, one can see that the choice of centres of circles has been roughly estimated and probably could be refined in some way. The parallel line (see Figure 1, B and C) is rather disputable because one don’t know why the distance has to be measured in the shown direction, and not in some other direction where the distance will be completely different! Figure 1. Determination of the outer limit of the territorial sea: A – method of arcs; B and C – method of drawing a parallel line to the straight baseline (B) and to the normal 2. Method of perpendiculars or a baseline when the coastline is relatively straight (C); ac- rectangle on a sphere cording to [Rudolf, 1985]

When talking with colleagues about the determination of the outer limit of the territorial sea as a line whose each point has a property that the distance to the closest point at the baseline is equal to the coastal state territorial sea breadth, one can usually hear that it is simple problem which can be solved by drawing the perpendiculars to the baseline. In this Chapter we will show that it is not true. Namely, if the earth would be flat, then the problem will be really simple, and it could be solved by drawing the appropriate perpendiculars to the baselines. However, due to the earth’s curvature the problem has to be solved directly on the sphere or ellipsoid, or in the plane of projection. Let us consider a rectangle. It is a four-sided polygon in a plane with four right angles and its opposite sides are of equal length. Now, let the perpendiculars AC and BD be drawn to the straight line segment AB in a plane and at the same side of the segment (see Figure 2). If the segments AC and BD are of equal lengths, then ABDC is a rectangle, and the length of CD is equal to the length of AB. This is known to everybody from the basic mathematics.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Let us try to do the same on a sphere instead of doing it in the plane. For this purpose, let us take a unit sphere with geographical coordinates f and l on it. Let AB be the arc of an orthodrome or great circle, e.g. equator, and let its length be ll (se Figure 2). At the end points of arc AB let the perpendiculars AC and BD be drawn. If the length of arcs of orthodroms AC and BD are equal, then ABDC is not a rectangle because the length of arc of orthodrome CD is not equal to the length of arc of equator AB, and the angles ACD and BDC are not right angles.

Figure 2. Rectangle ABDC in a plane and “rectangle” ABDC on a sphere The previous statement is a little bit unusual, or at least it is not obvious. Therefore, it has to be proved. Iin order to do so, let us denote the following points A(0, ë1), B(0, ë2), C(ö1, ë1), D(ö1, ë2). Let the length of arc of orthodrome from the point C to the point D be ó. The length of the arc of an orthodrome connecting points T1(ö1, ë1) and T2(ö2, ë2) can be computed by using formula [Lapaine, 1997]:

σ = ϕ ϕ + ϕ ϕ ∆λ cos sin 1 sin 2 cos 1 cos 2 cos (1) and in our case the length of the arc CD is σ = 2 ϕ + 2 ϕ ∆λ cos sin 1 cos 1 cos , (2)

or after the appropriate trigonometric transformation ∆λ cosσ = cos ∆λ + 2sin2 ϕ sin2 . (3) 1 2

From the last relation one can see that it is ó<Äë, except when ö1=0 or Äë=0, which are special cases when ABDC degenerates in an arc. In that way we proved that the length of the arc CD of orthodrome is generally always smaller than the length of the arc AB of orthodrome. Now we are going to show that the angles ACD and BDC can not be right angles.

It is known that the angle ã between the orthodrome passing through the points T1(ö1, ë1) and T2(ö2, ë2) and the meridian passing through the point T1 is given by the expression [Lapaine, 1997]:

λ − β γ = cos( 1 ) cos , (4) 2 + where k 1

tan ϕ sin λ − tan ϕ sin λ tanβ = 1 2 2 1 , (5) ϕ λ − ϕ λ tan 1 cos 2 tan 2 cos 1

sin(λ − λ ) k = 1 2 . (6) 2 ϕ + 2 ϕ − ϕ ϕ λ − λ tan 1 tan 2 2 tan 1 tan 2 cos( 1 2 )

Charting and Navigation / Cartographie marine et navigation 12 Index

If the orthodrome is going through the points C and D, then after substituting their coordinates in (5) and (6), the appropriate transformations can give the following:

λ + λ π (7) β = 1 2 − , 2 2

∆λ = − ϕ k cot 1 cos . (8) 2 Substituting the expressions for â and k from (7) and (8) in (4), after some manipulations one can get the relation

sin ϕ cos γ = 1 . ∆λ (9) cot 2 + sin 2 ϕ 2 1 From the last formula (9) one can see that the angle ã between the orthodrome and the meridian in the point o o C is the right angle, if and only if it is ö1=0 or Äë=0 . These are special cases when the four-sided curved o o polygon ABDC degenerates in an arc. If it is for instance Äë=90 , then it is cosγ = 1/ 3 , for ö1=45 , while o o o for ö1=60 , it is cosγ = 1/ 5 . If it is for instance Äë=180 , then for any ö1, it is cos ã =1, i.e. ã = 0 , which means that in curved polygon ABDC two angles are 90o, and the two other 180o (see Figure 3)!

From the previous elaboration it can be concluded that there are no rectangles on a sphere in the way we are used to it in a plane. On the sphere a “rectangle” ABDC can have two neighboring right angles and two opposite sides of equal length, but two remaining angles generally will not be right angles and the other two sides will not be of equal length.

Figure 3. “Rectangle” on a sphere having two Figure 4. “Duality” of the right angle right and two straight angles

According to that, if one moves from the point A belonging to an orthodrome and goes along a great circle making a right angle with the first orthodrome, then he will arrive to the point C belonging to some other orthodrome (see Figure 4). If he moves back going from the point C along the great circle that is perpendicular to the orthodrome passing trough the point C, he will not come back to the point A. In other words, on the sphere there are no parallel “straight lines” at all.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

3. Territorial sea of Croatia

It is a surprising fact that, out of the 22 independent states which have been created since 1990, 11 are land- locked. Of the remainder, some have relatively short coastlines. Croatia is one of the most fortunate in terms of the length of coastline although the large number of Croatian islands makes comparison difficult. This coastline is without question one of Croatia’s most important assets, which will have to be carefully managed and protected [Blake, 1994]. The use of straight baselines is in international law permitted where coasts are highly indented, or fringed with islands. They are important as the baseline from which the width of the territorial sea is measured. Baselines also enclose internal waters, in which the coastal state enjoys the same extensive rights of sovereignty as on land. Former Yugoslavia was one of the first states to adopt straight baselines in 1948. This system of straight baselines was extended in 1965 to an almost continuous line, and has been praised by commentators as a model of a modest and correctly applied straight baseline. Many states in the world have made exaggerated claims to straight baselines, contrary to the guidelines in Article 7 of the 1982 Law of the Sea Convention. Yugoslavia’s baselines have never been disputed by other states, and Croatia continues to apply the same baselines. The 1982 Law of the Sea Convention entered into force on 16th November 1994, and Croatia joined it on 5th April 1995. Croatian Maritime Code determines the Croatian sea and seabed in details, organises safety of navigation, material and law relationships, etc. The Croatian baselines are defined in the Article 19 of the Croatian Mari- time Code [1994]: 1) lines of average low waters along the cost of land and islands, 2) straight baselines closing the entrances in ports and bays, 3) straight baselines connecting the following points at the coastlines: a) cape Zarubaca – SE cape of island Mrkan – South cape of island Sv. Andrija – cape Gruj (island Mljet), b) cape Korizmeni (island Mljet) – island Glavat – cape Struga (island Lastovo) – cape Velje more (island Lastovo) – SW cape of the island Kopiste – cape Velo dance (island Korcula) – cape Proizd – SW cape of the island Vodnjak – cape Rat (island Drvenik mali) – rock Mulo – rock Blitvenica – island Purara – island Balun – island Mrtovac – island Garmenjak veli – the point at the island Dugi otok having the coordinates 43o53’12"N, 15o10’00"E, c) cape Veli rat (island Dugi otok) – rock Masarine – cape Margarina (island Susak) – shoal Albanez – island Grunj – rock Sv. Ivan na pucini – shoal Mramori – island Altiez – cape Kastanjija. The baselines are drawn in the maritime chart S101 Adriatic Sea, the northern and the middle part, published by the State Hydrographic Institute (first edition on 1st March 1971, then supplemented editions in 1973, 1980, 1986, 1990, 1996). The Croatian territorial sea spreads from the straight baseline towards the continental shelf boundary up to the distance of 12 nautical miles (1 nautical mile = 1852 m). In its southern part, the territorial sea is measured from several islands (Jabuka, Svetac, Palagruza etc.), and not from the straight baseline. At the places where the straight baseline is not drawn the territorial waters are measured from the low water mark along the coast. The outer limit of the territorial sea is the line that has each point 12 nautical miles away from the nearest point of the baseline.

Charting and Navigation / Cartographie marine et navigation 12 Index

4. Gauß-Krüger projection

The official map projection for the territory of former Yugoslavia was Gauß-Krüger projection by means of which the entire state territory was mapped in three zones. Prof. N. Francula from the Institute for Cartography at the Faculty of Geodesy, University of Zagreb has suggested the Gauß-Krüger projection for the maps of the entire Croatia at the scale of 1:300 000, or smaller, with the central meridian 16o30’E (Francula 1973, 1981). In order to reduce the linear distortions, the linear scale 0.9997 has been introduced on the central meridian. The question is raised whether the same projection could be applied in the determination of the outer boundary of the Croatian territorial sea. This question will be answered in the next Chapter.

5. Replacement of a geodesics or loxodrome with a straight line

In order to examine the possibility of replacing the geodesics image with the portion of the straight line, three

points have been chosen. The point T1 lies in the northern part of the Adriatic Sea, point T2 in the middle, and

point T3 in the southern part of the Adriatic Sea (see Figure 5). The circles with the radius of 12 nautical miles are delineated around the images of these points in Gauß-Krüger projection. On each of these circles 180 equally arranged points have been selected. For these points their geographic coordinates have been computed. Then, it was possible to compute the length of the geodesics on the Bessel’s ellipsoid from a single point of the circle to the corresponding centre. The analysis of the obtained results shows that the difference between the length of the geodesics on the ellipsoid and the length of the straight line in the plane of projection never goes over 10 m (see Figure 6). The same conclusion can be made after applying the formula for computing the linear distortion d that is defined as dS d = − 1, (10) ds where dS is the infinitesimal linear element in the plane of projection, and ds corresponding linear element on the ellipsoid. According to Borcic (1976), there is in Gauß-Krüger projection:

 2 4  =  + y + y  − (11) d m0 1  1,  2R 2 24R 4 

where the linear scale on the central meridian for the selected projection is m0=0.9997, R the mean radius of the ellipsoid at a point, and y the ordinate of the observed point. Although the previous formula gives the amount of the linear distortion in the point, this formula can be used also for the computing the linear distortion of relatively short segments. R and y can thereby be taken in the middle point of the segment. From the performed research it can be concluded that if we are satisfied with the accuracy of 10 m, then a part of the straight line can be used instead of the geodesics in the selected Gauß-Krüger projection. We shall furthermore take this as a presumption, and thus have the possibility of solving the problem in the plane. A similar procedure has been done with a loxodrome in place of geodesics. The results show that for the distance of 12 NM the length of geodesics and the length of loxodrome differ less than 0,01 m that is certainly negligible regarding the error introduced by map projection. This result was confirmed by computing the most remote point of loxodrome from orthodrome. The computation was based on the formulas derived by Viher and Lapaine [1998] and showed that the most remote point of loxodrome from an orthodrome between two points at the distance of 12 NM never goes over 10 m. Therefrom it can be concluded that in the frame of given accuracy, and in the selected Gauß-Krüger projection there are no differences between a geodesics and a loxodrome. That means that we are able to perform further geometrical research in a plane.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Figure 5. The circles in the Gauß-Krüger projection Figure 6. Residuals of geodesics lengths in three from 12NM points on the ellipsoid surface

6. Mathematical definition of the territorial sea boundary

On the basis of the previous explanations we can give an abstract definition of the territorial sea boundary. We shall say that G is the boundary of the territorial sea with the width d of some set A in the plane if for each point T belonging to the boundary G there is a point P from the set A so that the distance between the points T and P is equal d. More precisely,

G(A,d ) = {}T min {}d(T, P) P ∈ A = d (12)

Thereby d(T, P) means the euclidean distance between the points T and P.

Figure 7. Boundary G of the territ. sea of the point A Figure 8. Boundary G of the territ. sea of the line segment A

When the baseline would consist of only one point A, then the boundary G of the territorial sea would be the circle with the centre in the point A and with the radius d (see Figure 7). If the baseline is the line segment A, the boundary G of its territorial sea is the curve consisting of two semi-circles and two straight line segments (see Figure 8). If the baseline is a polyline A composed of two line segments, as shown in the Figure 9, then the boundary G of its territorial sea is more complex and consists of two semi-circles, one circle arc and four straight line segments.

Charting and Navigation / Cartographie marine et navigation 12 Index

Let us imagine that the baseline A is the polyline consisting of a larger number of segments. The boundary G of its territorial sea will consist of circle arcs and straight line segments. Circle arcs have the centre always in the breaking points of the baseline and the radius d. The parts of the boundary G that are straight lines segments are located at the distance d from a single segment of the baseline.

Figure 9. Boundary G of the territorial sea of the polygonal line A

A geographic information system (GIS) is a system for capture, storage, retrieval, analysis, and display of spatial data. A GIS usually enables to develop topologies, which means a series of defined relationships between nodes, links, and polygonal regions. This information can be analysed to provide data on spatial relationships. Topologies are also a very efficient way to store polygonal or area-based data. Using buffer analysis, or buffering, one can easily identify objects within a specified offset of elements in node, network, and polygon topologies. A buffer is a zone that is drawn around a topology. For example, one can specify a buffer on either side of a river to show the extent of a flood plain. By specifying a buffer offset a new polygon topology is created from an existing node, network, or polygon topology. To conclude, the determination of territorial sea limit can be interpreted as a construction of a buffer of a given width around a network topology made of polygons representing the baselines of a state.

7. Construction of the territorial sea boundary of Croatia by using network topology

As it has already been stated in Chapter 2, the Croatian Maritime Code brought in 1994 defines in its 19 Article the baseline as well. However, this definition is of a descriptive character and as such inconvenient for compu- tations. Since we have not received the list of the coordinates of the baseline points from the state institutions, we have taken them from the diploma thesis (Javorovic, 1993). The list was made by digitising the map S101 Territorial Sea Boundary of SFRJ and the Republic Italy, and the forbidden areas along the Yugoslav coast at the scale of 1:650 000 in Mercator projection. Knowing the standard parallel of the used Mercator projection, it was possible to compute the belonging geographic coordinates with regard to the Bessel’s ellipsoid. The list of coordinates makes a discrete record of the baseline. By applying direct equations of any projection, this baseline can be drawn. We have used known equations of the Gauß-Krüger projection with the scale on the central meridian m0=0.9997 (see Chapter 4). Coastlines of Croatia and Italy and the islands in the Adriatic Sea have been acquired analogously in a few seminars and diploma thesis by digitising from the maps at the scale of 1:1 000 000 to 1:50 000. For the official purpose of determination of outer limit of Croatian territorial sea, one will need to have better numerically defined Croatian baselines. Our list of points of the baseline together with a few islands located separately (see Figure 10), contains 850 points. Drawing the territorial sea boundary manually, by using the definition from the Chapter 6, would really

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

be rather hard work. In a previous paper (Tunjic, Lapaine 1998) the method of solving the entire problem by using AutoCAD has been described.

Figure 10. Croatian baselines and the application Figure 11. Croatian baselines and the outer limit of of network topology with the the Croatian territorial sea appropriate buffer

Due to the AutoCAD Map (Autodesk, 1997), a mapping version of AutoCAD, the whole story can be short- ened. By creating the appropriate network topology and defining a buffer with 12 NM width, AutoCAD Map draws the buffer presented in the Figure 10 in a few seconds. Hence, simply by erasing the unnecessary parts one can easily come to the final solution (see Figure 11).

References

Autodesk (1997). AutoCAD Map, Release 2, Using AutoCAD Map, User’s Guide. Baezley, P. B. (1982). Maritime Boundaries, International Hydrographic Review, Vol. 59, No. 1, 149-159. Blake, G. (1993/94). Croatia’s Maritime Boundaries. In: Croatia – a New European State, Proceedings of the Sympo- sium held in Zagreb and Cakovec, 1993, Department for Geography and Spatial Planning, Faculty of Science, University of Zagreb, 39-46. Borcic, B. (1976). Gauß-Krüger Projection of Meridian Zones (in Croatian), University of Zagreb, Zagreb. Carrera, G. (1987). A Method for the Delimitation of an Equidistant Boundary Between Coastal States on the Surface of a Geodetic Ellipsoid, International Hydrographic Review, Vol. 64, No. 1, 147-159. Francula, N. (1973). Mathematical Basis and Numerical Procedures in the Map Production of SR Croatia at the Scale of 1:1000 000 (in Croatian), Symposium Cartography in Spatial Planning, Ljubljana, A4/1-9. Francula, N. (1981). Application of Computers in the Map Production of SR Croatia (in Croatian), Faculty of Geodesy, University of Zagreb, Zagreb, Proceedings, Series D, Volume 2. Javorovic, I. (1993). Digital Map of Waters of Croatia and Bosnia and Herzegovina (in Croatian), Diploma thesis, Faculty of Geodesy, University of Zagreb. Lapaine, M. (1997). Vector Analysis (in Croatian), University of Zagreb, Faculty of Geodesy. Maling, D. H. (1989). Measurements from Maps, Principles and Methods of Cartometry, Pergamon Press, Oxford, New York.

Charting and Navigation / Cartographie marine et navigation 12 Index

Maritime Code (in Croatian), Narodne novine No. 17 of 7 March 1994, 404-503. Mayer, F., Szychta, D., Cravacuore, M., and Veron, D. (1992). Algorithm for the Calculation of Geodetic Distances for Maritime Jurisdictional Boundaries, International Hydrographic Review, Vol. 69, No. 1, 133-141. Rudolf, D. (1985). The International Law of the Sea (in Croatian), JAZU, Zagreb. Shalowitz, A. L. (1962). Shore and Sea Boundary, U.S. Dept. of Commerce, Coast & Geodetic Survey Publication 10- 1, 2 Vols. Washington, Government Printing Office. Thamsborg, M. (1974). Geodetic Hydrography as Related to Maritime Boundary Problems, International Hydrographic Review, Vol. 51, 157-173. Tunjic, I., and Lapaine, M. (1998). Croatian State Boundary at the Sea, 8th International Conference on Engineering Computer Graphics and Descriptive Geometry, Austin, Texas, Proceedings, Vol. 3, 716-720. Viher, R., and Lapaine, M. (1998). The Most Remote Point of Loxodrome from Orthodrome (in Croatian), Geodetski list, Vol. 52 (75), No. 1, 13-21.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Session / Séance 06-B Production of Thematic Nautical Charts and Handbooks for the Sea Area of the Eastern Adriatic Coast

Slavko Horvat, Zeljko Zeleznjak Ministry of Defense, Zvonimirova 4, 10 000 Zagreb, Croatia Phone: +3851 4567426, Fax: +3851 4567973, e-mail: [email protected]

Tea Duplancic Hydrographic Institute of the Republic of Croatia, Zrinsko-Frankopanska 161, 21 000 Split, Croatia Phone: +38521 361840, Fax: +38521 47242, e-mail: [email protected]

Abstract This paper briefly reviews the maritime hydrographic and nautical cartographic activities in the Republic of Croatia through history and nowadays, especially considering the production of thematic nautical charts and handbooks for military purposes.

1. INTRODUCTION

The Adriatic Sea is an elongated basin (ca 800 km long and 200 km wide). The northern part of the basin is a concave-shaped shelf, whose maximum depths (around 280 m) occur in the Jabuka Pit. The bottom rises at Palagruza Sill (130 m), then deepens in the southern part – the South Adriatic Pit – to about 1200 m, and rises again in the Otranto Strait (780 m). The western coast of the adriatic Sea is smooth, isobaths run parallel to it, and depth increases gradually seaward. The eastern coast is composed of many islands and headlands rising abruptly from the deep coastal water. A thousand islands and more than 5000 km of coastline, numberless straits, passages and other areas dangerous for navigation along the east coast of the Adriatic Sea, causes this area to be an exceptionally complex navigational whole. Adriatic subsea topography, as a part of the macrotectonic origin subsea valley between the Apennines, Alps and Dinarides. Such area needs constant hydrographic and oceanographic survey, as well as production of high quality nautical charts, handbooks and publications, and various thematic charts and handbooks.

Charting and Navigation / Cartographie marine et navigation 12 Index

2. SHORT HISTORICAL REVIEW OF HYDROGRAPHIC AND CARTOGRAPHIC ACTIVITIES

Organized and systematic hydrographic and nautical chart production for the east coast of the Adriatic Sea has a long tradition. French hydrographer Charles – François Beautemps Beaupré performed the first hydrographic survey along the east coast of the Adriatic Sea (Figure 1). From 1806 to 1809 during his campaign, he surveyed the most significant Croatian harbours, performing also astronomic, geodetic, hydrographic and geomagnetic measurements. An atlas with 15 maritime plans and 2 panoramas was given as the result of this campaign (Duplancic, 1999).

Figure 1. – A part of Beautemps Beaupré’s maritime atlas

During the campaign from 1822 to 1824 Austro – Hungarian Navy performed a systematic survey of the east coast of the Adriatic Sea. The result of campaign were 2 general charts, 22 sailing charts and 7 harbour panoramas. The Kingdom of Austria – Hungary established the first Hydrographic Office in Trieste during 1861, which moved to Pula the next year. Since then till nowadays hydrographic service has had almost a century and a half of continuous activity (with breaks during world wars) over the east coast of the Adriatic Sea. Over 1000 hydrographic originals on different scales in Gauss – Kruger’s projection according Bessel 1841 were produced during that period of time. Few hydrographic institutions could boast such a long hydrographic, oceanographic and cartographic activities. From 1991 hydrographic service take place in Hydrographic Institute of the Republic of Croatia (HHI).

3. HYDROGRAPHIC AND CARTOGRAPHIC ACTIVITIES NOWDAYS

Nowadays, hydrographic survey uses integrated hydrographic systems with DGPS positioning methods, which provides exceptional accuracy. Specific topography of the sea bottom is given by side-scan sonar image correction survey, as well as multibeam technique survey. Hydrographic information system – HIDRIS is used to

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index collect and process the collected hydrographic and oceanographic data. Hydrographic survey data are partly measured in the local coordinate system and partly in WGS 84. Solution for all cartographic activities, for civil maritime economy as well for the Croatian Navy, are certainly the existing interactive computer graphical systems within HHI, with modern hardware / software equipment. Wide cartographic data bank, analogue and digital, is the basis of cartographic information system – CIS. Croatian Hydrographic Institute covered the Adriatic Sea with 8 general charts (scales from 1:750 000 to 1:2 500 000), 30 sailing charts (scales from 1:150 000 to 1:300 000), 21 coastal charts (scales from 1:50 000 to 1:100 000), INT charts (scales 1 : 250 000) and various plans on scales 1:3 000 to 1:40 000 in several editions and publications. Apart from nautical charts this Institute provides numerous different informative, auxiliary and special thematic charts. Cartographic activity, apart from new analogue and digital chart production, proceeds in several mutually connected directions: transfer of all analogue charts to digital form, digitalization of hydrographic originals, cartographic data bank production, IHO’s recommendation standardization for display on charts, solving problem of incompatible nautical charts of HHI’s production towards the neighbouring countries nautical chart production. These are the significant steps nowadays.

4. MILITARY CARTOGRAPHIC SYSTEM PRODUCTION

C3I (Command Control Communication and Intelligence) system optimized a quantity of information on the Adriatic Sea area. The developed military hydrographic – cartographic information system essentially follows and processes battle activities, using the most advanced computer technologies (Horvat, 1992). At the moment, the production of this system is in the phase of initialization. New maritime navigation charts, publications and handbooks are produced in digital form together with formerly published nautical charts and publications, which are to be slightly modified and involved in the system. For military purposes HHI produces special charts in analogue and digital form, as well as various handbooks and other products such as: navigational handbooks, nautical – topographic charts scaled 1: 25 000, raster system of coastal, sailing charts, plans and sedimentologic charts of the Adriatic Sea.

4.1. NAVIGATIONAL HANDBOOK Navigational handbook is produced both in analogue and digital form. Handbook is the result of systematic data collection in the hydrographic, oceanographic and nautical survey, as well as cartographic processing of the collected data. Handbook displays all harbours and marinas, most of the bays, anchorages, quays, berthing facilities and other navigational objects within Croatian part of the Adriatic Sea. Besides graphical display handbook includes the data necessary to get the knowledge of local characteristics: orientation, meteorological data, currents, sea transparency, high and low waters, possibilities of entering ports, mooring and anchoring possibilities, repair service, coastline descriptions, offshore coastline description, strand- ing points, food and water supply, medical insurance, travel guide to neighbouring places, harbour installations (cranes, dock entrances, workshops), heights of quays and other data.

4.2. NAUTICAL – TOPOGRAPHIC CHARTS, SCALE 1 : 25 000 For areas of special navigational interest, HHI produces nautical – topographic charts, scaled 1:25 000. Detail display of land contents on these charts uses standards valid for topographic chart production, as well as symbols valid for topographic contents. Nautical contents are displayed with details according to rules that are

Charting and Navigation / Cartographie marine et navigation 12 Index valid for nautical chart production, with the symbols used on nautical charts. Till now, 12 of the planned 37 charts have been produced in analogue form (Figure 2). These nautical - topographic charts are produced on Gauss – Kruger’s projection, according to Bessel 1841 data. Formerly produced charts have been modified by overprinting the cartographic grid in magenta, according to WGS 84. New charts are produced in vector form with the cartographic grid within the state coordinate system and according to WGS 84.

Figure 2. – Part of maritime – topographic chart

4.3. SEDIMENTOLOGIC CHARTS The features and structure of the Adriatic seabed are not sufficiently displayed on sedimentologic charts as requested by economic, scientific, military and other reasons (Juracic, 1993). For the entire Adriatic Sea area, sedimentologic chart scaled 1 : 1 000 000 was produced as well as on four leafs scaled 1 : 750 000. For navigational areas of special interest, 10 charts scaled 1 : 100 000 and 1 : 25 000 are produced in Gauss – Kruger’s projection in analogue form (Figure 3).

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Figure 3. A part of sedimentologic chart

4.4. NAVIGATION CHARTS IN RASTER FORM (RCDS) PRODUCTION Due to limited funds for ECDIS production, as a transitional stage for military purposes a system of nautical charts in raster form (RCDS) is being developed. The existing INT charts, sailing charts scaled 1: 300 000, coastal charts scaled 1: 100 000 and 1: 50 000, sedimentologic charts scaled 1: 100 000 and 1: 25 000 as well as plans are scanned with 127 dpi resolution and geocoded, e.g. transferred from Mercator’s projection to Gauss – Kruger’s projection.

5. CONCLUSION

Hydrographic and cartographic activity in the Republic of Croatia has a long tradition. The existing nautical charts, publications and handbooks provide safe navigation and help in carrying out other activities within the Croatian part of the Adriatic Sea, also representing the base for the ECDIS military command system production, as well as for production of other projects. In the following period, full implementation of the standards in hydrographic and cartographic activities according to the IHO and IMO recommendations is planned.

Charting and Navigation / Cartographie marine et navigation 12 Index

REFERENCES

Canadian Hydrographic Service (1997): S - 57 ENC Product Specification, Version 1.1, CHS, Duplancic, T. (1999): Electronic charts in nautical cartography, M.Sc. Thesis, Geodetski fakultet Sveucilista u Zagrebu, Zagreb (in Croatian). Horvat, S. (1992): Digitalization of hidrographics originals as basis for formating digital cartographic data base, M.Sc. Thesis, Geodetski fakultet Sveucilista u Zagrebu, Zagreb (in Croatian). IHO (1990): Chart Specifications of the IHO and Regulations of the IHO for International (INT) Charts, International Hydrographic Organization, Monaco. IHO (1994): Hydrographic Dictionary – Special Publication No.32, V Edition, International Hydrographic Organiza- tion, Monaco. IHO (1996): Specifications for Chart Content and Display Aspects of ECDIS, International Hydrographic Organiza- tion, Monaco. IHO (1997): Glossary of ECDIS – Related Terms, III Edition, International Hydrographic Organization, Monaco. Juracic M. (1993): Prolegomena za izradu geoloskih karata dna Jadranskog mora, Vijesti Hrvatskog geološkog drustva, Vol. 30, No. 2, Zagreb. Vidovic, B. (1998): Automatization of activities in complex data bases, M.Sc. Thesis, Fakultet elektrotehnike i racunarstva Sveucilista u Zagrebu, (in Croatian).

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Session / Séance 06-A The use of global mathematical models in the cartography of marine sandbanks

Tom Vande Wiele RUMACOG - Research Unit for Marine and Coastal Geomorphology Universteit Gent Vakgroep Geografie Krijgslaan 281 - S8 9000 Gent Belgium E-mail : [email protected]

Abstract For the mapping of sandbanks we will make use of global mathematical models. The general models available are the trendsurfaces (algebraic) and the double Fourier series (trigonometric). The justification why to use these models falls within the framework of a general research on the influence of point patterns on different models for the representation of the relief of submarine sandbanks. Our hypotheses states that these global methods are less subject to the influence of certain point patterns with as result that these models can serve as a base for a further mapping. A second justification is that there is always a general trend in the data, which can be described by global models. For the interpretation of these mathematical methods we state that the deviations with respect to the model are significant and thus represent a local component. This gives the opportunity to map the differences between the data and the model with local methods, like the classical interpolation models. The research module now tries to determine to which extent the models can be used for mapping, whereby the influence of certain settings are examined. By settings we not just mean parameters as the power of the equation (in the case of trend surfaces) or the number of terms in the series or the fundamental wavelengths (in the case of double Fourier series), but also the orientation of the axis and the size of the research area (this with respect to the possible edge effects). One of the tools we will use to test the effectiveness of the model is the semivariogram. Thereby we foresee the possible use of kriging as the local interpolation technique. When the global methods enable us to predict the trend, then a simplification of the semivariogram (with zonal and geometric anisotropy) must be the result. This was one of the obstacles to use the kriging technique for mapping the submarine sandbanks.

Double Fourier Series

A function can be approximated by an infinite trigonometric series. If we consider the relief as an unambigu- ous finite continuous and periodic function of x and y, then the following equation can be used for the representation of that relief.

Charting and Navigation / Cartographie marine et navigation 12 Index

∞ ∞  2nxπ   2myπ  Z = ∑ ∑α cos i  cos j  ij nm  λ  λ n=1m=1 12  ∞ ∞  2nxπ   2myπ  + ∑ ∑ β cos i  sin j  nm  λ  λ n=1m=1 12  ∞ ∞  2nxπ   2myπ  + ∑ ∑γ sin i  cos j  nm  λ  λ n=1m=1 12  ∞ ∞  2nxπ   2myπ  + ∑ ∑δ sin i  sin j  nm  λ  λ n=1m=1 12  λ λ α β The unknown parameters are the fundamental wavelengths 1 and 2, together with the coefficients nm, nm, γ δ nm and nm. To put this theory into practice, some simplifications and approximations are necessary. This means among other things that the fundamental wavelengths are approximated by the distances L1 and L2, which makes it possible to avoid the repeating behavior by choosing these distances greater than the selected area of study. A second, obvious, approximation concerns the number of terms of the series. This has a direct effect on the necessary computing time. Two important questions arise now. First, how do we adapt our data to the equation, and secondly how do we interpret the deviations from the calculated equation ? Concerning the first question, we will make use of the least square method, which states that the square deviations must be minimal. As for the interpretation, we consider that the deviations from the equation are composed of a significant local component and a small error component. =++() ε zfxyLiTiiii,

The data.

The area of study is situated before the Belgian coast. It is part of the Kwintebank, which forms part of the Flemish Banks. The sandbank is characterized by his asymmetric form, steepest slope faced NW, and the occurrence of sandwaves and megaripples. The data acquisition is performed with the research vessel Belgica by making use of the Sercell DGPS positioning system and the Deso 20 echosounder in combination with the TSS heave compensator. The data itself exists out of 60 SW - NE oriented tracks and 30 NW - SE oriented tracks. The distance between the tracks is about 50 m while the distance between points on the track amounts to about 2.5 m. Besides this data set for the construction of the model, we also have a data set for testing the model performances. This data set is obtained during a sediment survey in the same area. Because a sediment survey is performed at a much lower speed the distance between two successive data points is only 0.9 m. Out of the first data set, different patterns can be derived. The first parameter we will make use of to distinct the different patterns, is the orientation. In this way we can derive two main patterns, according to the direction of the tracks sailed and a third one as a combination of the two preceding patterns. The second parameter that can be used is the distance between the tracks, this is achieved by a systematic reduction of tracks in the pattern.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Hypothesis and method.

The parameters we want to examine with regard to the Double Fourier Series model, consists of model proper-

ties (like n, m, L1 and L2) as well as the properties of the data set used for the calculation of the best fitting equation.

Figure 1. Hypothesis. The diagram shows a test method and a test set. The latter we described already in the previous paragraph. The test method will now compare the values predicted by the model with the observed values in the test set. The predicted values are acquired by importing the x, y - coordinates into the model.

Results and analyses.

The first results include a comparison between the different patterns and this in function of the number of Fourier coefficients. To get an overall impression on the model performances we have chosen to use Pearson’s product moment correlation. This coeffi- cient gives the correlation between the predicted values by the model and the observed values. This first graph compares 3 patterns (differing in distance be- Figure 2. r-correlation pattern 1. tween tracks) based on SW - NE oriented tracks. As we expected, the use of more terms leads to a better performance. And secondly there seems no influence of the pattern on the results. These conclusions can also been drawn for the other oriented patterns. This could lead to a rather rash conclusion about the influence of the point pattern. The influence of the patterns is not shown as a model improvement but acts on the stability of the model. This can easily be proven by the next diagram which shows the same results, but now with more (10 x 10) terms used.

Charting and Navigation / Cartographie marine et navigation 12 Index

Here we can see when the model becomes unstable. This phenomenon is related to the distance between the tracks, the greater this distance is, the sooner the instability occurs. We can also remark that combining the two track directions results in a positive effect on the stability of the model. Like I mentioned in the beginning of this paper, we have to do with a best fitting equation or surface. This means that the performance can’t decrease with an increasing number of terms. Figure 3. r-correlation all patterns. Why this occurs anyway has to do with the test set used. If we let the model predict the data that has been used for the calculation of the equation (the construction set) and plot these results together with those obtained by the test set, then the difference between both results must be clear. The explanation of these differences can be shown by means of a map that plots these deviations as they occur in space. The map also shows the relief and the data set used. The relation between the data Figure 4. Influence test set. set used and the results obtained must be very clear now. The largest deviations occur precisely between the meshes of the data net. It is also remarkable that they occur at the edges of the area and are successively positive and negative. The next step in our research concerns the influence of the point density in the pattern. We have chosen to perform the calculations on the pattern that consists of both directions and with the smallest distance between the track. The results obtained with 100%, 50%, 25%, 10%, 5% and 1% of the total data set are shown in the next graph. The use of even 5% of the available data doesn’t result in a performance decrease of the model. However when we make use of only 1% of the data, the model becomes unstable. The big advantage of low percentages of data expresses itself in shorter calculation times.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Figure 5. Deviations map - lt4.

Figure 6. Influence of point density on track.

Charting and Navigation / Cartographie marine et navigation 12 Index

Conclusion. Table 1. Parameters model 10x10 lt1. Mean D 0.20 About the use of the model we have made the following conclusions. Fist Variance D 0.36 we will consider the model for a direct mapping. The most important Min D -1.80 factor is then the accuracy of the mapping that can be achieved with the Max D 1.35 model. MAE 0.3328 RMSE 0.4161 For the model based on the total data set and with the settings : m = 10, n RMSEs 0.2039

= 10, L1 = 3500 and L2 = 3000, the results can be found in the preceding RMSEu 0.3627 table. Important are of course the extreme deviations, which are in this a 0.1611 case -1.80 m and 1.35 m. In spite of the very high correlation r = 0.9972, b 0.9973 these deviations are not to tolerate, and we are interested were these de- r 0.9972 d 0.9982 viations occur. Therefore we mapped these deviations together with the relief, as can be seen in the following figure. This map shows that there is an explanation for the larger deviations. They are as it happens related to the relief, and more particularly with the occurrence of sandwaves on the sand bank. With the exception of these structures (the sandwaves) the model is capable to restrict the deviations to the interval [-0.20, +0.20]. From which we must conclude that the model is able to map the general structure of the bank, but not the smaller structures that occur on the sandbank. This leads us to a next step, were we see the model as a tool for a further and more accurate mapping by means of local interpolation methods like kriging.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Figure 7. Deviations map - lt1.

References.

Bracewell, R. (1965) The Fourier Transform and Its Applications. McGraw-Hill. Davis, J. (1986) Statistics and data analysis in geology. J. Wiley, New York. Hwei P. Hsu. (1967) Fourier Analysis. Simon and Schuster, New York. Kreyszig, E. (1988) Advanced engineering mathematics. John Wiley & Sons. Tokstov, G. P. (1962) Fourier Series. Prentice-Hall, Englewood Cliffs, New Jersey. Willmott, C. (1984) On the evaluation of model performance in physical geography. Spatial statistics and models. D. Reidel publishing company, p443-460.

Charting and Navigation / Cartographie marine et navigation 12 Index

Session / Séance C6-C Making practical and effective electronic aeronautical charts

Sonia Rivest DMR Consulting Group Inc. [email protected]

Rupert Brooks Canada Centre for Remote Sensing, Geomatics Canada [email protected]

Bob Johnson Aeronautical and Technical Services, Geomatics Canada [email protected]

Abstract In order to take advantage of the opportunities presented by new display technologies, Aeronautical and Technical Services (ATS) was involved in research to determine the best way to reproduce and distribute digital charts which are based on the current paper charts produced by ATS. These charts had to be designed to meet many difficult constraints, while still maintaining the overall goals of clarity, accuracy and aesthetics. Safety and integrity of data is a major concern when producing aeronautical charts. Not only does the electronic version have to contain all the same information as the paper chart, it must also be easily perceived by the end user to be of the same quality and authority as the paper version. The best way to accomplish this is to make the electronic chart look as much as possible like the paper chart. The resulting charts had to be transferable across a network or by CDRom media with only minor modifications, and be compatible with all major computing platforms. At the same time, software and hardware costs needed to be kept to a minimum, particularly for the client viewing the chart. Finally, the disruption to the existing paper chart production environment needed to be minimised. Many different software packages can be used to produce charts to be visualised on a screen. The choice is based on many criteria like the users’ needs and the existing chart production process. At Aeronautical and Technical Services, Adobe’s portable document format (PDF) was chosen and a combination of software was used to transform the digital files used for the paper chart production to PDF digital files. The transformation process was designed in order to fulfil all the requirements described previously. The process ensures a consistency between the paper and the electronic charts because they are produced from the same digital files. In addition to overcoming the cartographic challenges of a screen medium as opposed to a paper medium, the new electronic charting technology can enhance the capabilities of a chart. Elements on charts were linked to other charts, publications and databases, using the interactive capabilities of the PDF. This interactivity greatly enhances the usefulness of a chart, by placing more information at the user’s fingertips more quickly. Creating a system to provide this information reliably in the face of frequent changes requires a fundamental understanding of the technology behind the scenes. This paper will describe the work that is ongoing in the Aeronautical Charting group in this area, and will discuss the achievements to date. How a screen medium differs from a paper medium, how the electronic aeronautical charts have been designed and how digital charts can exploit their connections to the database both on and off the Internet will be discussed. Charting and Navigation / Cartographie marine et navigation 12 Index

Introduction

In 1997, Aeronautical and Technical Services (ATS) began actively researching and developing technology for electronic distribution of aeronautical charts. ATS produces many different types of aeronautical charts and documents but the research focused on the Enroute charts, the Canada Flight Supplement (CFS) and the Canada Air Pilot (CAP) which are actually being produced digitally. The Enroute charts are used for instrument flying. They cover large areas, and show little to no topographic base information, but show copious amounts of airspace, navigational aid beacon and airport information. The CFS is a handbook describing all the airfields in Canada, with the information that a pilot might need. The CAP describes approaches to each airport. There are corresponding military publications for each of these that differ slightly in form and contain a bit more information. All information was put online in this project, but we will continue to refer to these publications by their common, civilian names. This paper first presents the fundamental differences between paper and digital charts. The second section explains the specific requirements that the new digital documents would need to fulfil. Then, details about the technology available during the research are given. The final section presents the process developed and implemented at ATS.

Screen vs. Paper: What is the digital medium?

Like many other aspects of communication, the art and science of cartography has been revolutionised by the use of computers. Maps are not only produced and analysed on computers, but they are increasingly being viewed on computers, and it is not uncommon for a map product to be heavily used yet never put on paper. It is clear that the production of a map for computer use is not as simple as just displaying a picture of the paper map on a computer screen, but how it should be done is still not well understood [Petersen, 1995]. This change in the medium of expression has led inevitably to a change in the methods of using the map [McLuhan, 1964]. Despite this change, the ultimate purpose of cartography remains unchanged. The intent is still to communicate information about a spatially defined set of information to the map user. Furthermore, maps have traditionally been documents to be studied intensely with the user learning more the longer they spend examining it. This type of use is still valid, although the way the user interacts with a map has changed greatly from the paper to the screen. Because of differences in the capabilities and methods of using a map on the screen, there are differences in the effective cartographic techniques for each medium. The screen-based map has forgone the high resolution of paper, but has gained instead the potential of interactivity. Although computer screens are being improved all the time, their effective resolution is still only a fraction of the detail that can be shown on a piece of paper. Furthermore, the average computer screen is between 15 and 21 inches on the diagonal, while paper maps are usually on the order of 2 to 3 times that size. Fortunately, the computer gives the capability of zooming in, and changing the level of detail if necessary. The user can also make queries of objects on the map, which changes greatly the need for labelling. The capabilities of zooming and querying change the whole process of cartography. Scale – once a defining characteristic of the paper map – has now become arbitrary, and a whole host of information that used to be implied by scale (such as accuracy) now must be made explicit. Furthermore, the requirements of generalisation have changed. In the paper map, generalisation was driven by the requirement of readability, printability and the desire to create an accurate summary. On the screen, generalisation is still related to readability, but in an entirely different fashion. The printability criterion has been replaced by a need to conserve bandwidth. Nevertheless, the final criterion, which is to create an accurate summary, remains the same. To add further difficulty to the cartographer’s task, it is often required to produce paper and digital versions of effectively the same map from the same source information. While the requirements of the end products are

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

quite different, it is necessary to use the same processes wherever possible to minimise error, and to enhance efficiency. Also while it might be ideal for technical reasons to completely change the appearance of the chart on the screen from the one on the paper, this would not be practical for most users. The average user of aeronautical charts has been trained on paper charts, and if the differences between paper and screen are too radical, they will lose confidence in the chart, or worse still, be unable to interpret it quickly and correctly.

Requirements at ATS

When publishing to a digital medium, the potential seems limitless. However, there are a number of practical issues that should guide any design. One of the most important is the clients intended use, and the machines and software that they use. In the case under discussion, however, it was found that the possible client platforms were diverse. The intended platforms for use of digital aeronautical charts ranged from high-powered UNIX workstations isolated from external networks, to Web-aware desktop PC’s. Furthermore, although it was not explicitly part of the design, some thought was given to in-flight displays. ATS believed that whatever technology was chosen should not lock out that possibility. This range of target systems had a profound influence on the technology chosen. Secondly, ATS produces different types of aeronautical products designed for different users. Each product necessitates a particular production process. Since 1994, the Enroute charts have been digitally produced on a 56-day cycle basis, in accordance with the International Civil Aviation Organisation (ICAO) publication cycle. The effective date of a chart is fixed and the deadline cannot be moved. During the first 28 days of every cycle, changes in the aeronautical information, provided by Transport Canada, NAV Canada and the Department of National Defence, are validated and included in the Canadian Aeronautical Charting (CANAC) database. The changes are then extracted and symbolised in digital vector files using a suite of software designed exclusively for aeronautical charting. A manual cleaning is done on every vector file in order to integrate the changes to the previous version of the chart. Only the changes are extracted from cycle to cycle in order to reduce the cleaning operation. The aeronautical charting software then combines the vector files and transforms them to raster files to produce the negatives. Once these are printed, the paper charts are finally ready to be distributed. The CAP and CFS are also revised and published every 56 days. Because the time requirements for the production of the paper documents (charts and books) are very tight, the new digital document production process would have to be integrated with the existing process as much as possible. This would help minimise the production costs and would require minimal extra effort from the product specialists. Also, in order to make sure that the digital documents contain the same information as the paper documents, both would have to be produced from the same source of information. This is a very important safety factor. As in any project, cost was an important requirement. There are three main sources of cost in digital publication. These are the production of the data and software, licensing fees, and support costs. As the data was already being produced, the costs were managed by integrating the digital production as tightly as possible to the existing paper process. A COTS (Commercial Off The Shelf) software solution was preferred for cost- effectiveness and speed of implementation. When licensing software products for publishing, there are two major ways of applying the costs. One method requires a large outlay for software that will be used to produce the media, but special viewing software, if needed at all, is free. Thus there is no per user seat licensing cost with such a system. Because ATS expected a large client base, this type of solution was considered the most appropriate. Alternatively, there are systems with lower up front cost, but which require a license fee or royalty to be paid for each user who ends up using the client software. Clearly, as the number of users increases, this becomes a less appropriate option. Finally, the support costs may be extremely difficult to quantify. These may include opportunity costs of lost business due to product failure.

Charting and Navigation / Cartographie marine et navigation 12 Index

Also, the design of the new digital documents would have to take into account the need for clarity. This could be implemented by using a level of detail appropriate for the zoom level. On a computer screen, it is not possible to show everything that is on the paper chart at the same time, so it is important to allow the users to choose the information they need or want to see. It is also important that the users feel comfortable with the new charts. These should then have the same or similar colours and symbols as the paper documents. Finally, the digital aeronautical document production process would have to be adaptable to other ATS products like the Air Traffic Control (ATC) charts. These ATC charts are produced on demand and contain a subset of the elements found on the Enroute charts.

Investigation of Available Technology

The World Wide Web (WWW) has shown tremendous success in the realm of electronically distributing information. It has grown astronomically in the short time since its development. Working under the assumption that the WWW must be doing something right, ATS had decided to investigate “web-enabled” technologies for use in publishing these digital products. As research progressed, it became apparent that there were a number of related technologies involved in the web’s success, and they shared several attributes that contributed to the web’s effectiveness. By understanding and harnessing those attributes, ATS has been able to develop a very flexible and expandable process for publishing electronic aeronautical charts. In common usage, the term “the web” refers to a vague collection of technologies. Some of these technologies are protocols, some are languages, some data formats, and some are proprietary software. There were two important and independent aspects of these technologies that related to the core of the problem of producing digital aeronautical documents. One was the use of open, widely understood data formats, and the other was the use of client-server computing. Both of these concepts are well tested in practice, and ATS has woven these ideas into the core of the digital chart design.

Open, Widely Understood Data Formats There are a number of data formats, for different types of data, which are closely associated with the Web. Examples include hypertext markup language (HTML) [Raggett et al., 1998], joint photographic experts group image format (JPEG) [CCITT, 1992], and ECMAScript (better known as JavaScript or Jscript) [ECMA, 1997]. These examples run the gamut from describing formats for rich text, to programming languages, but they share a number of common attributes. · Client-Side platform independence Notwithstanding the efforts of many software companies, there are many different computer hardware systems, operating systems, and viewing software. As mentioned, the potential clients for any digital electronic charts would have a diverse number of different platforms. The exact details of how a programming or scripting language can be platform independent are beyond the scope of this paper. In general, it is accom- plished through the use of an abstract virtual machine to isolate the programming environment from the diverse hardware on which it may exist. More detail may be found in McComb et al. [1997] and ECMA [1997]. The popularity of these formats is attributable to the availability of software to work on them on any platform. · Open standards The standards for each of these formats is publicly available to any interested party, and may be freely used without patent or copyright infringement. In part, this enables client-side platform independence, because any company or individual can create or adapt software to work with these formats. Furthermore, the open-

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

ness of the standard allows experts to evaluate and comment on the technical details, which can reveal flaws, or provide insight. · Inexpensive viewing software The openness of the standards involved encourages the proliferation of software to deal with information in that format. In each case, either through competition, or the explicit creation by promoters of the format, tools to view and work with the format are widely available. · Wide usage Partly because of the client-side platform independence, but mainly because of the availability of inexpensive software, these formats are in wide use. This wide use means that the overhead of maintaining the viewing software becomes very low, because of familiarity, and because of its use for many purposes. Aeronautical charts contain symbolised geographic objects, which are described in a very abstract way. These are ideally represented as a vector GIS dataset. To date, however, the web formats do not include any vector format. This greatly limited the choice of formats available to ATS to use for digital aeronautical charts. In all cases, some software that was not, strictly speaking, a web standard, would have to be used. Any format must be supported on the machine used to view it by some software. It was certainly not reasonable to ask all ATS clients to support an entire GIS system just to view digital aeronautical charts. Therefore some vector format which met the criteria listed above was required to allow keeping all the advantages of the client-server computing involved in the web, while using the advantages of the format. There has been much excitement about using plugin components that understand GIS type data to view spatial datasets in a web browser environment. Because of ATS requirement for extremely fast production time, this approach showed great promise. The transformation from the GIS data formats used internally in ATS were not too severe, and were easily automated. In fact, the Intergraph GeoMedia Web Map Version 1 plugin system was used during the first phase of the project. However, none of the plugin systems available from the major GIS / drafting vendors has cross-platform support at this time and all have only limited graphics flexibility [Limp, 1997]. (Note that Autodesk has very recently added Java-based cross-platform compatibility to their system.) This severely weakened the viability of this approach for ATS, as the client base was known to use a wide variety of machines and operating systems. There are XML based vector graphic being developed currently, which show great promise, but are not yet ready for implementation. There was only the Adobe PDF format available which possessed the attributes considered necessary to produce an effective product.

Portable Document Format (PDF) The Portable Document Format (PDF) was developed by Adobe Systems Inc. to provide a means of displaying graphically rich, page-based information. Although originally not intended to support maps, the paradigm used by PDF easily adapted to this use. PDF is a page description language which uses a subset of the Post- script drawing engine to produce graphical representations on screen which maintain the layout, colour and graphic design of the original document [Adobe Systems Inc., 1997]. The format is open for public review, and the viewing software is free, and available on a wide variety of platforms. Due to its use of the PostScript drawing model, PDF presents very high quality graphics and text and is very flexible on the colours and fonts side. It is a hybrid raster - vector format, and its support of vector information allowed for effective production of small files. Every single element of the document is described separately in the file so it is possible to manipulate it, for example to create customised links. Finally, the format supports the description of http based hyperlinking, and the use of ECMAScript / JavaScript to provide custom interactivity. This means that the use of this format does not in anyway preclude the use of client-server computing or possible future development.

Charting and Navigation / Cartographie marine et navigation 12 Index

Client-Server Computing The client-server model of computing involves the division of computing processes into client and server processes. Server processes listen for and respond to requests from the client processes. A client and a server process may reside on the same machine, or on greatly separated machines connected by a network [Taylor, 1997]. The use of client-server processing by the WWW involves the use of the HTTP protocol to request service from servers that may be widely separated from the client processes [McComb et al, 1997]. This is completely independent of the use of the data formats described in the preceding section. The HTTP protocol can be used to distribute any type of data, and there is no requirement to use a standard web port (80) or to use a web browser to make requests [Fielding et al, 1997]. The primary advantage of client server computing is that it allows an author to create a division of labour between client machines, which can be extremely diverse, and server machines, which the author may have full control of. The alternative is to have all required software running on and developed for, the end user’s machine. When the end user may be using any one of a diverse variety of platforms, developing custom software to run on all of them becomes a tremendous chore. Nevertheless there are advantages and disadvantages to both methods. Having all the software located on the end users machine has two major advantages. The resulting product will be self-contained, and independent of any network connections. Furthermore, once the product is produced it requires no further effort. It can be stored and distributed on a fixed medium, and will continue to work even if the producer is not actively supporting the machines involved. The disadvantages of this method, however, can be severe. ATS was in a situation where the potential users operated a wide variety of systems, and developing custom software for all of these systems would have been prohibitively expensive. This meant that for this solution to work, all the desired functionality would have to be available in a COTS solution. Using a client-server approach provides the major advantage that most or all custom programming can be done on the server side of the system. This means that the programs can be written for a very specific machine, or set of machines, which the developer has access to and control over. Furthermore, it may be easier to restrict access to the server to authorised users. However, the convenience of this server has a price. It is up to the publisher to maintain it, and keep it running and up to date. Clearly, using a server only becomes an advantage when custom programming is required. Custom programming will be required when the author desires functions from the published product which are not explicitly available in viewers for the format, and cannot be precalculated and prepared in advance. For example, web browsers support the viewing of images, but do not support zooming. For an author to support zooming into an image in this environment, they must either have a program that will do the zooming, or they must pregenerate all possible zoom levels, and generate the links to them within the data. Some types of interactivity, therefore, require custom programming and can add greatly to the cost and difficulty of a project. When planning interactivity in a digital publication, it is critical to differentiate between interactivity that is implied by the medium, and is therefore packaged for free with existing software viewers, and interactivity that must be programmed by the author. The differences are subtle, because some of the formats being discussed are actually programming languages that may have a great deal of functionality. The key to making this differentiation lies in the level of abstraction of the language, and how successfully platform details have been hidden from it. Viewing, zooming and panning and linking is available with the software for any of the formats that ATS was considering. The capability of querying was the area where issues of customisation came to the forefront. There are two approaches to querying. In one, all query results are precalculated and treated as links. This type of interactivity is still quite simple to implement. However, it requires the author to imagine all possible user queries in advance, which may not be possible. Even when all possible queries can be accurately imagined, the

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index implementation of such a plan may be impossible because the number of possible queries (and therefore answers which must be precomputed) may quickly become astronomical. The other approach is to be able to specify a query that requires processed at run-time. This query could be either spatial or non-spatial in nature. For this type of interactivity some programming must be done. The processing can be done either at the client end or the server end or with some combination thereof. On the whole, the choice of whether or not to use a client-server approach depends on what types of queries would be supported, and to what degree the responses to them can be precalculated.

Experimentation, Development and Implementation

ATS possesses a large database of information about airports and other objects on the aeronautical charts. In addition, two publications, the CFS and the CAP encapsulate the information about airports in a book format. ATS was able to experiment with both a database query approach, and a pregenerated approach.. In the database query approach, the database was loaded onto a machine with an http server, and Common Gateway Interface (CGI) programs were used to provide a query service. By passing the name of an airport, for instance, to the CGI process, the database would be queried, and relevant information formatted and returned. The advantages of this approach were that it was low in cost, as the database was already being maintained, and it was quick and effective. When using a plugin based approach, the links to the database key were already present because the geographic data was already linked to the database. It was also possible to use the precalculated approach because of the existence of the CAP and CFS publications. The individual pages of the CAP and CFS books were already prepared in postscript format for the press. It was a simple matter to convert these to PDF format for online viewing. As the CAP/CFS pages were already identified by airport, it was fairly simple to establish the link to them from the database. As the database was associated with the plugin-based chart, this provided a nice binding between the different products and information available, and placed all information that a user might desire within one or two mouse clicks. In the beginning of the research, a plugin approach was used, and this required a server process to be running continually. This prevented the precalculated queries from reaching their full effectiveness at this stage. The plugin approach, however, was seriously flawed by its lack of graphic flexibility, its platform dependence, and its requirement for a server process even for precalculated responses. For that reason, ATS chose to look at the PDF option for publishing digital charts. This option resolved both the platform dependence and cartographic appearance issues, but as a visual representation was no longer directly linked to the database. It was necessary to re-establish the link between the chart, and the database. In the beginning, this appeared to be a technical headache to overcome, but it allowed even further flexibility in the long run. It is possible to define a link within the local filesystem in the PDF file, and this enabled the system to be free of its network dependence for those queries that had precalculated answers. Because the PDF format supported HTTP based hyperlinking, using PDF did not rule out using client-server based queries. As the necessary information to create these links was completely represented in the database already, it was only necessary to generate the link information for each chart, and to insert that link into the PDF file. Based on this choice, an appropriate process had to be designed to transform the three types of documents under study into the PDF format. The actual paper document production processes used at ATS facilitated the development of the new process. For the Enroute charts, the software that actually produces the digital vector files during the extraction and symbolisation has the capability of plotting files in postscript. Having documents in this format, it is very simple to produce the corresponding PDF files. Also, the CAP and CFS are both produced directly in the postscript format.

Charting and Navigation / Cartographie marine et navigation 12 Index

Figure 1: Web browser interface

The final prototype system that has been developed allows PDF charts to be linked as stand-alone documents, or to be viewed through a web browser interface. The web browser interface also allows the use of the Intergraph Geomedia Web Map plugin for viewing data. Many queries can be handled in a precalculated way, but if further sophistication is required, there is no reason why a server cannot be used. The web browser interface is shown in Figure 1.

Digital Enroute Charts production process In order to minimise the duplication of efforts, the transformation process starts with the cleaned vector files. A copy of these files is transferred to the digital charts production process while the original paper chart production process remains unchanged. The cleaned files are then customised prior to the postscript plotting. This customisation step is necessary because the colours and symbols originally used for the paper charts production process need to be changed to reflect the printed paper charts. The postscript files are then produced, transformed into PDF files and an application is run to generate the links to the database. The resulting PDF files can also be linked together and to other digital publications. This digital Enroute charts production process can be completely automated and can be adapted to produce the new digital CAP and CFS as well. Figure 2 shows the integration of the digital charts production process within the paper charts production process. The two processes can work in parallel.

Design details The design of the visual representation of the new digital Enroute charts addresses many aspects. The main ones are the colours and the levels of details. The choice of colours was primarily based on the colour standards actually used for the paper charts. Each colour has a particular signification and it was important to keep this association when designing the new digital charts, so these are easy to read by current paper charts users. On the paper charts, aeronautical information is coded in shades of green and grey. The same colours have

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index been used on the digital charts. The important symbols like the airports and navigational aids have been made brighter so they are easily distinguishable from the background [Kaufmann, 1987]. Also, a light background has been used as it facilitates the colour perception of symbols [Eaton, 1993]. In order to make the chart clearer, three levels of details have been defined. The first one contains only the features needed to locate the area of interest. The second level adds the aeronautical information symbols and the last one includes all the text. Figure 3 shows an example of PDF chart.

Figure 2. Actual Enroute charts production process

Figure 3. Sample of an Enroute paper chart (left) and of an Enroute digital chart (right)

Charting and Navigation / Cartographie marine et navigation 12 Index

Conclusion

After investigation and experimentation, the PDF format appeared to fulfil all the requirements of the Aero- nautical and Technical Services for the online publication of aeronautical information. The PDF format is platform independent, it allows high quality graphics and text as it is derived from the postscript and the viewer is free. This format has the ability to contain links and because of its support of the JavaScript / ECMAScript language, it allows the creation of customised interactive functions. The process developed at ATS integrates into the existing paper document production process. The digital documents are produced from the same digital files as the paper products, so they contain the same information. This guarantees the integrity and consistency of the aeronautical information. This process is flexible and can be adaptable to other charts and publications that ATS produces with minor modifications. A prototype system has been developed that allows the charts and publications to be linked together as a stand-alone product with some query capability, or to be integrated into a web browser environment using server processes to do more sophisticated queries.

References

Adobe Systems Inc. (1990). Postscript Languages Reference Manual 2nd Edition. Addison Wesley. Adobe Systems Inc. (1997). Portable Document Format Reference Manual, Version 1.2. Available from www.adobe.com. CCITT, (1992). Information Technology – Digital Compression And Coding Of Continuous-Tone Still Images – Re- quirements And Guidelines. CCITT (the International Telegraph and Telephone Consultative Committee) recomendation T.81. Also known as ISO/IEC International Standard 10918-1. Commonly known as JPEG image format specification. Eaton, R.M. (1993). Designing the electronic chart display. The Cartographic Journal, 30, 184-187. ECMA, (1997). ECMAScript Language Specification. ECMA - European association for standardizing information and communication systems. http://www.ecma.ch Fielding, R., Gettys, J., Mogul, J., Frystyk, H., Berners-Lee, T. (1997). RFC2068 Hypertext Transfer Protocol — HTTP/ 1.1. http://www.cis.ohio-state.edu/htbin/rfc/rfc2068.html. Kaufmann, R. (1987). Colour considerations for electronic charts. Technical Memorandum, Department of National Defense, Canada. Limp, W. F. (1997). Weave Maps Across the Web. GIS World. September, 1997 pp. 46-55. McComb, G., Bower, M., Robinson, M. (1997). Web Programming Languages Sourcebook. John Wiley & Sons. McLuhan, M. (1964). Understanding Media: The extensions of man. McGraw-Hill, Toronto. Netscape, Inc. (1997). Netscape DevEdge Library Documentation. http://developer.netscape.com/library/documentation/index.html Peterson, Michael P. (1995). Interactive and animated cartography. Prentice Hall, Englewood Cliffs, N.J. ; Toronto Raggett, D., Le Hors, A., Jacobs, I. (Eds.)(1998). HTML 4.0 Specification. World Wide Web Consortium recommendation. http://www.w3.org/TR/REC-html40/ Taylor, L. (1997). Client / Server Frequently Asked Questions. http://www.abs.net/~lloyd/csfaq.txt

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

Session / Séance 06-D Environmental Mapping of Russia’s Seas Using GIS

I. Suetova and L. Ushakova Dept. of Cartography and Geoinformatics, Faculty of Geography, Moscow State University, Vorob’evy Gory, Moscow, 119899 Russia Phone: 7(095) 939 37 93 Fax: 7(095) 932 88 36 E-mail: [email protected]

Preface

The intensive anthropogenic impact on Russia’s inland and adjacent seas in the process of utilization of their resources in recent decades resulted in the pollution of some of their areas, disturbance of the entire complex of natural conditions, and decrease in the natural ability of sea ecosystems for autopurification. The information system for observation and analysis of the state of natural environment, first of all, of the pollution and its effect in the biosphere, brought the investigation of natural phenomena to a new level. The study of the ecological results of the human economic activities at the seas takes two forms. One of these is the geochemical monitoring of sea environment, which involves the monitoring of abiotic factors including the hydrographic ones such as the temperature, salinity and pH value of sea water, the biogenic elements such as oxygen, nitrogen, phosphates and silica, and the monitoring of the factors of anthropogenic impact such as the spreading of chlorinated hydrocarbons, pesticides, petroleum hydrocarbons, heavy metals. The other form is the biological monitoring of sea environment, i.e., the analysis of the behavior of sea ecosystems under these conditions. The inadequate amounts of available information add to the importance of monitoring and of the accumulated data base from the standpoint of analysis of both the natural processes, such as the seasonal and annual changes in sea ecosystems and hydrodynamic and climatic variations, and the anthropogenic impact on the sea surfaces located in different latitudes. Because the local pollution and its negative results at the seas may have a large-scale and even global effect, GIS mapping of the seas, which enables one to take into account the natural relationship between elements of sea ecosystems and the dynamics of natural phenomena, assumes special importance in developing the system of integrated ecological monitoring.

Mapping of the Active Sea Zones

The results of numerous investigations of recent years demonstrate that the sea pollutants come from atmospheric precipitation and river and terrigenous runoffs. Along with large masses of industrial and sanitary sewage that flow to the sea year after year without any purification, major sources of pollution include irrigation systems, agricultural washoff, coastal and offshore oil and gas fields, water engineering, and emergency dumping of oil from tankers and pipelines. The seas are polluted the year round with harmful pollutants such as oil, phenols, detergents, heavy metals, biogenic elements.

Charting and Navigation / Cartographie marine et navigation 12 Index

The pollutants are distributed nonuniformly in the seas and form high-level zones in the top euphotic layer, in littoral areas, and in the regions of convergence of water masses, that is, in the ecologically important sea biotopes in which the bulk of biological products are created. In mapping inland and adjacent seas, special attention should be given to the active zones, i.e., places of higher intensity of geographic processes, in view of the fact that the intensity of the processes corresponds to the intensity of transformation of matter and energy. The zones of higher transformation of matter and energy as a result of hydrodynamic, physical, chemical, and biological processes usually emerge at places of intersection of several boundary surfaces. These are coastal zones such as the ice edge (Arctic seas) or frontal zones such as places where water masses of different origins and different characteristics meet. The regions of development of the most active hydrodynamic and thermo- dynamic processes in the atmosphere and on the see surface include submerged springs, volcanoes and wells, river mouths, canyons, straits. They serve as channels for intensive exchange of matter between the sea and deep-lying layers of lithosphere (for example, emergency dumping of oil in the Caspian Sea may cause local ecological shocks). River mouths are places in which slow global cycles of matter accelerate and the concentration of matter occurs, and where the role of industrial waste is rather important. The estuaries and coastal sea waters are the most fertile places in the world. In the canyons, matter is transported from the coastal surface zone to the deep sea regions. The dynamics of the coastal zone, the shore destruction or accretion, the migration of river mouths, the balance of pollutants depend in places on the magnitude of this runoff. Straits and submerged rapids are of unique importance to the circulation of sea water, redistribution of heat, slats and dissolved gases.

Description of Environmental and Anthropogenic Factors

The following environmental and anthropogenic factors have been treated in mapping Russia’s seas. - The main oceanographic features: the temperature, salinity, and density of water of the surface sea layers, as well as dissolved oxygen that plays an important part in the distribution of biomass and primary productivity of the seas being explored. - The oceanologic conditions such as sea currents, circulation of water, hydrofronts of the upwelling zone. The pollutants coming to the shelf are redistributed there and then carried out to the sea. The sea circulation and hydrofronts of the upwelling zone play an important part in the transfer of pollutants. - The biogenic elements that form the basis of mineral nutrition of algae in the process of photosynthesis. These are phosphates, total phosphorus, nitrogen (nitrite, ammonia, total), and silicon. - The coastal processes responsible for the delivery of material to the littoral areas; special attention is given to shoal, where secondary pollution takes place as a result of the wave and wind activities. - The bottom sediments form the zone of special critical intensity, in which the coefficients of accumulation of pollutants are much higher than their concentration in surrounding water. Most of pollutants, such as heavy metals, molecularly stable chemical compounds, are sorted in the mass of water on settling particles, pass through this mass, accumulate in the bottom sediments and in benthos. - Atmospheric circulation, especially, on the land-sea boundary. The contribution by atmospheric pollutants to the overall balance of anthropogenic pollution of the sea environment is compared at present with river discharge. Higher concentrations of pollutants, involved in biological cycles, have been discovered in surface water and, especially, in the surface microplankton. - The biological sea and coastal resources as a measure of stability of sea ecosystems. The correlation between the ecosystem biomass and primary productivity is calculated by the actual data on a unified scale for land and sea. The ecosystems of the coastal zone are most productive.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

The most significant anthropogenic factors are as follows: - The industry, especially, that related to mineral resources (oil, gas, coal, uranium ore in the coastal zone and on the shelf) and transport in the coastal zone, its types and load capacity. The industrial centers are classified by the combination of branches of industry, degree of hazard, and the values of maximum permissible concentration of air pollutants. The low, moderate, increased, high, and very high ecological levels of intensity are identified for industrial centers. - The level of river pollution in mouth areas, defining the influx of pollutants to the seas, the overall volumes of their discharge, and the volumes of purified and polluted sewage. - The level and sources of sewage pollution of the sea water surfaces proper. - The main sources of atmospheric pollution, their structure, volumes of emissions in the coastal zone. - Three levels of ecological hazard are identified for transportation centers and communications, namely, moderate, high, and very high, with due regard for freight turnover and traffic, frequency of train movement, presence of polluting freight. Special attention is given to sea waterways, oil pipelines, bases of merchant and fishing fleet. - Potential sources of pollutants such as objects that present the hazard of radioactive contamination of the environment, places of dumping and burial of solid and liquid radioactive waste at sea, bases of atomic fleet, nuclear reactors, and uranium mining and processing plants. - Population density.

Investigation Methods and Results

The systems ecological analysis is usually based on a large body of data about the characteristics of the environment. The data storage is a necessary, but not sufficient, part of the project. The data must be readily accessible on request. The information from different sources must be correlated, compared, analyzed, and visualized in the form of a table, scheme, map, or chart. The modern GIS technology provides the framework for such studies and offers diverse possibilities for the acquisition, integration and analysis of spatial data. Base layers of spatial information were automated using ARC/INFO. GIS ARC/INFO has a topological vector data model. The topology creation is accompanied by the creation of the attribute table for each coverage. The attribute table contains information about the type of the object, area and perimeter for polygons and lengths for linear objects, the internal and user identification number of the object. Such data model enables one to perform various spatial topological operations and correlate data of different types using location in space as the common key. Subsequent work (visualization, editing, combining and analyzing different layers of information, modeling, creating and editing legends and attribute tables, charts, layouts) was done in ArcView GIS 3.1. A geographic map on the 1:10 000 000 scale was used as the base map when drawing the ecologo-geographic map of the Arctic seas. This map was automated. ARC/INFO coverages included layers such as hydrography, sea shore, populated areas, latitude and longitude grid. The outlines of the polygon features, which represented the basic thematic content of the map were likewise automated from separate paper maps. Such information layers on the topological vector data model of ARC/INFO represented sea areas with different stability of sea ecosystems, sea shore stability, geomorphology of the sea shores, water quality of the streams, potential of atmospheric pollution, location of wells and platforms on the shelf, areas of dumping of waste, industrial and mining centers. Linear coverages represented sea ice borders, ice openings, sea water circulation. Based on these layers, a digital map composition of the ecologo-geographic map was compiled in ARC/INFO. The Digital Chart of the World (DCW) was used as the base map for spatial location and modeling of distribution of the point data about pollution of the Caspian Sea. This chart is a comprehensive, 1:1 000 000-scale,

Charting and Navigation / Cartographie marine et navigation 12 Index vector base map of the world, which consists of 17 geographic thematic layers, attribute and textual data that may be accessed, queried, displayed and modified using ARC/INFO software. ArcView was used to create tables in DBF format, containing data on the concentration of different pollutants in different years for different points of the sea areas, as well as the geographic coordinates of the sampling points. The use of the ArcView Spatial Analyst extension containing a wide range of new powerful tools for modeling and analysis of spatial data enabled us to calculate grid themes (georeferenced matrices of the distribution of different pollution parameters) from point themes. We used spline interpolation (with the tension parameter) and IDW interpolation to compile maps of the distribution density of different pollutants in the Caspian Sea. As a result, maps and diagrams of pollution by industrial sewage, oil, detergents, phenols, heavy metals, and pesticides were compiled, as well as maps of the dynamics of the water pollution index. The analysis of the resulting maps and diagrams for a ten-year period (1985-1996) revealed the dynamics and regularities of distribution of the sources and streams of pollutants in the sea. At present, all of the results of ecologo-geographic mapping are in the ArcView GIS environment, which makes possible a multilayer map analysis. These maps are open for editing, union, and transformation into new maps, as well as for operative updating. The next important stage of environmental mapping is the estimation of negative ecological results of the anthropogenic impact on the sea environment. The results of anthropogenic influence on sea ecosystems may show up as changes in the average biomass of plankton and benthos, a simpler community structure and poorer variety of species, a larger number of indicator species of microorganisms, disappearance of the bottom fauna, reorganization of cenoses and appearance of species-installators, eutrophication and supereutrophication of water in coastal areas, gulfs, and bays. The empirical scale of qualitative criteria of the ecological state of sea ecosystems includes four grades, namely, stable, transitional (from stable to critical), critical, and catastrophic. In the process of ecologo-geographic mapping of Arctic seas and of the Caspian Sea (Dagestan coast), an attempt was made at assessing the state of sea ecosystems using the above-identified hydrochemical and hydrobiological indicators characterizing the sea environment and its ecosystems. The final estimate that char- acterizes the ecological intensity of the sea water surfaces being investigated involves the following indicators. 1. The water pollution index. There are five grades of the state of water, namely, relatively clear, moderately polluted, polluted, dirty, and extremely dirty. 2. The amount of basic pollutants in fractions of maximum permissible concentration for fishing. These include petroleum products, phenols, detergents, heavy metals, pesticides, and biogenic products. 3. The oxygen regime. 4. The changes in the average biomass of plankton and benthos. 5. The simpler community structure and poorer variety of species. 6. The growth of the number of indicator species of microorganisms. 7. The disappearance of the bottom fauna. 8. The reorganization of cenoses and appearance of species-installators. 9. The eutrophication and supereutrophication of water, especially, in coastal areas, gulfs, and bays. The integrated geographic analysis of the ecologo-geographic mapping reveals the ecological situation characterized by a rapid increase of concentration of pollutants in the coastal areas and by low concentration of pollutants in the distant deep sea zones. As regards the pollution level, the explored part of the Caspian Sea may be divided into three conventional zones, namely, those of the most, medium, and least concentration of pollutants. The zone subjected to the maximum anthropogenic impact borders on the coast line and on the 10-meter isobathe on the outside.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

The hydrological regime of the Caspian Sea promotes the concentration of pollutants and their propagation with sea currents along the coastal area. The zone of medium concentration (off-shore) is characterized by a more homogeneous composition of pollutants compared with the coastal area and is located between the 10- and 25-meter isobathes. The water surface zone between the 25-meter isobathe and open sea is characterized by the minimum concentration of pollutants. The analysis of the ecological situation of the Dagestan coast and Caspian Sea reveals that sea water in the coastal area is polluted dangerously. Water in this area is characterized as “dirty” or “polluted” based on the integrated estimation of the hydrochemical index of water quality. For the past ten years, the copper content increased eight times, that of zinc - ten times, lead - seven times, and of cadmium - five times. During the same period, the content of mercury, petroleum hydrocarbons (up to seven times the maximum permissible concentration), and phenols (up to 13 times the maximum permissible concentration) increased as well. The eutrophication of water, accompanied by plankton bloom and subsequent dying off, covers the entire Caspian coastal zone. The most dangerous ecological situation is observed at the mouth and off-shore of River Terek. The amounts of ammonia nitrogen and copper pollutants coming from farming lands to the Terek mouth demonstrate that this river water cannot be used for fish breeding without preliminary purification. Waste water from oil-processing and nonferrous metallurgy discharged from River Terek into the Caspian Sea led to the creation of zones of biological death in this area. The ecological situation deteriorated appreciably in the regions of the towns of Makhachkala, Caspiisk, Derbent, and Izberbash, and at the mouths of Rivers Sulak and Samur. One can describe the ecological situation of the northern part of the Caspian Sea as a small disturbance. The ecologo-geographic mapping of Arctic seas made it possible to identify different classes of anthropogenic impact on the water surface and coastal zones. The regions with the catastrophic state of ecosystems include the Teribersk Inlet, the Barents Sea, the Bulunkan Gulf of Tiksi Bay in the Laptev Sea. For the Dvina and Onega Gulfs in the White Sea and for the southern areas of the Ob’ and Tazovsk Inlets, the state of ecosystems may be described as transitional from catastrophic to critical. The regions with the critical state of ecosystems include the water surfaces of the Kandalaksha and Mezen’ Gulfs in the White Sea, the central part of the Ob’ Inlet, Chaun Inlet, Pevek Channel, and the coastal zone of the Chukchee Sea. The ecological situation of the water surface near the Novosibirsk Isles, west of Novaya Zemlya, in the northern parts of the Ob’ and Enisei Inlets, and in the central part of the White Sea may be described as transitional from stable to critical. The situation of the water surface of the coastal areas and open sea in the East Siberian Sea, and in the central part of the Chukchee, Laptev, Kara and Barents Seas may be described as favorable. As a result of analysis of the geoecological map of Russia’s Far Eastern seas, the following may be regarded as ecological risk regions: in the Sea of Okhotsk - the north-eastern shelf of Sakhalin, the Terpeniya, Aniva and Tauiskaya Bays, and the water surface near the Oktyabr’skii settlement; in the Bering Sea - the Kamchatka and Avachinskii Gulfs and the Anadyr’ Lagoon; and in the Sea of Japan - the Gulf of Tatary and the inner part of the Peter the Great Bay. Serious disturbances of the land ecological balance, which affected directly or indirectly the sea water surface conditions, were observed in the north-eastern and southern parts of Sakhalin, in the south of the Far-Eastern region, in the mining regions between the Amur and Khabarovsk Provinces, in the upper reaches of the Indigirka and Kolyma rivers, in the vicinity of the towns of Petropavlovsk-Kamchatskii and Anadyr’. The environmental mapping of Russia’s seas using GIS will help assess the results of human activities in different natural zones for subsequent ecological monitoring.

Charting and Navigation / Cartographie marine et navigation 12 Index

Session / Séance 06-C Airborne remote sensing for water quality mapping on the coastal zone of Abruzzo (Italy)

Claudio Conese I.A.T.A. – C.N.R., Via Caproni n. 8, 50145 Firenze (Italy) e-mail: [email protected]

Marco Benvenuti Ce.S.I.A. – Accademia dei Georgofili, Via Caproni n. 8, 50145 Firenze (Italy)

Paola Grande Ce.S.I.A. – Accademia dei Georgofili, Via Caproni n. 8, 50145 Firenze (Italy)

The main goal of this work is to monitor the coastal environment of Abruzzo, a central Italy region, by means of airborne remote sensing techniques. In particular the physical and biological dynamics of both natural and anthropogenic phenomena are pointed out. This work puts in evidence the importance of remote sensing in mapping and managing the territory and the environmental resources. Some processing methodologies defined to support the creation of a coastal monitoring service will be shown. The IATA and the CeSIA institutes developed algorithms and analysis procedures of airborne remote sensing data to provide methodologies for mapping the sea water quality along the Abruzzo coast. The airborne multispectral scanner, VIRS 200 (Visible InfraRed Scanner), was used to collect images on different sites of the coast near the principal rivers foci in order to monitor the presence and the distribution of water pollutants. The airborne system acquires in the visible and near-infrared bands (400-1000 nm) and the channels are properly set for marine applications. A series of sea truth campaigns was carried out to collect the physical and bio-chemical parameters. For this purpose the multiparametric probe IDROMAR IM51 was used. The measured parameters were correlated with synthetic indices, obtained from the remote sensed data, putting in evidence environmental phenomena. Furthermore, the thermal camera ETS 512, detecting the thermal emission in the 8-12 mm band, was used in order to create the thermal maps near some mouths of rivers. The temperature gradient between river-water and sea-water allows to evaluate the distribution of the fresh-water into the sea. Both VIRS 200 and ETS 512 was produced by Officine Galileo, Florence (Italy) and were available at the Environmental Control Centre of Atri (Italy). The two acquisition systems have spectral, radiometric and spatial characteristics useful for a detailed analysis of the environmental parameters.

Study area

The study area is located along the coast of Abruzzo, a central region of Italy, on the Adriatic sea, in proximity of the mouths of the following rivers: Alento, Pescara and Sangro. The coast is characterised by the presence of breakwater, built during the last twenty years, in order to prevent the erosion phenomenon that reduces the width of the beach. This resolution has brought, as a consequence, to the permanence of the water pollutants and the sediments, transported by the river to the sea, nearby the rivers

Charting and Navigation / Cartographie marine et navigation 12 Index

mouths. The problem of water pollution can be studied by processing remote sensing images, acquired in the opportune spectral bands with the desired spatial resolution. The remote sensing approach allows to collect in a simple way information about different sites in several dates, that can be related to the measured parameters providing a description of the observed phenomenon (in this case the distribution of water pollutants). The environmental problem increases near the ports, where the pollution is due also to the maritime traffic. Pescara river mouth is located near the port creating a channel extended up to the sea, and it is characterised by an offshore dam. This particular structure of the port has produced a pollutants distribution with a typical “mushroom” shape.

Figure 1. Geographic location of The acquisition systems Abruzzo

The used airborne multispectral system is the VIRS 200, designed and produced by Officine Galileo of Florence (Italy) for environmental monitoring. It is characterised by an high spatial and spectral resolution. The VIRS 200 is a passive system, able to detect and record the spectral radiance scattered from the ground in the 400-1000 nm band of the electromagnetic spectrum. It is possible to program 20 channels, with a minimum band-width of 2.5 nm, among 240 possible channels, whose Figure 2. VIRS 200 system. choice depending on the specific application. This system is a line scanner using a CCD (Charge-Coupled Device) matrix of 240 arrays of 512 pixels. Each array is sensible to a particular band. Only twenty arrays among the 240 available can be selected and, consequently, the resulting multispectral image has 20 bands. The image is digitised and recorded on a magnetic tape in a Band Interleaved Line format. The spatial resolution of an airborne acquisition system depends on the flight altitude and is determined with IFOV the relation: l = (H − h)*arctg( ) 2 where: l: pixel dimension; H: absolute altitude of the aeroplane; h: elevation of the observed surface; IFOV: Instantaneous Field Of View. Thus, the flight altitude is established in function of the elevation of the observed area and of the required spatial resolution.

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

The system consists of two principal components: the avionic section and the ground section. The avionic component includes the detection unit, the acquisition and control unit and the data recorder, while the ground part is the device used to read digital data and the software used to record the image on a computer in a compatible format.

Table 1. VIRS characteristics.

Spectral range 400-1000 nm Scanning Pushbroom Spectral resolution 2.5 nm Pixels per line 512 pixels Selectable spectral bands 20 over 240 Dynamic range 10 bit Field of View (FOV) 37.6° Roll stabilisation 0.1 mrad rms (±15°) Instantaneous FOV (IFOV) 1.66 mrad Recorder data rate 1 Mbyte/sec

Table 2. VIRS spectral channels selection.

Channel λ0 (nm) Channel λ0 (nm) Channel λ0 (nm) Channel λ0 (nm) 1 448.75 6 521.25 11 641.25 16 801.25 2 451.25 7 531.25 12 661.25 17 831.25 3 461.25 8 558.75 13 681.25 18 858.75 4 483.75 9 578.75 14 721.25 19 901.25 5 501.25 10 601.25 15 771.25 20 961.25

The thermal camera ETS 512 (Environmental Thermal Surveyor) is a system for airborne remote sensing for environmental monitoring. The sensor acquires the thermal emission in the 8-12 mm band. The scanning format adopted to create the image is the standard television one and the acquired images are recorded on a videotape. The IR images on the surveyed territory are partially overlapped because of aeroplane motion. The flights are performed at constant altitude and are parallel and rectilinear. The infrared camera is located on the aeroplane in order to carry out a nadir view of the surveyed territory. The system is characterised by: • high spatial resolution; • high radiometric resolution; • high field of view; • thermal references of the instrument.

Table 3. ETS characteristics.

Spectral range 8-12 µm Horiz. Field of View (FOV) 38° Vert. Field of View (FOV) 28° Instantaneous FOV (IFOV) 1.5 mrad Pixels per line 512 pixels Precision < ±0.5 °C Accuracy < ±0.5 °C

Charting and Navigation / Cartographie marine et navigation 12 Index

The two thermal references are provided by two thermoelectric elements based on the Peltier effect, for which a junction of two different conductors (in this case drugged semiconductors), crossed by the current, produces an increase of the temperature in one side of the junction and a decrease of the temperature in the other side. The precision and the accuracy of the thermal references are better than ±0.5 °C. The complete system is formed by the avionic subsystem and the ground subsystem. The first one includes the thermal camera, the control electronic unit, the remote control panel and the recording system. The ground section, including the hardware necessary for the vision of the acquired images, allows to convert the signal from analogue to digital, to record the images line by line and to process them by means of a computing system.

Truth campaigns

The sea and rivers truth campaigns have taken place at the same time of the remote sensing campaign, because of the very rapid dynamics of the observed physical and bio-chemical parameters. The collected data were compared to the multispectral data acquired by the VIRS 200 in the different selected bands and to synthetic indices, calculated by combining the spectral bands. The sampling points along the coast were chosen in order to collect data in the nearby of the rivers mouths and along the coast considering how far from the foci the river water arrives but also taking into account the width of the swath observed by the sensor.

3.1 River campaign The mouth river campaigns described in this paper took place in the following dates: • 10.03.97: Vomano river (mouth survey); Pescara river(mouth survey); • 11.03.97: Sangro river (mouth survey). The following methodology was applied for each river: 1. the multiparametric probe IDROMAR IM51 was used to monitor continuously for an hour the parameters: pH, temperature, dissolved oxygen (concentration and percentage), conductivity, turbidity, salinity; 2. acquisition of two samples for each measurement point and laboratory analysis of the typical chemical and physical water parameters.

3.2 Sea campaign (14.03.97) The coastal water campaign was conducted by using an equipped boat. Twenty sampling and measurement stations were determined and the surveys were carried out as follow: · n. 2 transepts with sampling points at 150, 300, 1000 and 2000 m from the coastal line, in correspondence of the Saline and Pescara rivers; • n. 2 transepts with sampling points at 150, 300 and 1000 m from the coastal line, in correspondence of S. Silvestro zone (Pescara sud) and Alento river (mouth); • n. 3 transepts with sampling points at 150 and 300 m from the coastal line, in correspondence of Zanni zone (Pescara nord), Montesilvano sud and Silvi Marina zone. The methodology used for the survey is: 1. the multiparametric probe IDROMAR IM51 was used, with a survey of a vertical profile on each station, on the following parameters:

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

pH, temperature, dissolved oxygen (concentration and percentage), conductivity, turbidity, salinity, relative density, chlorophyll “a” (through coupled fluorometer); 2. determination of 2 samples for each measure point (surface and floor) and laboratory analysis on some parameters. The monitoring programme produced quite positive results although the sea conditions were not optimal. Data acquisition was useful for the setting of the airborne equipment used during the remote sensing mission.

4. Methodology

The VIRS images of the Pescara port and the Alento mouth were acquired on 11.04.97. The platform altitude was 3000 m and the consequent average ground resolution was about 3 m. A series of pre-processing techniques were applied in order to obtain corrected images useful for the subsequent processing. The considered images presented a strong periodical noise, called “striping”, more evident in the lower bands, in particular in the blue band, more sensible to the absorption of the atmosphere. An algorithm, based on the FFT (Fast Fourier Transform) is able to determine the particular frequencies (that in the frequency domain appear like straight lines) producing the striping. Afterwards these frequencies were eliminated by applying an Hanning filter and than the IFFT (Inverse Fast Fourier Transform) was performed. The so obtained image is almost completely without striping noise and sharper. The georeferencing procedure belongs to the pre-processing techniques and allows to gives map coordinates to each point of the image. The used geographic reference projection is the UTM, ellipsoid International 1909, datum European 1950, zone 33. For the projection the orthophoto (ed. 1982, Abruzzo Region), at scale 1:10000 was used in order to determine the Ground Control Points. Principally two indices were calculated, the turbidity and the sediment concentration, making use of the spectral bands previously corrected. On the basis of the results of the spectral profiles analysis, it was possible to find the correlation formulas defining the two indices: = + + turbidity a b * Log10 (R87 2)

where a=12.729677, b=-46.020389, R87=(B8-B7)/(B8+B7). For the turbidity the bands 7 and 8 resulted more correlated to the sampling data. In Figure 3 and Figure 4 the turbidity maps for Pescara and Alento mouths are shown.

Figure 3. Turbidity - Pescara mouth. Figure 4. Turbidity - Alento mouth.

Charting and Navigation / Cartographie marine et navigation 12 Index

The turbidity unit is the FTU (Formazine Turbidity Unit - ISO). Figure 5 shows the experimental relationship between the turbidity samples, collected during the sea truth campaigns, and the coast distance. It is possible to observe that there is a rapid decrease of the turbidity with an increase of the coast distance. Obviously the highest values are reached near the river mouth or where the water is rather still (near the breakwater).

Coast Saline Pescara Alento Distance mouth mouth mouth 150 58.43 46.985 9.655 300 56.44 53.95 4.18 1000 6.97 25.11 6.015 2000 5.24 13.14

Figure 5. Relation between measured Table 4. Turbidity values (in FTU) on the transepts. turbidity and coast distance for three rivers.

As concern the soil suspended index the experimental founded formula, using the band 1 and the band 8 of the VIRS image, is: = + ssi a * Ln(B18 ) b

where a=-0.0409, b=-0.7893, B18=(B8-B1)/(B8+B1). From the sediments concentration, measured during the river campaign, in sampling points located at different distance from the coast, a sediments index was extracted:

Figure 6. Suspended Solid Index (Alento river).

Ottawa ICA / ACI 1999 - Proceedings / Actes 12 Index

The ETS 512 flied on march ’97, covering all the coast of Abruzzo. The flight altitude was 1000 m on sea level, with a swath of about 660 m and a spatial resolution of 1.5 m.

Figure 7. Thermal map (March ’93). Figure 8. Thermal map (March ’97).

Conclusions

The described activities were carried on to define a methodological approach able to provide regular and objective information about the water quality parameters. In fact, the use of remote sensing techniques, validated by the acquisition of truth data, can be useful to provide systematic and comparable information on a large area. Thanks to the possibility of the VIRS instrument of setting the spectral position of the twenty acquisition channels, the system and the developed methodology can be tuned depending on the specific needs. The study shown how the combined use of a multispectral high resolution scanner and thermal camera can represent a very useful tool to monitor the coastal environment and, in particular, the water parameters. An interesting result of this research arise from the multitemporal comparison of the thermal images collected on the Pescara. This analysis shown the different distribution of the freshwater into the sea before and after the offshore dam in front of the port was built up (see Figure 7 and 8). This phenomenon is quite evident also in the result of the multispectral image processing (Figure 3). From this analysis arise that the dams and the breakwater placed along the coast to reduce the coastal erosion process, can be considered as one of main the causes of the increase of suspended solid concentration and water pollutants along some portions of the Abruzzo coast. An important role is played by the truth campaigns necessary to collect the real parameters for the calibration of the correlation models to be applied to the remote sensed images in order to have a spatialisation of the related index and, consequently, a global view of the studied phenomena. However, some problems came from the absence of an automatic georeferencing system on the used remote sensing equipment, problem that is much more heavy in marine environment where it can be very difficult to find the reference points necessary for the registration of the acquired image on respect to a cartographic reference.

Charting and Navigation / Cartographie marine et navigation 12 Index

References

Alberotanza, L., Masserotti, M.V., (1992). Fondamenti del telerilevamento ed applicazioni all’ambioente marino. AIT Lillesand, T.M., Kiefer, R.W. (1994). Remote Sensing and Image Interpretation. John Wiley & Sons Inc., New York Slater, P.N., (1980). Remote Sensing: Optics and Optical Systems. Addison-Wesley Publishing Co., Reading Massa- chusetts Morel, A., Prieur, L. (1977). Analysis of variation in ocean color, Limnology and Oceanography. Toselli, F., Bodechtel, J., (1992). Imaging spectroscopy: fundamentals and prospective applications. Klwer Academic Publisher, London

Ottawa ICA / ACI 1999 - Proceedings / Actes