Вы находитесь на странице: 1из 105

Diploma Thesis:

Design and Implementation of a Digital Geographic


Representation Framework - as exemplied by the
IMPETUS Atlas (Version 2.0)

Presented by Stefan Alfons Krüger

Supervised by Prof. Gunter Menz

January 27, 2008


ACKNOWLEDGEMENT ii

Acknowledgement

This thesis has been developed in cooperation with the IMPETUS project where I have

been actively supported by all participants.

Special thanks go to Dr. Hans-Peter Thamm for his advice and creative energy which

provided the backbone of the IMPETUS atlas idea and thus this thesis. I would also like

to recognize my supervising tutor Prof. Dr. Gunter Menz for his valueable contribution.

Further I have to thank Michael Judex for all the subtle advice that I gained from

him in countless brainstorming sessions. I also want to thank Martin Schmitz and the

whole community of programmers that devote their work to the public domain. Without

Open Source Software and their contributions a thesis such as this would not have been

possible.

Last but not least I want to thank my dear friend Gregor Fellenz who supported me

in many long nights with valuable advice and great company.

Bonn, January 2008


CONTENTS iii

Contents

1 Introduction 1
1.1 The growing pool of geospatial products . . . . . . . . . . . . . . . . . . . 1

1.2 Ineciencies in the communication . . . . . . . . . . . . . . . . . . . . . . 1

1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.3.1 Design of a framework for the publication of multimedia atlases . . 3

1.3.2 Implementation of the framework as prototype software . . . . . . 3

1.4 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Theory 5
2.1 Evolution of mapping: descriptive to prescriptive to perceptive . . . . . . 5

2.2 From coordinates to geoproducts . . . . . . . . . . . . . . . . . . . . . . . 7

2.2.1 Coordinate Reference Systems . . . . . . . . . . . . . . . . . . . . . 7

2.2.2 Geodata, geoinformation, and geoproduct . . . . . . . . . . . . . . 8

2.3 Digital Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.3.1 Cartography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.3.2 Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.3.3 Map Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.4 Digital Geographic Representation . . . . . . . . . . . . . . . . . . . . . . 13

2.4.1 Separation of storage and visualization . . . . . . . . . . . . . . . . 13

2.4.2 Cartographic representation rules . . . . . . . . . . . . . . . . . . . 14

2.5 Communication via maps . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.5.1 Map makers and the map encoding phase . . . . . . . . . . . . . . 15

2.5.2 Map users and the map decoding phase . . . . . . . . . . . . . . . 18

2.6 Multimedia Atlas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.6.1 Atlas - denition and history . . . . . . . . . . . . . . . . . . . . . 20

2.6.2 Multimedia and cartography . . . . . . . . . . . . . . . . . . . . . 21

2.6.3 Maps in Multimedia Products . . . . . . . . . . . . . . . . . . . . . 22

3 The Interactive IMPETUS Digital Atlas 1.0 25


3.1 IMPETUS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3.2 The IMPETUS Interactive Digital Atlas (1.0) . . . . . . . . . . . . . . . . 26

3.2.1 Shortcomings of atlas viewer software . . . . . . . . . . . . . . . . 28

3.2.2 Diculties lling the IIDA1 with data . . . . . . . . . . . . . . . . 29


CONTENTS iv

4 The Digital Geographic Representation Framework 31


4.1 Atlas Authoring Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4.1.1 Geo-spatial Import . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4.1.2 Cartographic Representation Rules . . . . . . . . . . . . . . . . . . 35

4.1.3 Multimedia Import . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.1.4 Content Management . . . . . . . . . . . . . . . . . . . . . . . . . 36

4.1.5 Internationalization . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

4.2 Atlas Package Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

4.2.1 Flexible distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 37

4.2.2 Integrity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4.2.3 Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4.3 Atlas Visualization Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4.3.1 Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4.3.2 Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.3.3 Usability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4.4 Openness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.4.1 Interoperability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.4.2 Open Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

5 AtlasCreator-AtlasViewer Implementation 48
5.1 The AtlasViewer application . . . . . . . . . . . . . . . . . . . . . . . . . . 50

5.1.1 Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

5.1.2 Interactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

5.1.3 Subsumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

5.2 The AtlasCreator application . . . . . . . . . . . . . . . . . . . . . . . . . 61

5.2.1 Import . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

5.2.2 Content Management . . . . . . . . . . . . . . . . . . . . . . . . . 67

5.2.3 Subsumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.3 Atlas Package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.3.1 Integrity and Compression . . . . . . . . . . . . . . . . . . . . . . . 73

5.3.2 Flexibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

5.3.3 atlas.xml conguration le . . . . . . . . . . . . . . . . . . . . . . 76

5.3.4 Subsumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

5.4 Preview of IMPETUS Interactive Digital Atlas (version 2.0) . . . . . . . . 77


CONTENTS v

6 Discussion and Conclusion 78


6.1 Evolution of the DGRF design . . . . . . . . . . . . . . . . . . . . . . . . 78

6.2 Implementation of the AtlasViewer - AtlasCreator framework . . . . . . . 79

6.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

References 83

A Appendix 91
A.1 Preparing data to be imported into the AtlasCreator . . . . . . . . . . . . 91

A.2 Image data compression algorithms . . . . . . . . . . . . . . . . . . . . . . 91

A.3 Preparing Tiled Image Pyramid (TIP) layers for import into AC . . . . . 92
LIST OF FIGURES vi

List of Figures

1 The Map use cube showing dierent visualization strategies. Adapted

from MacEachren & Kraak (1997) . . . . . . . . . . . . . . . . . . . . . . 12

2 Map encoding phase. Adapted from Haggett (2001, S.677). . . . . . . . . 15

3 Map decoding phase. Adapted from Haggett (2001, p.677). . . . . . . . . 15

4 Map communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

5 Map eectiveness in thought and communication. Adapted from Robinson

et al. (1995, p.17, gure 2.7) . . . . . . . . . . . . . . . . . . . . . . . . . . 16

6 The visual variables. Van Elzakker (2004, p.26). . . . . . . . . . . . . . . 18

7 An animated map: Sim City game screenshot. . . . . . . . . . . . . . . . . 22

8 Interactive Digital IMPETUS Atlas 1.0 . . . . . . . . . . . . . . . . . . . . 26

9 Digital Geographic Representation Framework . . . . . . . . . . . . . . . . 32

10 Global distribution of the Digital Opportunity Index. ITU/UNCTAD (2007) 38

11 AtlasCreator-AtlasViewer Framework . . . . . . . . . . . . . . . . . . . . . 48

12 AtlasViewer - Example for map visualization. . . . . . . . . . . . . . . . . 51

13 AtlasViewer - Geo-visualization component. . . . . . . . . . . . . . . . . . 52

14 Tiled Image Pyramid (TIP) for ecient visualization of big rasters . . . . 54

15 AtlasViewer - Interactive legend . . . . . . . . . . . . . . . . . . . . . . . . 55

16 AtlasViewer - Map tools. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

17 AtlasViewer - Info-tool querying a land-use raster . . . . . . . . . . . . . . 57

18 AtlasViewer - Info-tool querying vector objects . . . . . . . . . . . . . . . 58

19 AtlasViewer Geo-object lter tool. . . . . . . . . . . . . . . . . . . . . . . 59

20 AtlasCreator - Authoring the IIDA2. . . . . . . . . . . . . . . . . . . . . . 62

21 AtlasCreator - Testing compatibility during video import . . . . . . . . . . 67

22 Model for structuring atlas content . . . . . . . . . . . . . . . . . . . . . . 68

23 Topology of data pool entries . . . . . . . . . . . . . . . . . . . . . . . . . 68

24 AtlasCreator - Map composer . . . . . . . . . . . . . . . . . . . . . . . . . 70

25 AtlasCreator - Translation of data pool entries and maps . . . . . . . . . . 71

26 AtlasCreator - Translation of database attributes . . . . . . . . . . . . . . 71

27 AtlasCreator - Translation of raster legends . . . . . . . . . . . . . . . . . 72

28 JPEG compression artifacts in a TIP layer at low resolution . . . . . . . . 94


LIST OF TABLES vii

List of Tables

1 Theme groups in the Atlas of Switzerland 2.0. (Hurni & Räber, 2004, p.4) 24

2 Atlas communication phases and corresponding DGRF modules. . . . . . 32

3 Operating systems visiting w3schools.com in 2007. (W3Schools, 2007) . . 45

4 Creators, consumers, and required functionality for IMPETUS. . . . . . . 49

5 Selected contents of the IIDA2 demo. . . . . . . . . . . . . . . . . . . . . . 78


ABBREVIATIONS viii

Abbreviations

AAT Atlas Authoring Tool

AC AtlasCreator

AC-AV AtlasCreator-AtlasViewer framework

AIS Atlas Information System

AoS Atlas of Switzerland

AoS2 Atlas of Switzerland 2.0

AtlasML Atlas Markup Language

AV AtlasViewer

AVT Atlas Visualization Tool

BMBF Bundesministerium für Bildung und Forschung (en.: Federal Ministry of

Education and Research)

CRS Coordinate Reference System

CS Coordinate System

DEM Digital Elevation Model

DGRF Digital Geographic Representation Framework

DOI Digital Opportunity Index

EPSG European Petroleum Survey Group

ESRI Environmental Systems Research Institute

FOSSGIS Free and Open-Source Software for GIS

GDAL Geospatial Data Abstraction Library

GIS Geographic Information System

GML Geographic Markup Language

GUI Graphical User Interface


ABBREVIATIONS ix

GPS Global Positioning System

GViz Geographic Visualization

HTML Hypertext Markup Language

HVO Haute Vallée de l'Ouémé (en.: Upper catchment of the Ouémé river)

ICT Internet and Communication Technologies

IIDA1 IMPETUS Interactive Digital Atlas (version 1.0)

IIDA2 IMPETUS Interactive Digital Atlas (version 2.0)

IMPETUS Integratives Management-Projekt für einen Ezienten und Tragfähigen

Umgang mit Süÿwasser (en.: Integrated approach to the ecient management of

scarce water resources)

JAR Java Archive

JMF Java Media Framework

JNLP Java Network Launching Protocol

JVM Java Virtual Machine

JWS Java Web Start

LGPL Library GNU Public License

LUCC Land-Use and Land-Cover Change

LZW The Lempel-Ziv-Welch lossless data compression algorithm

OGC Open Geospatial Consortium

OLPC One Laptop Per Child

OOP Object Oriented Programming

OS Operating System

OSI Open Source Initiative

PDF Portable Document Format


ABBREVIATIONS x

RSRG Remote Sensing Research Group

SDI Spatial Data Infrastructure

SLD Styled Layer Descriptor

TIP Tiled Image Pyramid

W3C World Wide Web Consortium

WCS Web Coverage Service

WFS Web Feature Service

WKT Well-known Text

ZFL Zentrum für Fernerkundung der Landoberäche (en.: Center for Remote Sensing

of Land Surfaces)

XML Extensible Markup Language

XULU Extendable Unied Land Use modelling platform


1 INTRODUCTION 1

1 Introduction

1.1 The growing pool of geospatial products


In the last decades the volume of geospatial data has grown exponentially. One im-

portant factor is the technical progress in air- and spaceborne surveillance. Hundreds of

dierent satellites are circling Earth and the world can be observed in ever-higher spatial,

temporal, and spectral resolution (Haggett, 2001, p.710).

In addition, the distribution of aordable receivers for the Global Positioning System

(GPS) - often embedded in vehicles, digital meters, cellular phones, or mobile computers

- contributes to the accelerated provision of geodata (Haggett, 2001, p.667).

Not only the physical parameters of the Earth's surface, but also socio-economic

data like shopping behavior, market opportunities, beliefs, and behaviors become widely

available as geo-referenced data.

Map based applications have become ubiquitous (Morita, 2004; Hrebícek & Konecný,

2007; Dietze & Zipf, 2007). Geographic Information System (GIS) experts and research

institutions were once the only users of geospatial data. Today in developed countries

where Internet and Communication Technologies (ICT) are widespread and aordable the

general public is developing a growing interest in mapping applications. Location aware

mobile phones and online services like GoogleMaps


1 are just some prominent indicators

that a growing demand for geospatial products exists among the general public (Morita,

2004).

Geospatial analysis and processing techniques are diusing. Therefore the demand for

homogenization is growing. The international Open Geospatial Consortium (OGC) aims

at the creation of standards for geospatial data exchange. The standards are slowly being

implemented on a wider scale (Lake & Farley, 2007; Lake, 2007). The new generation

of standardized and open geospatial interfaces makes it easier to create new geospatial

products by combining dierent data sources and technologies (Bartelme, 2005, p.363).

Decision-makers and stakeholders at all levels of public and private enterprise as well as

the general public can benet from the growing wave of geospatial products.

1.2 Ineciencies in the communication


Regrettably, many of the geospatial products are only generated for a singular purpose or

one consumer. Too seldom third parties get access to the data. This has many reasons:

1 GoogleMaps webpage: http://maps.google.com


1 INTRODUCTION 2

Often projects that generate geoinformation simply lack the conceptual foresight

needed to catalog or export the data in a format others can use after the project nished.

Much of this data has been tediously collected and processed to answer one specic matter

but never generate any value (scientic or scal) on their own, many are never shared.

To date, the manufacturers of many of these products have failed to see the intrinsic

value of this data and third parties are seldom privy to the information.

Other organizations, like the IMPETUS project (see 3.1), have the clearly stated aim

to transfer their research ndings to other scientists, stakeholders, and local decision-

makers (Thamm et al., 2006). Even for them, the transfer and re-use of the geospatial

information can be a very cumbersome, expensive, and ineective task. This is primarily

due to trying to preserve the detailed metadata and embedded question sets.

Growing demand for geoproducts is being generated increasingly by people who are

not educated in geospatial data management, GIS, or map-reading. Stakeholders are

experts in their specic socio-economic domains and the geospatial products are supposed

to support them without the need for GIS training courses.

Transferring geospatial data from a GIS to a novice user can be troublesome due

to the inhomogeneous formats and the amount of metadata (Bartelme, 2005, p.371).

Two GIS professionals sharing a layer with the popular Environmental Systems Research

Institute (ESRI) Shapele format is possible. That same layer, when presented to a

decision-maker without GIS knowledge, is nearly useless. It has to be imported and

visualized in a GIS before it can be of any use. When a set of layer les is transferred,

they must rst be assembled into meaningful maps. This can overburden the occasional

user.

All geospatial exchange formats which solely contain geodata are not well suited to

comprehensively transferring a complex geospatial problem. From a recipients perspec-

tive, a ready-to-use packaging for the geoproduct is desired. It should incorporate a

stand-alone visualization tool to represent all geodata in a meaningful and understand-

able manner, as well as auxiliary information like documents, statistics, or multimedia.


1 INTRODUCTION 3

1.3 Objectives
This work aims to design and implement a framework which improves the communication

of geoproducts in the form of a digital, interactive, multimedia atlases. The actors of

that communication are institutions which run GISs and a growing public that wants to

prot from the new geoproducts but is not educated in GIS.

1.3.1 Design of a framework for the publication of multimedia atlases


The framework will span the whole workow of creating, transferring, and using mul-

timedia atlas products. It will not be bound to specic contents, but be generic and

re-usable for diverse topics. It assumes that the geoproducts are created in GISs and the

recipients wish to use them on standard computers. The design of the framework will be

based on theories of geo-visualization (2.3.3), map communication (2.5), and multimedia

cartography (2.6).

Using open standards in cartographic multimedia products is a design decision which

positively aects the sustainability of the product (Jobst, 2007, p.425). Hence the frame-

work will be designed according to open standards wherever appropriate.

1.3.2 Implementation of the framework as prototype software


A prototype software of the designed framework will be implemented. A time-frame of

three months is scheduled for programming. During that time as many of the conceived

features as possible will be realized. The desire is for the software to reach a level of

practical usability, i.e. the prototype intends to be usable for creation and publication

of multimedia atlases. Creation and publication of the new digital IMPETUS atlas shall

serve as a test case. Specic translations in dierent languages or detailed preparation

of the contents are not prioritized. The implementation does not focus on creating a

beautiful multimedia atlas, but on the creation of tools which empower scientists to

publish eective multimedia atlases by themselves.


1 INTRODUCTION 4

1.4 Outline
Theory
In the second chapter the basic concepts related to the design and implementation of a

digital geographic representation framework are introduced. 2.1 traces the evolution of

mapping from descriptive to prescriptive and perceptive. Concepts for the communica-

tion via paper and digital maps are further reproduced. Examples of existing multimedia

mapping applications are presented in 2.6.2.

Experiences with the rst Interactive Digital IMPETUS Atlas


This third chapter describes the objectives of the IMPETUS Interactive Digital Atlas

(version 1.0) (IIDA1) and reviews its eectiveness. Based on experiences gained during

its creation process in 2005, a qualitative evaluation of discovered shortcomings and

desired functionality is given.

The Digital Geographic Representation Framework


The Digital Geographic Representation Framework (DGRF) (gure 9) is designed as

a facilitation for generic digital atlas productions on the basis of map communication

concepts. A pair of tools and a generic interchange format are conceived. A catalog of

features and requirements is deduced from experiences with the creation of the IIDA1.

AtlasCreator-AtlasViewer Implementation
This chapter describes how the AtlasCreator - AtlasViewer (AC-AV) framework has been

implemented following the concept of the DGRF. An overview of the applied technologies

and their interaction is given, as well as an introduction to the programmed tools and

Graphical User Interfaces (GUIs). Some specic technical solutions to abstract problems

are described, like the development of the Tiled Image Pyramid (TIP) layer format to

visualize large rasters fast on standard computers.

Discussion and Conclusion


The last chapter summarizes how the basic concepts of traditional paper maps have

inuenced the design of the DGRF (78). In 79 a critical discussion reviews how the

requirements of the DGRF have been met with the AC-AV implementation. 82 gives a

nal conclusion.
2 THEORY 5

2 Theory

This chapter will introduce basic terms and theories needed to approach the concepts of

modern multimedia atlas products.

2.1 Evolution of mapping: descriptive to prescriptive to perceptive


In the beginning of the 20th century the technical and mathematical solutions were good

enough to create accurate maps, which described virtually every place of interest on our

planet's surface. In that time maps were manually created and analyzed (Berry, 1993,

p.3). But manual analysis of analog maps is dicult and limited. Therefore the focus of

the analog map and its manual processing was descriptive - to describe occurrences and

distribution of landscape features.

The descriptive property of a map has never been lost but as describing the physical

reality with a map became easier, people had more time for analysis. In the 1960s,

separating maps in layers of dierent feature types was common and allowed to compose

on-demand maps with focus on special topics. Land and resource planning embraced the

new technologies and started to use the maps as a decision basis for management. That

was a turning point in the use of maps, from techniques that emphasize the physical

descriptions of geographic space to those that spatially characterize management actions

(Berry, 1993, p.3).

This shift from descriptive to prescriptive mapping was the precursor to embrace

computer-assisted analyses in mapping. With the digital revolution in the 1970s, maps

moved from analog images representing a spatial distribution to geographically referenced

digital data - thereby quantifying a physical, social, or economic system in prescriptive

terms (Berry, 1993, p.3).

Spatial analysis techniques were developed which allowed to derive new geoproducts

by combining digital maps. Computer-aided environmental modeling created predictions

for planning and decision making that were based on digital maps (Kraak, 2002, p.320).

The rst GIS functionality implemented solutions for the daily tasks of a map user,

like measuring distances (measure tool), or overlaying maps on a light table (layered vi-

sualization). What rst started as computerized cartography or computer mapping, grew

to powerful analytical and prescriptive GISs. The mathematical correctness of analysis

functions is seldom questionable, but the solutions to real-world problems are often not

congruent with the mathematically optimal solution. In the domain of resource plan-

ning, for example, the mathematically optimal distribution between rivaling agents will
2 THEORY 6

probably not be as well accepted as a solution created on the basis of consensus, spa-

tial reasoning, and dialog. The thematic maps generated by sophisticated GIS-functions

should play the role of an objective input into the dialog of the agents. It would be ideal,

if every participant of a resource conict had access to the same geoinformation. Sadly,

this is not feasible with today's GIS, because the professional skills needed to understand

and examine geoinformation with the system are too high. The solution must be the

eective communication of GIS results to GIS-laymen (Berry, 1993, p.203).

The step from technically feasible to socially acceptable options is not so much one

of increased scientic and econometric modeling as it is one of communication. (Berry,

1993, p. 203)

Understanding a map as a perceptive medium for spatially communicating decision

factors is emerging. Supporting a decision making processes with spatial geoproducts is

a matter of communicating the spatial situation or spatial scenarios to all participants.

Novel information technology gives room for many more ways than just transferring a

traditional descriptive map.

Many mapping and GIS applications have emerged in the last decades, but they

are still far away from an optimal communication with the map user. The concept

of the Digital Geographic Representation Framework (introduced in chapter 4) tries

to contribute a small step in this direction by creating a homogeneous, open-standard

framework from the producers of geoproducts to the general public.


2 THEORY 7

2.2 From coordinates to geoproducts


2.2.1 Coordinate Reference Systems
Positions on the Earth's surface can be described by systems of spatial reference (Robin-

son et al., 1995, p.95). Some systems are based on geographic identiers like country

codes or postal addresses, but more common and more digitally versatile are systems

that use a Coordinate Reference System (CRS). Coordinates are positions in a Coor-

dinate System (CS). A coordinate can be represented as a tuple, where the dimension

of the tuple equals the dimensionality of the CS. Per se, a CS (with all its possible

coordinates) does not have any reference to any position on Earth. By linking the origin

of the CS to one explicit position on Earth all coordinates of the CS become related to

an unambiguous position in the real world space (Robinson et al., 1995, p.97). The ad-

ditional information that links the CS to Earth is called a Geodetic Datum. It describes

the relationship of the 3D or 2D CS to Earth. The combination of CS plus Geodetic

Datum is a CRS (Romagnoli & Giannecchini, 2006; Bartelme, 2005, p.215).

The Earth is a very inhomogeneous body with a boundless level of detail. Often it is

useful to simplify the surface as if it were just a at, two-dimensional plane. This is the

case for the Gauss-Krüger CS much used in Germany in combination with the Potsdam

Datum. In its application area it approximates the Earth's surface as a two-dimensional

plane. This greatly simplies planning and visualization, but also creates problems of

imprecision and ambiguity.

On the other hand, other application domains need to dene coordinates in three-

dimensional space with precise distance from the Earth's center.

Trade-os must be made between the exactness of the CRS and the extend where

it can be used. Whenever a geodesist or cartographer plans to survey inside an area

or volume, they will select the combination of CS and Geodetic Datum that is most

suitable. Hence a great number of dierent CRSs is in use worldwide.

Where the application areas of multiple CRSs overlap, coordinates can be trans-

formed. A location in real-world can hence be represented in dierent CRSs. To trans-

form a coordinate from one CRS to another all parameters of both CS and their Geodetic

Datums are needed. When transferring coordinates, their metadata must therefore con-

tain a full mathematical description of the CS and the Geodetic Datum (Hake et al.,

2002).

Section 5.2.1.2 describes two ways for dening CRSs for the implemented prototype

framework.
2 THEORY 8

2.2.2 Geodata, geoinformation, and geoproduct


Coordinates come in many dierent forms. They can, for example, be gathered by the

physical sensors of a satellite as digital numbers. Before they are processed or utilized,

they are seen as geodata (Bartelme, 2005, p.10). By giving semantics to geodata, they can

be transformed to geoinformation. This can be done by structuring (normalization, rec-

tication, etc.), usage, or application (Bartelme, 2005, p.13). When information becomes

valuable and is sold on a market, geodata turn into an information product. Processing

and generation of geoinformation for a market (or to fulll an order) creates a geoproduct

(Bartelme, 2005, p.11).

A geoproduct is not as at as the geodata that it has been created from: Geoprod-

ucts are enriched with metadata that help to comprehend and use them correctly. For

geoproducts the metadata typically contain a description of the processing chain, the

methods or simplications applied, origins and timestamps of underlying data, author-

ship, legal information, and contact information. Communicating this metadata from

the maker to the user is important as it ensures that the facts visualized in the map are

not misinterpreted.

2.3 Digital Maps


2.3.1 Cartography
Man has been creating maps or map-like gures for a long time and it is impossible to

know when the rst map was made. Key data are cave wall drawings from between

10,000 and 20,000 years ago and Mesopotamian maps drawn on clay tablets dating back

about 4,500 years (Peterson, 2003, p.2).

Historically, the main challenge in cartography was the accurate representation of

physical features. A central function of maps was to facilitate the navigation in unknown

areas. For centuries cartographers had to cope with the task of exact georeferenciation

on the Earth's surface. Advances in chronometers and innovations like the Mercator

projection in 1569 slowly but steadily increased the accuracy of mapping (Hake et al.,

2002; Haggett, 2001, ch.21.1).

Georeferenciation has been extremely simplied by technological innovations like the

GPS, remote-sensing and computers. As described in section 2.2.1, today the default

method to store coordinates is in digital form. The map, which used to be a medium for

storage and representation at the same time, has shifted to a medium for visualization of

geodata. The geodata itself is not primarily stored in the map. This has led to a wider
2 THEORY 9

understanding of cartography (Wiesmann, 2007, p.8): Cartography is seen as an area

of expertise dealing with the collection, processing, storage, and analysis of geospatial

information with the purpose of visualizing it in the form of maps or other media (Hake

et al., 2002).

2.3.2 Maps
A map is created to represent a spatial aspect for a particular purpose. There are many

dierent types of maps due to the great diversity of purposes that they serve. Before a

cartographer decides what kind of map to create, he/she has to deal with the following

questions:

ˆ What is the purpose of the map?

Maps can be created for a vast variety of purposes like navigation, game hunting,

tax collection, inventorying resources, risk management, Land-Use and Land-Cover

Change (LUCC) and many other.

ˆ Who are the targeted users of the map?

Van Elzakker (2004, p.24) emphasizes that knowledge of the needs and character-

istics of the audience can be used to improve the eectiveness of map products. It

makes a great dierence whether a map is being created for primary schools, as an

evacuation map on an international airport, for stakeholders in Africa, or to depict

recent vegetation changes to a scientic audience.

One has to know: What is their level of geospatial literacy? How much do they

already know about the topic of the map?

ˆ Which part of the real world shall be represented by the map?

Maps can represent the whole globe - a world map - or just a small part of it.

ˆ Under which conditions will the map be used.

Will the users take the map into the eld? Should the map be dependent on

information technology (for example published on some digital medium)? How

much time will the map users invest in reading the map?

ˆ What is the budget and time-frame for the mapping project? How many copies

will be published at what price? Are there any legal restrictions when publishing

the map?
2 THEORY 10

Depending on the answers to these questions, a multitude of possible maps can be created.

The ultimate type of map does not exist (Haggett, 2001, Chapter 21.3).

2.3.2.1 Classication of maps Robinson et al. (1995, p. 11) argue that  there are

recognizable groupings of objectives and uses for maps, which permit us to catalog them

to some degree. They distinguish three criteria to classify maps:

ˆ Scale

The ratio between the dimensions of the map and those of reality is called the
extend of area on map
map scale: map scale = extend of area in reality . When a map presents a large area
on a much smaller area (e.g. piece of paper), the map is called a small-scale map.

Examples are maps of nations or continents.

A map which describes only a small part of the world is called a large-scale map.

An example is the land register that shows the real property of a housing estate.

There is no consensus on the quantitative limits of the terms small, medium and

large scale map (Robinson et al., 1995, p.12). A scale of 1:500,000 can be called

small-scale in Germany but large-scale in a remote area like Siberia.

ˆ Function

Every map is created with a function in mind (Haggett, 2001, ch 21.3). Maps can

be used in many dierent ways and new functions are still being discovered (An-

drienko et al., 2003, p.137). Therefore, a functional classication of maps cannot

create strictly separated classes that each map can be classied by. The borders

between the classes are oating. Having this inexactness in mind, Robinson et al.

(1995, p. 12) recognize three main classes of functional maps: topographic maps,

thematic maps, and charts.

Topographic maps portray the Earth's surface as accurately as possible (with re-

gards to the map scale). These maps are often standardized by national authorities

and have the validity of legal documents. An example is the topographic series of

Germany in the scale of 1:25000 (TK25) published under the authority of the fed-

eral states of Germany.

Thematic maps (or special purpose maps ) have the objective to represent the spatial

distribution of particular attributes (or the relationship among several attributes).

They usually contain basic topographic features in order to facilitate the map user's

orientation. Examples are maps of population density and precipitation as well as

bit-masks of cloud-cover for satellite images.


2 THEORY 11

Charts are a class of maps that are especially designed for the use by nautical and

aeronautical navigators. Chart series do not have a uniform scale. They focus on

details that are essential for safe navigation (e.g. lighthouses, shallows, towers)

(Francula, 2005, p.6).

ˆ Subject matter

Maps are communication tools which are used by many dierent application do-

mains. Grouping maps by their subject matter leads to classes like cadastre, eco-

nomic, or transportation maps.

2.3.3 Map Visualization


Visualization has always been an important tool to describe or communicate features of

real landscape.

DiBiase (1990) denes a model for map-based scientic visualization. It covers both

the communication and thinking functions of the map. He distinguishes between public

visual communication and private visual thinking. The traditional process of layout,

print, and distribution of maps is an example of public visual communication, because

here visualization is used to communicate to a public audience of map users.

Private visual thinking applies to situations where the users of a visualization software

 play with their data. Examples are trial and error or testing of spontaneous hypotheses.

2.3.3.1 Classication by use MacEachren (1994) introduced a map use based ap-

proach to map visualization presented as a cube which was later modied together with

Kraak (MacEachren & Kraak, 1997). The map use cube (gure 1) denes three aspects

of map visualization via the cube's three axes: Type of audience, level of interaction and

character of the data. The audience axis is used to describe the type of audience the

map is made for. Audience is dened from public (in the cube's origin) to private (at

the cube's end). The data axis dierentiates between the quality of data that are visu-

alized. Data (or relationships within data) can be known or unknown. The interaction

axis reects the level of interactiveness. A traditional paper map is not interactive. An

interactive map allows the user to change parameters of visualization like map scale or

order of layers (The role of interactiveness in maps is further discussed in section 2.6.2:

Multimedia and cartography).

Kraak (1999, p.159) positions the three visualization strategies of presentation and

exploration into the map use cube (gure 1). Using visualization for presentation, com-
2 THEORY 12

Figure 1: The Map use cube showing dierent visualization strategies. Adapted from
MacEachren & Kraak (1997)

municates facts which are well known to the map maker. The interaction is supposed

to be low because maps for presentation are still mostly presented on xed paper maps.

Presentation via map is generally generated for a public audience.

The exploration strategy is located at the opposing corner of the cube - the process is

more private, new insights into the data are still unknown and there is high interaction

with the audience because the map makers themselves are the main audience of the

visualization.

The development of faster computer graphics software and hardware has made com-

plex and highly realistic landscape visualizations possible. It has become a widespread

output medium for maps. Husdal (2001) argues that computer-based visualization does

not only allow to communicate known data to others: When tools and applications allow

researchers to visualize dierent aspects or relations of their data pool within seconds,

visualization becomes a tool for private thinking and exploration of unknown facts.
2 THEORY 13

According to Kraak (2002, p.320) the opportunities oered by hardware and software

development shifted the research eorts in map visualization from presentation-oriented

to exploration by the individual.

2.4 Digital Geographic Representation


Geographic representation is dened by Longley et al. (1999, p. 59) as  the construction

of a digital model of some aspect of the Earth's surface. Representations are versatile,

because they allow users  to learn, think and reason about places and time that are

outside their immediate experience.

The rst concepts of how to present entities of geography's object of investigation

digitally, dealt with the distinction between two perceptions of geographic objects: Either

as continuous elds or as discrete objects. For both approaches suitable digital geographic

representation models have been developed: Raster and vector. Raster is adequate to

represent features such as surface temperature or atmospheric pollution. Vector is more

suitable to represent discrete objects which can be constructed with points and lines like

trees, streets, or borders (Bartelme, 2005, ch.2).

None of the concepts is limited to features of the physical world. Raster can repre-

sent average real estate values and vector is able to represent administrative boundaries

(Casati et al., 1998, pp.79).

Both perceptions of geographic objects are interchangeable in the sense that any scene

can be represented as either raster or vector. A map of vegetation classes can either be

seen as a function that maps location to a nominal variable vegetation type (raster

view), or as a fragmentation of the area into non-overlapping, space-exhausting, discrete

areas (vector view) (Goodchild et al., 2007, p.241).

2.4.1 Separation of storage and visualization


A traditional map - whether printed on paper or carved into wood - stores and visualizes

the geo-spatial information it contains inseparably (Hake et al., 2002). This has been an

inherent feature of maps until the advent of the information age.

Kraak (1999, p.159) states, that  today a clear distinction is made between the

database and the presentation functions of a map. When stored digitally, data can

be structured to separate spatial data from its visual representation. This separation has

many advantages: Hardy et al. (2004, p.1) reports that many cartographic production

organizations tend to optimize their workow by maintaining a single geo-spatial model


2 THEORY 14

in a central database and by deriving a range of dierent cartographic products from

it. The information of how the products are rendered is stored product-specically and

separately. The database provides what to visualize, the visualization parameters store

how to visualize.

2.4.2 Cartographic representation rules


The information of how to visualize geo-spatial information is expressed by sets of carto-

graphic representation rules. These rules dene the way geo-features are rendered to an

output medium (Hardy et al., 2004, p.2). Cartographic representation rules can range

from simple:

ˆ Paint forest polygons in green with a black border of 1mm.

to complex:

ˆ Fill cemetery polygons with symbols of 2mm size. Depending on the religion at-

tribute of the cemetery use a tombstone, crescent, or star symbol. Align the grid of

the symbols along the angle of the long axis and rotate the symbols by this angle.

Suppress symbols that cross the boundary of the polygon.

When cartographic representation rules are comprehensively dened, updates of a central

geospatial database can be automatically transferred into cartographic products. This al-

lows automating the generation of maps. Trévisan (2004) mentions that once established

this approach saves budget and time in the map-making workow.

2.5 Communication via maps


Haggett (2001, ch. 21.3) describes the process of making maps as a process of encoding.

The map is seen as a mid-point in an information transfer which has a map encoding

phase (see gure 2) and a map decoding phase (see gure 3). The map maker's task is

to capture aspects of the real world into a map. The map user interprets the map and

tries to decode the information that was encoded into the map.

The two phases are linked by a transfer phase in which the map (or atlas) is being

distributed to the users. The ow of information is primarily one-way from the creator

to the user. Figure 4 visualizes how the processes of encoding, transfer and decoding are

linked.
2 THEORY 15

Figure 2: Map encoding phase. Adapted from Haggett (2001, S.677).

Figure 3: Map decoding phase. Adapted from Haggett (2001, p.677).

Robinson et al. (1995, p.17) consider the link between the creator and the user as

the central aspect of cartography:  We can liken cartography to a drama played by two

actors (map maker and user) with two stage properties (data and map) (Robinson et al.,

1995, p.19). The undisturbed interplay between makers and users is seen as the central

element for successful communication with maps. They dene a map's eectiveness as

the result of how well map makers and map users interplay with each other (see gure

5).

2.5.1 Map makers and the map encoding phase


The rst maps were rare and valuable treasures. To duplicate a map was a tedious and

slow process. In the late-1400s cartography started to use relief printing methods: The

original map was directly engraved into wood or copper by the cartographer. Once that

process was nished, copies of the original map could be printed to paper at reasonable
2 THEORY 16

Figure 4: Map communication

Figure 5: Map eectiveness in thought and communication. Adapted from Robinson et al.
(1995, p.17, gure 2.7)

speed. This allowed many more people access to maps (Crane, 2002; Peterson, 2007a,

p.68).

The introduction of printing-techniques at the beginning of the 15th century has been

compared with the impact of the Internet on map distribution at the end of the 20th

century by Peterson (2007a, p.68). Thousands of maps are distributed through computer

networks every second. As the maps are digital, their copies are perfect duplicates of

the original. The worldwide distribution is possible in a fraction of the time needed to

distribute physical maps on paper.

This impact also augmented the tools and technologies involved in cartography. In

the  analogue world, the distribution of a geoproduct deals primarily with printing

techniques. A map maker who wants to publish digital maps needs to deal with new

technical issues arising from multimedia and ICT.


2 THEORY 17

The map encoding phase (see gure 2) is dominated by the map maker. He/she has

to use a combination of techniques to encode a spatial situation or process into a map.

There is no single best way to encode a map. Every cartographer will create a somehow

dierent map because there is a number of decisions he/she has to make.

Basic parameters the cartographer has to deal with are :


2

ˆ Selection and Simplication ( Generalization)

A map always has a macroscopic function: It presents a large area of reality on

small extends. No map contains everything that exists in reality in the area it

represents. Therefore maps are selective in what they show. Sometimes multiple

symbols have to be replaced with only one to increase the readability of the map.

The process of selecting, simplifying or combining many symbols to fewer - but

more comprehensible - symbols is called cartographic generalization. Weibel &

Dutton (1999, p.126) describe the main objective of cartographic generalization

as to create maps of high graphical clarity, so that the map image can be easily

perceived and the message the map wants to deliver can be easily understood.

ˆ Classication

Exact representation of every feature's singularity is usually not the purpose of a

map as the user will be overburdened by details. The diversity of reality should

be classied into similar groups to make the information more accessible to the

map user. Statistical considerations of grouping and number of classes have to be

applied (Robinson et al., 1995, p.520, box 26.B).

ˆ Symbolization

To present the geoproduct in a meaningful way, the map user must be able to

identify and discriminate the symbols on the map. The selection and positioning

of symbols on a map is called symbolization. Good symbolization leads to less

information-loss in the encoding and decoding phase.

Symbolizing the basic graphic elements (point, line and area) can be structured
3
into adjusting the visual variables : form, size, orientation, and color . The com-

bination of symbols is called pattern and can be controlled by the position, texture

2 The toolbox of the cartographer is large and this section can not represent all the aspects involved

in creating an eective map. For an overview see Robinson et al. (1995).

3 Color can be further subdivided into the three separate variables hue, tone and chroma.
2 THEORY 18

Figure 6: The visual variables. Van Elzakker (2004, p.26).

and orientation (Robinson et al., 1995, ch.18). Figure 6 shows a comprehensible

overview by van Elzakker (2004, p.23).

ˆ Lettering

Lettering deals with positioning text labels in proximity to features on the map.

Too few labels make it hard to recognize features on the map - too many labels

make the map crowded and incomprehensible.

One location might have many names in dierent languages. Which language to

use in the map? Should the local name be used - so that one can nd a place by

the name that is written on a road sign - or the language of the targeted map user?

For an overview of functions and guidelines of lettering, see Robinson et al. (1995,

ch.22).

2.5.2 Map users and the map decoding phase


Everybody who just reads a weather map is a map user. The information encoded in the

map has been successfully transferred to the reader her when he/she opens the newspaper

and locates the weather column. At that point the map decoding phase begins (gure 3).

In his book Map Use: Reading, Analysis and Interpretation Muehrcke (2001) de-

scribes three sequential phases that constitute map use. Robinson et al. (1995, p.17)

refer to the same three phases in gure 5:


2 THEORY 19

1. Reading

According to Muehrcke (2001, p.17), to read a map means to translate its features

into a mental image of the environment. The rst task is to identify map symbols.

Then the reader of the map must make a creative eort to translate the arrangement

on the map into an image of the real world.

The reading of the map can be understood as a rst absorption of the elements,

such as the title, the legend, the colors, and the shapes. The map user will have to

understand the language used for legend and labeling. Foreign languages or formal

scientic terms can seriously hinder the decoding phase (Muehrcke, 2001, p.17).

To realize the geographic placement of the map is a basic step of map reading. If

one can not recognize the shape of a familiar country or the name of a location on

the map, one will have to interpret the coordinates given in the grid. The map user

has to have some  map-literacy to understand elements like scale-bar and north

arrow.

2. Analysis

The goal of map analysis is to reduce the amount of information of a map until it

can be understood (e.g. told to someone else) (Muehrcke, 2001, p.213).  Analysis

gives us descriptions, not explanations or interpretations (Muehrcke, 2001, p.214).

The analysis phase grasps the forms seen on the map and links them to things that

one can imagine existing at the mapped location.

3. Interpretation

While interpreting a map, a user will notice unusual or interesting patterns and seek

explanations for them. The answers to these questions are usually not found ex-

clusively on the map, but it includes clues or provides starting points for discovery.

 Intuition is an important part of map interpretation, just as it is in interpreting a

book, a poem, or a painting" (Muehrcke, 2001, p.431).

Map interpretation is a complex and creative act. Everything the map user has

learned about map reading and analysis will be put to use. In fact, everything

he/she has learned throughout his/her life will be helpful, because interpretation

requires an understanding of more than just the map. The map user must have

some background knowledge of the features depicted on the map (Muehrcke, 2001,

p.432).
2 THEORY 20

Whether the map user will be able to decode the information encoded into the map

depends a lot on his/her knowledge and abilities. The better the map makers know

their targeted user group, the more can the map be adapted to the skills, knowledge, and

experience of the audience.

Communication via maps always entails losses. The map image of the real world

which the map user interprets will never be exactly the impression of the real world

that the map maker intended to transfer (Haggett, 2001, 677). Muehrcke (2001, p.17)

emphasizes that there is a large gap between the two because  much of what exists in

the environment has been left o the map, while many things on the map do not occur

in reality".

2.6 Multimedia Atlas


2.6.1 Atlas - denition and history
The Atlas is a gure from the Greek mythology and describes one of the son's of the

titans Lapetos and Klymene. As a punishment for joining the titans in their ght against

the gods he has to carry the rmament on his shoulders ever since (Crane, 2002).

In cartography the term was rst used by Gerhard Mercator (1512-1594) in the title

of his work Atlas or cosmographic meditations about the creation of the world and its

shape
4 (1595). He used the term to refer to a holistic reection of the world (Crane,

2002; Bollmann & Koch, 2001).

The new term gradually gained recognition. In the 19th century an atlas described

a collection of map-sheets which dealt with some topic of the whole world. Since the

middle of the 20th century a dierentiation of atlases had begun: School atlases, home

atlases, and national atlases are just some examples (Cron, 2006a, ch.2).

Today the term atlas has lost some of the  global or  worldwide aspects and is

understood as an extensive and comprehensive information medium. It is perceived as a

thematic compilation of maps and semantically related information. Atlases can contain

(beside all kinds of maps) texts, images, charts, graphs, diagrams, statistics, etc. In

the 1990s the rst digital atlases were created on CD-ROM. This further moved the

perception of the term atlas into the direction of multimedia (Cron, 2006a, ch.2).

4 lat.: Atlas sive Cosmographicae Meditationes de Fabrica Mundi et Fabricati Figura


2 THEORY 21

2.6.2 Multimedia and cartography


The term multimedia stands for a sum of hard- and software technologies for the integra-

tion of digital media like text, pixel images, graphics, video, and sound. But this is only

the media focused aspect of the term - the multimediality of multimedia. Other aspects of

multimedia are interactivity, multitasking (simultaneous execution of multiple processes)

and parallelism (concerning the parallel presentation of media) (Issing & Klimsa, 1997,

p.3).

Keates (1982, p. 91) demands that  a map does not only contain a geographic

representation of some part of the world, but is also accompanied by a legend, description

with metadata and hints on how to interpret the map. The legend - even though of

fundamental importance - only provides very brief explanations of the map's symbols.

It should be extended with auxiliary information on topics related to the map. Written

in 1982, Keates could not have anticipated the role that multimedia can play today to

provide this additional metadata.

Golledge et al. (2006, p. 182) state, that  the information ow from a computer to a

learner is less vulnerable to shifts in attention of the user as the information is presented

in dierent modalities. This amplied attention by the user is needed to eectively

communicate more complex metadata.

Cartwrigth (2007, p.29) concludes that  the challenge for map designers and produc-

ers is to use multimedia, in its broadest sense, as a new tool for cartography. Francula

(2005, p.6) proposes that multimedia shall be interlinked with digital cartographic prod-

ucts to provide an  additional level of information and a more complete view of reality.

He agrees that multimedia is an ideal carrier for context information and will help map

users not only to decode the information encoded into a map, but also to get a deeper

understanding of the presented topics.

Using multimedia, a map on water infrastructure in Africa could be accompanied by

an introductory text, a digital library with related studies and scientic papers, videos on

how to manage water pumps, as well as georeferenced precipitation statistics and photos.

The list of potential add-on information to a map is as manifold as its purpose.

Multimedia is the most promising concept to successfully communicate this additional

information.
2 THEORY 22

Figure 7: An animated map: Sim City game screenshot.

2.6.3 Maps in Multimedia Products


According to Ormeling (2007), there is no single best way to use maps in multimedia

products. They can fulll many dierent functions, just as the nal product can have

many dierent purposes.

In the following, two examples of maps in multimedia products are briey introduced

to show the wide range of possible applications.

2.6.3.1 Sim City One example that reects the diversity of maps in combination

with multimedia is the city-building computer game Sim City (Ormeling, 1995). A

screenshot of the game is shown in gure 7.

Maxis Software released the rst version in 1989 and their founder Will Wrigth de-

scribed it as a  digital playground. The game starts with a view on an empty landscape

which then has to be developed to a ourishing city. The user moves around the area
2 THEORY 23

using a cursor but the player himself is never visible. One can can build housing, employ-

ment, and city infrastructure with a budget that is linked to the number of taxpayers in

the growing city. The map visualization is both interactive (building sites can be located

and houses can be clicked) and dynamic (the game is played at a specic speed and city

growth and trac is constantly animated).

Ormeling (1993) sees this  digital playground as one of the earliest  animated maps.

2.6.3.2 Atlas of Switzerland 2 The Atlas of Switzerland (2001) and Atlas of

Switzerland 2.0 (2003) have been extensively discussed in cartographic research as a

prominent example of multimedia atlas products. Examples are: Sieber & Huber 2007;

Hurni & Räber 2004; Huber & Schmid 2003; Huber et al. 2003; Bär & Sieber 1999; Cron

2006b; Bär & Sieber 1997.

The Atlas of Switzerland 2.0 (AoS2) combines high quality thematic maps and

panoramic 3D views with a number of modularized analytical functions. Basic features

are (Bundesamt für Landestopograe, 2004):

ˆ Integrated 2D and 3D maps, both with interactive layer concept

ˆ CD-ROM and DVD edition

ˆ More than 100 map topics

ˆ Analytical tools for 2D maps

ˆ Multimedia elements: 600 text panels, hundreds of pictures, sound and video

ˆ Image export and printing

ˆ Content update via Internet

Swisstopo, the Federal Oce of Topography of Switzerland, published 20,000 copies of the

rst edition. The second edition reached 6000 copies within the rst year of publication

(Sieber & Huber, 2007, S.162). The project is a joint venture by dierent academic and

administrative groups, including the Atlas team ETH Zurich, sowisstopo, the Duplex

Design software company, 50 data owners mainly from public adiministarion and more

than 20 professional translators (Sieber & Huber, 2007, p.168).

Design concepts and descriptions of new functions for the third edition are currently

under development. A prototype GUI for the Atlas of Switzerland 3.0 has been concep-

tualized and implemented by Cron (2006a).


2 THEORY 24

First level Second level theme groups in the Atlas of Switzerland 2.0

Base Map Cities, Villages, Lakes, Rivers, Glaciers, Railway Lines and

Stations, Cable Car, Roads, Forest, Administrative

Boundaries, Mountains, Passes, Contour Lines, Relief,

Satellite Image

Nature and Weather and Climate, Geology and Resources, Soils, Water,

environment Ice and Snow, Landscapes, Vegetation, Fauna

Society Population, Languages, Religion, Education, Health,

Households

Economy Labor, Agriculture, Industry, Services, Tourism,

Construction, Habitation

State and Politics Spatial Units, Referendums, Elections, Public Finances

Switzerland and Nature and Environment, Society, Economics, State

Europe

Table 1: Theme groups in the Atlas of Switzerland 2.0. (Hurni & Räber, 2004, p.4)

According to Huber & Schmid (2003, p.1398), the rst edition of the atlas was planned

as a  dual concept with a print and a multimedia version. This concept was never

realized and was dropped in favor of multimedia.

Sieber & Huber (2007, S.162) describes the metaphor used to structure the GUI as

 the atlas as a story book. The GUI was designed with the idea to make users feel

comfortable - whether they are layperson or scientist. The map functions as a main

structure for journeys through the map. To ensure that the users do not get lost, they

can always undo the steps they took (Sieber & Huber, 2007, S.162).

This concept corresponds to the idea of the map as a spatial organizer in multimedia

products (Ormeling, 2007, p.107). The area covered by the maps is the frame that keeps

the multimedia content together. Via geographical linkage all contents of the product

can be reached.

Additionally, the content of the AoS2 is explorable by a thematic structure and a

search functionality. The atlas themes are divided into six thematic domains which are

then subdivided as shown in table 1.


3 THE INTERACTIVE IMPETUS DIGITAL ATLAS 1.0 25

3 The Interactive IMPETUS Digital Atlas 1.0

The following chapter will introduce the interdisciplinary Integratives Management-

Projekt für einen Ezienten und Tragfähigen Umgang mit Süÿwasser (en.: Integrated

approach to the ecient management of scarce water resources) (IMPETUS) project and

how their repository of research ndings and geo-spatial data sets has been distributed

in form of the rst IMPETUS Interactive Digital Atlas (version 1.0) (IIDA1) (3.2).

Shortcoming during the usage of the software and while creating a new release of the

atlas are summarized.

3.1 IMPETUS
The IMPETUS (Integratives Management-Projekt für einen Ezienten und Tragfähigen

Umgang mit Süÿwasser (en.: Integrated approach to the ecient management of scarce

water resources)) project is a collaborative research project of the Universities of Bonn

and Cologne, in co-operation with partners in West-Africa. IMPETUS is part of the


5
GLOWA framework initiated by the Bundesministerium für Bildung und Forschung (en.:

Federal Ministry of Education and Research) (BMBF). The goal is the investigation of

the changes of the hydrological cycle and their impact on the ecosystem and the socio

economic conditions. The interdisciplinary investigations are carried out within two

catchments in Northwest and West Africa, namely of the river Drâa in the southeast of

Morocco and of the river Ouémé in Benin.

Tropical West Africa has suered from a prolonged drought since the early 1970s.

It has aected all climatic zones, from the semi-arid Sahel down to the humid Gulf

of Guinea. The lack of rainfall has had deteriorating eects on the economic and so-

cial development of the West African countries among which are Benin and Morocco

(IMPETUS, 2003, p.10).

The variations of the hydrological water cycle of the two selected catchments and the

socio-economic implication are investigated interdisciplinary from four dierent perspec-

tives: alimentation and subsistence, hydrology, LUCC, and society (IMPETUS, 2003,

p.1).

Since the start of the rst IMPETUS project phase in 2000, a great number of geospa-

tial and socio-economic data, as well as documents and reports, have been aggregated.

5 The GLOWA project analyses the global change and the hydrological cycle. GLOWA project website:

http://www.glowa.org
3 THE INTERACTIVE IMPETUS DIGITAL ATLAS 1.0 26

Figure 8: Interactive Digital IMPETUS Atlas 1.0

Many have been collected and processed, others are the results of the many IMPETUS

models. This interdisciplinary and long lasting repository holds great potentials, not only

for other researchers, but also for stakeholders on regional and national level in Benin

and Morocco. Since 2004 IMPETUS has addressed structuring and facilitating access

to this pool by creating the IMPETUS metadatabase Search


6 (IMPETUS, 2006, p.360).

This metadatabase helps to link researchers among each other, but is not designed with

a focus on reaching the local stakeholders which often have more practical interests in

the data.

3.2 The IMPETUS Interactive Digital Atlas (1.0)


In 2005 the IMPETUS Interactive Digital Atlas (version 1.0) (IIDA1) product (gure

8) has been compiled at the Remote Sensing Research Group (RSRG) at the University

of Bonn. It is an interactive, digital multimedia atlas, combining the presentation of

6 IMPETUS public metadata search portal: http://www.impetus.uni-koeln.de/en/data.html


3 THE INTERACTIVE IMPETUS DIGITAL ATLAS 1.0 27

geo-data with related documents and geo-referenced photographs. Contents have been

provided by the IMPETUS sub-projects and the software has been developed exclusively

for the IIDA1.

IMPETUS followed a parallel approach with a printed and a digital version with

similar content. In comparison to the printed version, which costs about 25¿ per piece,

the digital version can be replicated at no cost.

The user interface is simple and intuitive: The user can select a topic or region of

interest and will be oered corresponding maps and articles. The digital atlas has made

it possible to transfer the research ndings of the IMPETUS project to people with no

special GIS knowledge. It also intends to avoid the unfavorable situation where costly

created geo-data is left unused after a project has been completed (IMPETUS, 2006,

p.366).

The IIDA1 was inspired by the Atlas of Natural and Agronomic Resources of Niger

and Benin, developed in 1999 by the Institute of Soil Science and Land Evaluation of

the University of Hohenheim, Germany. Their basis for the development was that the

accessibility of scientic data for management purposes, rather than [its] mere existence,

is often a restriction for scientists, decision-makers, and extensionists in developing coun-

tries (Stahr et al., 1999). The Atlas of Natural and Agronomic Resources of Niger and
7
Benin is available online and on CD-ROM and delivers simple thematic maps and socio-

economic data. The biggest limitation is that only one layer can be viewed at a time

and there exists no interactivity like zooming, panning or selection of geo-features (Stahr

et al., 1999).

The main component of the IIDA1 software is a minimal GIS for presenting geo-data

to map users. Meta information for geospatial layers can be shown as Hypertext Markup

Language (HTML) pages with images and text. For selected point-layers, the linkage

of points to HTML pages is possible. Additionally the IIDA1 provides the user with a

structured collection of related scientic articles. The IIDA1 software can be executed
8
from CD-ROM or accessed as a Java Applet online . Since the IIDA1 is written in the

Java programming language, its use is not restricted to a special Operating System (OS)

like Windows, Mac OS or GNU/Linux (Thamm et al., 2006; IMPETUS, 2006, p.366).

7 Atlas of natural and socio-economic resources of Niger and Benin online: https://www.
uni-hohenheim.de/~atlas308/startpages/page2/english/content/title_en.htm
8 IIDA1 online: http://www.impetus.uni-koeln.de/impetus.php?show=En_Be_At
3 THE INTERACTIVE IMPETUS DIGITAL ATLAS 1.0 28

The rst version of the IIDA1 was replicated approximately four hundred times on

CD and distributed to stakeholders and decision-makers. The concept of the IIDA1

proved to be fruitful and in demand (Thamm et al., 2005a,b).

3.2.1 Shortcomings of atlas viewer software


After a rst analysis of the existing source code, the following diculties have been deter-

mined: The code is not following the concept of Object Oriented Programming (OOP),

thus seriously limiting the extensibility of the program. Implementing international-

ization and localization would require widespread changes throughout the whole source

code. Additionally, the upgrade from the outdated GeoTools Library 1.x to its second

generation implicates many fundamental changes. Logic and content are interwoven in

the existing design. Clear separation of logic (algorithms, GUI, etc.) and content (maps,

articles, time-series, reports) is an important component in facilitating the re-use of the

software. These aspects lead to the conclusion that the development of the new atlas will

have to be started from scratch - but not without taking advantages from the experiences

gained.

The following list shall detail some of the shortcomings, suggestions, and new ideas

that have been noticeable during more than two years of experience with the atlas prod-

uct.

ˆ Improved usability: The selection and presentation of the published information

can further be improved. E.g. the IIDA1 provides the user with a very rigid way

of layer ordering: Every layer has a up and down button to move it in the legend.

When the button is pressed, the layer moves and another layer appears under the

mouse cursor. This has proven to be confusing for many users which are anyway

not familiar to the concept of GIS layers.

ˆ Multimedia: The atlas viewer software can display images and Portable Document

Format (PDF) documents (only on Microsoft Windows). Videos, animations and

sounds can not be played back.

ˆ Large raster: When displaying a raster, the IIDA1 software loads the whole image

into the system memory. For high resolution images


9 - as used in remote sensing

- this can easily exceed the system's available memory. Additionally, sub-sampling

9 To be preciser: Not the resolution, but the extend of pixels by width and height as well as the color

depth determine the needed amount of memory.


3 THE INTERACTIVE IMPETUS DIGITAL ATLAS 1.0 29

these images for viewing in lower resolutions (zoom out) is a big computational

eort that can interrupt the usage of the atlas for a noticeable time. A solution to

both problems would be support for viewing pyramid raster les. Pyramids allow

to load only the data at the specied resolution and level required for the display

into the memory (Shirley, 2005).

ˆ Raster queries: Imagery is internally managed as rasters of colors and not as rasters

of geophysical values. Therefor it is for example not possible to query a Digital

Elevation Model (DEM) for the height at a coordinate because the software is only

aware of the color at that position.

ˆ Internationalization and localization: The IIDA1 doesn't follow a separation of data

and logic which is needed for exible internationalization and localization support.

It was created with an English user interface. As the language is hard-coded into the

Java code, it is obligatory to recruit a programmer for every translation (Conner,

2007).

ˆ Export: An export feature could improve the application of the atlas, making it

not only a tool for presentation, but also a tool for archiving geographic output.

ˆ Printing: The IIDA1 doens't allow the users to save or print the views they have

created.

ˆ Simple analysis: The possibility to do basic spatial analysis and queries has been

requested be users. One possibility would be a attribute based lter that could be

used to display only geo-objects with special properties.

3.2.2 Diculties lling the IIDA1 with data


Two years after its introduction, the content of the atlas was partly outdated. New maps

and scenarios have been created and more papers have been published. The attempt to

update the existing atlas brought additional diculties to light. Even though the atlas

provides some basic conguration les to restock its layers, it is still very static software.

Some of the disadvantages when altering the content are:

ˆ Changing or adding the legend of a layer requires creating it manually as an image

le, implicating the use of a third-party graphics software like Adobe Photoshop.
3 THE INTERACTIVE IMPETUS DIGITAL ATLAS 1.0 30

ˆ Only three band RGB imagery can be imported. A raster layer with geophysical

values has to be converted to a three band color layer rst, thereby loosing it

geophysical information.

ˆ All geo-referenced layers have to be converted to the CRS of the digital atlas. A

better approach would be to enable the software to internally convert CRSs. The

user could then choose an arbitrary projection for displaying the content.

ˆ Publishing the atlas on CD-ROM requires knowledge of the Java Virtual Machine

(JVM).

Summing up, updating the IIDA1 with new content and the creation of frequent releases

has shown to be more dicult and time-consuming than expected. It often requires the

assistance of a Java programmer who is familiar with the code of IIDA1.


4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 31

4 The Digital Geographic Representation Framework

ˆ How can this ever-growing amount of spatially referenced digital information be

optimally used for the benet of a broad part of mankind?

ˆ How can spatial information (geo-products) be eectively distributed (sold) to a wide

range of users (consumers)?

ˆ How could a concept for storing and distributing such data and all its metadata look

like?

These are the motivating questions which drive this thesis. These questions are not

referring to the content of any specic atlas product but aim at the geoinformatic tools

needed by cartographers to fulll their mission.

When Robinson et al. (1995) position the map's eectiveness between the map-

making and map-use process (see gure 5), they emphasize that eective map com-

munication has to incorporate both sides with equal attention.  Mapping eectiveness is

best achieved by treating the making and using of maps equally (Robinson et al., 1995,

p.17). Map makers and map users are the main actors in cartography. Communication

between them is a central element which controls the map's eectiveness. Therefore a

concept for eective communication of multimedia mapping products should concern a

design matching of both side's software tools.

This chapter introduces a conceptual framework for spatial communication in the

form of standardized interactive multimedia atlas products. Cartographers need a system

which provides a more comprehensive support for the whole process of communication.

This framework shall help makers and users to bridge the technical barriers of the digital

publication workow. The Digital Geographic Representation Framework (DGRF) is a

conceptual approach to create such tools.

The DGRF spans the workow from the import of geoproducts up to the consumer

who interacts with the atlas product. It assists in three sequential phases of atlas com-

munication:

1. Phase: Import, structuring and interlinkage of maps and multimedia to create an

atlas product.

2. Phase: Packaging and distribution to transfer the product to the consumers.

3. Phase: Visualization and interaction on the map user's computer.


4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 32

Figure 9: Digital Geographic Representation Framework

# Phase DGRF module User realm

import, structuring and


AAT
1. interlinkage (content map makers
(section 4.1)
management)

Atlas Package
2. transfer both
(section 4.2)

visualization and AVT


3. map users
interaction (section 4.3)

Table 2: Atlas communication phases and corresponding DGRF modules.

The DGRF consists of two software tools: Atlas Authoring Tool (AAT) and Atlas Visu-

alization Tool (AVT). They provide the GUIs to interact with map makers and users.

The assumed target group consists not necessarily just of experts, but also of interested

laymen users who have an interest in the topic covered by the atlas. They are not ex-

pected to have special knowledge in GIS usa, remote-sensing or cartography. Table 2

shows how they correspond to the dierent phases of atlas communication. Figure 9

shows the DGRF embedded between map-encoding phase and map-decoding phase. The

transfer of the atlas - in form of an atlas package (4.2) - is encapsulated by the AAT

(4.1) and the AVT (4.3) so that neither the creators nor the consumers have to deal with

technological details.
4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 33

4.1 Atlas Authoring Tool


The need to communicate spatial information has undergone profound change over the

past decades. Whereas in the 1980s, the diusion process of GISs was still in its up-rise,

today it has become hard to nd an institution or organization that is not using some

type of GIS.  The change is stimulated by a ood of new georeferenced data combined

with new scientic and societal demands and uses for these data along with rapidly

evolving geoinformation technologies (MacEachren, 1998, p.1).

More sciences and professions start to analyze phenomena in a spatial context. Hy-

drology, market research, medical scientists, municipal service providers, planers, police

and even recreational hikers start to adapt GIS approaches to solve problems. They all

create or analyze geodata and accumulate them in their local databases. Many of these

actors would like to use modern technologies to communicate (or sell) their geoproducts

to the their respective audiences (customers) - but technological hurdles make their map-

ping eort costly, time-consuming and complex. Empowering  part-time cartographers

to publish their messages eciently via modern mapping technologies is a primary goal

of the AAT design. It provides a GUI which supports atlas creators in the rst two

phases of atlas publication: Compilation and packaging.

In brief the workow of using the AAT is as follows: The author imports geo-spatial

data sets (4.1.1) and denes how the layers shall be visualized (4.1.2). Additional mul-

timedia can be added (4.1.3). Content management (4.1.4) allows to dene thematic

structures and relationships between geo-spatial and multimedia contents. For multi-

lingual atlases, the AAT assists the creator to translate the contenta (4.1.5). Finally the

AAT manages the export to the atlas packages format, further described in section 4.2

on page 37.

4.1.1 Geo-spatial Import


The job of the DGRF begins when the building blocks of the atlas are available. Ap-

plicable geo-products have to exist prior to the use of the AAT


10 . It allows to import

geo-products in two dierent ways: File import and Spatial Data Infrastructure (SDI)

import.

10 The AAT is not a spatial analysis or manipulation tool. For previous processing of spatial data,

third party GIS and remote sensing applications provide the appropriate tools.
4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 34

File import Geo-spatial data les can only be imported into the AAT if the le format

is compatible. If the format is unknown, the les might be convertible with the help of

the open source Geospatial Data Abstraction Library (GDAL) tool


11 . At minimum the

following geo-spatial formats are supported by the AAT:

ˆ ESRI Shape le

ESRI Shapele format is an openly published vector data format which has become

sort of a a lingua franca for GIS vector data. It consists of three (optional four)

separate les. The le ending with .dbf is a DBase/FoxPro database and contains
attributes of the geo-objects (Erle et al., 2005, p.337).

ˆ GeoTIFF

GeoTIFF
12 is a TIFF based interchange format for georeferenced raster imagery.

GeoTIFF was initially developed by Intergraph in the 1990s and since evolved

to a widespread open standard for remote-sensing (Erle et al., 2005, p.470). It

extends the TIFF image format by storing geo-spatial meta information tags (like

georeferenciation and projection) inside the image le (Bartelme, 2005, p.130).

ˆ Standard image + projection le + world le

The combination of a standard (non spatial) image format (e.g. JPEG, TIFF,

GIF, PNG), a world le which denes georeferenciation, and a projection le may

be imported as a fully qualied raster image.

The world le is a plain text document containing the parameters: x scale, y

rotation, x rotation, y scale, x oset and y oset needed to transform the image

coordinates to real world coordinates (Erle et al., 2005, p.159).

The format of the projection le may be European Petroleum Survey Group

(EPSG) or Well-known Text (WKT), both further described in section 5.2.1.2. Its

le name ending is typically .prj (Erle et al., 2005, p.159).

Spatial Data Infrastructure import The AAT can participate in SDIs which are

based on OGC standards. It can connect to Web Feature Services (WFSs) and Web

Coverage Services (WCSs) and allows atlas authors to browse the geo-spatial data sets.

11 GDAL project webpage: http://www.gdal.org/


12 GeoTIFF discription at RemoteSensing.org webpage: http://www.remotesensing.org/geotiff/
geotiff.html
4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 35

They may then select subsets which are downloaded and imported into the atlas as

content.

4.1.2 Cartographic Representation Rules


As described in section 2.4.2, handling of geo-spatial information is separated into  What

to visualize and  How to visualize . The AAT provides two complementing ways to

handle cartographic representation rules:

Import Most GISs - like ESRI ArcMap - already include modules to dene the vi-

sual appearance of layers (Minami & ESRI, 2000, p.167). The AAT will therefore provide

ways to import cartographic representation rules created in external GIS applications.

Unfortunately, most GISs store their cartographic representation rules in proprietary

formats which do not follow open standards. Hence - without the cooperation and interest

of GIS companies - it is likely that the AAT will not be able to import from a wide range

of proprietary GIS products.

Visual editor The AAT shall not be dependent on any GIS product to create or

change the appearance of the layers. Perhaps atlas authors have just bought some geo-

spatial data and want to add it to an atlas. Thus the AAT provides a visual editor to

dene cartographic visualization rules, e.g. how the layers shall look like. The editor

must be able to modify all visual variables (gure 6) for vector and raster data.

Because the user can always change the scale of any map by zooming in the AVT,

the resulting maps don't have xed scales. Hence the cartographic representation rules

need to be scale-aware in order for selective generalization rules to be applied on-the-y

when the scale is changed.

Internally the AAT stores the cartographic representation according to open stan-

dards to achieve the highest degree of re-usability and connectivity.

4.1.3 Multimedia Import


Apart from geodata and their visualisation rules, the AAT can also import various other

media formats: videos, documents (PDF and Hypertext), sounds, and images. These

media elements can then be linked with the content management component (see next

section).
4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 36

A special task of the multimedia import is to ensure that the AVT will play back all

media on the map user's computer, despite it might run a dierent OS and no special

media player software can be presumed. When importing multimedia les, they must be

tested and converted to a compatible format if necessary. Especially for video formats,

platform interoperability is a critical task (see 4.4.1.1).

4.1.4 Content Management


Creators of an atlas must be given the facility to structure content and describe relation-

ships between multimedia entries and geo-spatial contents. These settings will aect the

GUI and interactiveness of the resulting product.

For thematic classication of contents the AAT provides tree structures that can be

arbitrarily changed by the author to reect any taxonomy of topics.

The following types of relationships can be expressed with the AAT:

ˆ Topics ↔ Multimedia

ˆ Topics ↔ Maps and Layers

ˆ Maps ↔ Multimedia

ˆ Geo-spatial objects ↔ Multimedia

On the basis of these relationships the GUI is dynamically adapted by the AVT.

4.1.5 Internationalization
Atlas internationalization refers to a design which facilitates the interchangeability of all

language specic aspects of the atlas product.

Software internationalization of the DGRF is the duty of the tool programmer and

discussed in section 4.3.3.1, but internationalization of atlas content and metadata is the

atlas author's task. When he or she decides to publish a multi-lingual atlas, the AAT

has to ensure that the translations are coherent and no label, text, sentence etc. is left

out before the product is published.

From the AAT's perspective, every atlas is translated into at least one language. This

implies that column names, legend, or labels are not limited by restrictions inherited from

their data sources. For example the ESRI Shapele database limits its column headers

to a maximal length of eleven characters. In such a case the AAT asks for translations
4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 37

of the column headers. This allows arbitrary length and the use of special characters to

properly describe the attribute column. Its a strong design aspect of the DGRF that

every widget
13 of the atlas can be labeled in a language that is most understandable for

the targeted users.

4.2 Atlas Package Format


Packaging is the nal step in the communication chain of the DGRF before the atlas is

transferred to the user. The AAT enables the author to generate atlas packages suitable

for distribution via various digital media. The reason to package the atlas is to generate

a single transferable atlas bundle that merges the content and the AVT application.

The atlas package format is compressed to minimize the size of the package. Further

aspects of the package format are exibility, integrity and security:

4.2.1 Flexible distribution


When compilation and layout of an atlas are nished, it has the be distributed to the

targeted users. This can be done via the Internet (online) or on portable media like

CD-ROM, DVD-ROM, memory stick, etc. (oine).

Distributed via the Internet the atlas product can reach a large community. By the

end of the year 2004, there were 935 million Internet users worldwide, which corresponds

to nearly 16% of the world's population (Peterson, 2007b, p.38). Providing atlases in

form of an Internet service - also called Atlas Information System (AIS) (Ormeling, 1998;

Schneider, 1999) - is unbeatable in up-to-dateness and reachability. Any changes to the

online map repository will directly aect the user's maps. Internet atlas users can be

sure that they are always looking at the latest version of the atlas.

Cartwright et al. (2001, p.13) emphasize that it is necessary to understand how users

interact in real world situations with geo-spatial visualization services and tools. An

online atlas allows to monitor the users and can thereby provide statistics to improve

layout and content of the product. Optionally the medium oers the possibility to restrict

access to the product with authentication mechanisms.

Herzog (2003, p.117) points out that the  World Wide Web oers cartography an

ideal platform for making communication with maps more feasible. He further accuses

cartography that it  has not taken sucient advantage of the Internet (Herzog, 2003,

p.129).

13 Widgets are the components of a GUI. The name is a contraction of  window and  gadget.
4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 38

Figure 10: Global distribution of the Digital Opportunity Index. ITU/UNCTAD (2007)

The atlas package format can be copied onto a standard web server. No installation

of services on the server is needed. The atlas (strictly speaking the AVT) can then be

accessed via any Java enabled web browser. A broad band Internet connection is required

by the users.

Figure 10 shows a map by the International Telecommunication Union (ITU), rep-

resenting the distribution of the Digital Opportunity Index (DOI)


14 . The DOI is an

index which reects a country's state of availability and accessibility to ICT. Dark colors

represent relatively good chances in the knowledge society, light colors represent poor

chances and correlate with low gross domestic product and less than 0.01 PC per capita.

The map shows that a major part of the world is not joining the transformation to a

 global village.

Users in the US (186 million), Germany (42 million), and Japan (78 million) make

up one third of all Internet users worldwide (Peterson, 2007b, p.38). Hence the global

distribution of ICT users is not uniform.

Oine distribution increases the chances to reach people without (or with limited)

Internet access. Also in-eld use on laptops becomes feasible.

Any digital media can be used for lossless distribution. Portable media like CD or

memory stick provide space for large amounts of data. A double-layer DVD can store

14 For more information on the DOI see International Telecommunications Union (ITU)'s Telecommu-

nication Development Sector webpage: http://www.itu.int/ITU-D/ict/doi.


4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 39

up to 8,5 gigabyte. If the atlas size exceeds a few hundreds megabytes, transfer via

the Internet will hardly be possible and oine distribution obligatory. Additionally, the

distribution of the atlas on a physical medium allows a more haptic handling of the

product like giving it personally to someone.

An atlas package can be copied to any digital media without special modications.

When the medium is inserted into a computer the atlas can be started locally without

the need for an Internet connection.

Cartwright et al. (2001, p.1) state that exible access should be provided to geo-

spatial services. Such exibility is achieved by following a hybrid online-oine approach

for the atlas package format. It allows the map maker to publish the same atlas either

online or oine.

4.2.2 Integrity
The integrity of an atlas package must be veriable. This allows to control whether the

package has been manipulated during the transfer to the user. The AVT veries the

validity of itself and its atlas content before starting the GUI. This mechanism ensures

that the atlas package has not been infected by any malicious code
15 .

4.2.3 Protection
Creators need a way to protect the content of the atlas for legal or other reasons. The

package format provides cryptographic mechanisms, which restrict the use exclusively to

visualization by the AVT.

4.3 Atlas Visualization Tool


The map decoding phase (see gure 3 on page 15) begins when the users receives (oine)

or accesses (online) the atlas product. They then have to decode the messages and

situations which the map maker encoded into the product. For that task they need a

user friendly software tool that visualizes digital maps and allows interaction with the

atlas content: the Atlas Visualization Tool (AVT). This human-machine interface
16 is

15 Malicious code is an umbrella term for computer viruses, Trojans, spy-ware and any application that

acts against the interest of the user.

16 The human-machine interface (or user interface) is the aggregate of means by which people (the

users) interact with a computer program. User interfaces can be textual or graphical. A GUI is a

graphical user interface that allows the user to interact with a program using a mouse and graphical
4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 40

an essential element in the communication ow. It provides the GUI through which

the atlas is used. If the GUI can not be eciently operated it essentially disturbs the

communication ow.

Cron (2006b) created a guideline for the implementation of interactive atlas GUIs.

She states that the design of GUIs for interactive atlases or maps is being approached very

dierently world wide (Cron, 2006b, p.9). No comprehensive compilation of requirements

or guidelines existed yet. Her work is based on the evaluation of twelve digital atlas GUIs

(six online atlases and six oine atlases). In (Cron, 2006a) she additionally implemented

a prototype of a GUI, which is used as a guideline for the third version of the Atlas of

Switzerland (AoS). While most aspects of her work are focused on the AoS use case,

some of it is highly applicable for the AVT.

The design of the AVT prioritizes its ability to visualize geo-spatial and multimedia

content (see 4.3.1), provide interaction between user and map (see 4.3.2), and do both

with a high level of adaptability and usability (see 4.3.3).

From a user's perspective the AVT GUI represents the multimedia atlas product. The

internal separation between data and tool is not visible to the user. Still it is essential to

the concept of the DGRF that the AVT can be used to visualize many dierent atlases.

4.3.1 Visualization
Atlas content can be distinguished into two types: geo-spatial content and multimedia

content. The AVT is responsible for the visualization of both.

4.3.1.1 Geo-visualization The main function of the AVT is to provide the geo-

visualization to present the dierent maps of the atlas. Such a geo-visualization tool - in

a broader sense - is part of every multimedia atlas, GIS, and remote sensing application.

Geographic Visualization (GViz) research investigates the design of such tools to

facilitate visual thinking, pattern recognition, and hypothesis generation (Kraak, 2002;

MacEachren et al., 1998, p.1). In terms of the classication by Kraak (1999, p.158)

(see 2.3.3) the AVT focuses on presentational functionality. Analytic and exploratory

functions can be added on special demand but are of lower priority. The AVT's geo-

visualization component is rather a tool for  public visual communication than for  pri-

vate visual thinking (DiBiase, 1990).

feedback. It is the standard for end-user software design since the introduction of Microsoft Win-

dows 3.11 and Mac OS 1.


4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 41

The AVT presents two-dimensional maps


17 which are combinations of vector and

raster layers. A core feature of the AVT is the visualization of very large raster images

on standard personal computer in reasonable time. This allows to present high resolution

remote sensing products with virtually no limitation in size.

To help novice users it is important to maintain similarity to traditional paper maps.

Standard map elements - orientation of the map (north arrow), the map scale (scale bar)

and the explanation of the symbols (legend) - are present. Grid bars at the border of

the view-port inform the user about the geographical extensions of the visible part of the

map.

The maps presented in the atlas are integrated into a thematic context. Meta infor-

mation and multimedia related to the topic of the map are accessible.

4.3.1.2 Multimedia visualization Some versions of OSs ship with media players

and codecs
18 installed to visualization the basic types of multimedia. In other cases the

user might have installed a missing player software or removed another. The task of

the AVT is to search for an appropriate way to playback the atlas content on any user's

computer.

4.3.2 Interaction
The AVT allows users to interact with the atlas content interactively. Wiesmann (2007,

p.19) states that users of ICT are aected by their experiences from the Internet and

expect a universal  clickablility. Therefore the mouse must be the primary interaction

device for the GUI. Besides the standard interaction widgets of computer GUIs (menu

bar, buttons, drop-down lists, tool-tips) the AVT features the following interactivity for

maps:

4.3.2.1 Spatial navigation Spatial navigation can be conceptualized as the hovering

of a view-port over a map (Wiesmann, 2007, p.23). The panning function moves the view-

port over the map without changing the scale. To enlarge the view-port the user has

17 Three-dimensional visualization is omitted because the implicated hardware and software require-

ments for the user's computer system are still too high for world wide use and simple deployment of

the atlas product.

18 A codec is a program capable of performing encoding and decoding on a digital data stream or

signal. Codecs are used to reduce the size of audio and video les. The term is a contraction of

 coder and  decoder.


4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 42

the option to zoom-out, to make the view-port smaller the user can zoom-in by either a

xed zoom level or by drawing a new view-port rectangle onto the map. Whenever the

mouse cursor moves over the map, the GUI presents the geographical coordinates of the

current location.

4.3.2.2 Interactive legend Wiesmann (2007) describes possibilities of interactive

legends in online mapping. Besides their traditional functionality to explain the content

of the map, they can also represent more aspects like the order of the map layers. Re-

arranging the layers in the interactive legend results in automatic update of the map's

layer sequence. Clicking a layer results in contextual information about the layer.

4.3.2.3 Queries of map content The atlas provides an information query tool

(short Info-tool ) which allows to query the map elements. A click on a vector feature

returns attribute values of dierent types. Standard numerical or textual values can be

presented with a description of the attributes. The Info-tool can also present attribute

values of multimedia type. For example clicking a feature of the map can lead to the

pop-up of pictures, documents, sounds or videos.

The Info-tool is sensitive to dierent raster layer types: For raster layers with geo-

physical values measured on a ratio scale the Info-tool presents the raster's value. For

nominal values (e.g. classes of land-use) the raster value is automatically linked to the

legend's description for that class.

When a feature of the map is queried, the corresponding layer is highlighted in the legend.

4.3.2.4 Attribute based geo-object lter The attribute-based feature lter allows

the selection (or masking) of spatial objects on the basis of their attribute values. The

lter's user interface features two modes.

In the basic mode, interesting and context-related queries like  Which of these water

pumps show a dangerous level of contamination? can be oered to the user for a layer

about water infrastructure. These lters are context related and the AAT provides ways

to the atlas author to predene them. Predened lter queries allow to guide the users

to understand special aspects of a layer.

In a second expert mode GUI for the lter tool, users will be able to manipulate the

predened lters and also create their own queries. Basic arithmetics and all attributes

from a layer can be used to create lter expressions.


4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 43

4.3.3 Usability
The important role of usability - or ease of use factor (Robinson et al., 1995, p.591)

- has been realized early as it is a general challenge in human-machine interaction.

Cartwright et al. (2001, p.13) postulate to extend the  approach for the design of usable

geo-visualization interfaces to meet the needs of the general population. They expect

a movement towards an ubiquitous use of geo-visualization by the general public be-

cause they have never before had access to such a broad range of spatial information.

Cartwright et al. (2001, p.13) complain that most geo-visualization tools developed thus

far have been designed for expert users with narrow application needs. Peterson (2007a,

p.71) states that it is a major challenge of multimedia cartography to bring interactive

map use to a larger audience by focusing on usability.

Andrienko et al. (2003) describe tests of usability of interactive maps in the

CommonGIS project. They discovered that without previous instructions people are

quickly overloaded with unknown functionality and make only limited use of new tools.

They recommend a simple interface which oers selected tools only when they are

indicated.

The AVT design is oriented towards general public use and assumes that the target

group is not educated in GIS use. The following aspects of the AVT intent to provide

high eectiveness and satisfaction for the users.

4.3.3.1 Software internationalization and localization Software international-

ization is a concept to separate translations and program code in software design. An

internationalized software has a pluggable language interface (Conner, 2007).

Most parts of the AVT GUI are translated by the atlas author with the help of the

AAT (see 4.1.5) and are not part of software internationalization. The text messages

inherent to the AVT are provided in English. The applied concept of software interna-

tionalization allows to add more languages without changing the program code.

Software localization is the adaptation of a software to a locality (in addition to the

language). Localized software is aware of currency, measure units, formatting of numbers

(decimal points, positioning of separators, character used as separator), and date/time

format used and can adapt to it (Conner, 2007).

The AVT determines (automatically or by interacting with the user) which language

to use for the software and content. Language and localization parameters are changeable

on the y.
4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 44

4.3.3.2 (No) installation To enhance usability the atlas product must not discour-

age the user with the need for complicated installation procedures. Some computer

systems reserve the right of installing new software to a special user (the administrator

or root user). If starting the atlas involves such an installation it presents an unnecessary

obstacle to the casual user. The software distribution concept used for the AVT must

allow to be executed without the need for system-wide installation (installation without

 administrator rights).

4.3.3.3 Export The atlas product will provide a way to also use the maps it contains

without the AVT software. Users must be able to print maps they created interactively.

According to Cartwright et al. (2001, p.6) a printed map has the very compelling feature

 that it is truly mobile and assists the user in the eld where it is highly valued.

To reuse the atlas maps in presentations and web-pages the AVT additionally allows

to export the view-port to image les.

The third export method allows to save layers to standard geo-spatial formats. They

can then be imported into GIS, desktop publishing or statistical software on the map

user's computer. The availability of the export-function is customizable on a by-layer

basis via the AAT to protect the atlas creator's interest.

4.4 Openness
The DGRF targets reduction of atlas production costs and an extended distribution

range of the multimedia atlas. Both objectives can be supported by openness in terms

of interoperability and open source software.

4.4.1 Interoperability
4.4.1.1 Operating Systems To run applications on computers an intermediate layer

of software is needed: the Operating System (OS). The OSs (or platforms) for consumers

can be grouped into three families: Microsoft Windows, GNU/Linux, and Mac OS.

Traditionally a computer application is developed for a specic OS family only and does

not run on any other OS


19 (Tanenbaum, 2002).

The true shares of operating systems installed on the world's computers can only be

estimated. Peterson (2007b, p.39) uses the visitor statistics of http: // www. w3schools.
com as a reference for the operating systems used by Internet users. Table 3 shows the

19 This is also the case for most commercial GIS applications.


4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 45

2007 Win XP W2000 Win98 WinNT Win.NET GNU/Linux Mac

November 72.8% 5.1% 1.0% 6.3% 2.0% 3.3% 3.9%

October 73.9% 4.9% 0.9% 5.6% 1.9% 3.3% 3.9%

September 74.3% 5.4% 0.9% 4.5% 2.0% 3.4% 3.9%

August 74.4% 5.7% 0.9% 4.0% 2.0% 3.4% 3.9%

July 74.6% 6.0% 0.9% 3.6% 2.0% 3.4% 4.0%

Table 3: Operating systems visiting w3schools.com in 2007. (W3Schools, 2007)

statistics for the last months of 2007. These numbers must be interpreted with caution

as they only represent visitors of the specic web-page. Beyond doubt Microsoft Win-

dows XP is the most popular operating system and the whole Microsoft Windows family

counts for nearly 90% of all users.

Android
20 , Google's OS for mobile devices to be released in 2008 is based on

GNU/Linux and Dell has been shipping computers with pre-installed GNU/Linux

to business and private customers since 2007. The GNU/Linux platform is of special

interest in developing countries and growing economies because it is free of charge and

may be adapted freely. The One Laptop Per Child (OLPC) project selected GNU/Linux

as the operating system of their 100$ laptops (Collins, 2007, p.36).

OS interoperability is the capacity of an application to run on dierent operating

systems without modications. It can be achieved by developing applications in a plat-

form independent programming language. The DGRF tools are platform independent to

reach a wide user community.

4.4.1.2 Open standards Computer standards are formalized descriptions and exact

denitions of interfaces like protocols, le formats etc. Open standards are publicly

available and have various rights of use associated with them.

Even if the name might suggest, open standards are not exclusively bound to open

source (4.4.2). Adobe Systems, for example, has been disclosing its PDF specications

since 1997 which allows everybody to write applications that can read and write PDF

documents. The Adobe Acrobat Reader application though has never opened its sources.

20 The Open Handset Alliance (http://www.openhandsetalliance.com) - a group of more than 30

technology and mobile companies - is developing the Android OS to be released in 2008. Google is

the initiator of the alliance.


4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 46

This combination has led to the wide diusion of the PDF and today many applications

(by competing vendors and open source) support the format (Jobst, 2007, p.420).

In the eld of geo-informatics, the Open Geospatial Consortium (OGC)


21 is the main

driving force for standardization. It is an international voluntary non-prot organization

which develops standards for geospatial services. The members include GIS vendors,

governments, and academia. (Jobst, 2007, p.422). Kiehle et al. (2006) and Gietler (2003)

also see the OGC as the single proper authority for the interoperability and convergence

in geo-spatial applications.

Jobst (2007, p.425) concludes, that the use of open standards in cartographic mul-

timedia products positively aects the sustainability of the digital content. The com-

patibility of data, software, and hardware in respect to the multimedia product will last

longer. Huber & Schmid (2003, p.1405) report that an augmented use of open standards

facilitates the authoring of maps and multimedia.


22

The DGRF uses open standards where ever possible. The denition of the atlas

package format is well documented and publicly available to allow other third-party

software to export or import the package format.

One possible application for an open standard atlas package format is the extension

of spatial modeling platforms like XULU


23 : Extended with an export to atlas feature,

results of multiple scenarios could be exported automatically into an interactive atlas

which would include all relevant metadata.

4.4.2 Open Source


The idea behind open source is to provide other programmers the means to understand,

modify, and redistribute a program. Thereby the intellectual property is transferred to

the public domain (Grassmuck, 2004). The Open Source Initiative (OSI)
24 is a non-prot

organization dedicated to the promotion and protection of the concept of open source.

OSI maintains a list of software licenses which they regard as compatible with the ideas

of open source (Jobst, 2007, p. 423).

21 Webpage fo the Open Geospatial Consortium: http://www.ogc.org


22 Their experiences are based on the development of the AoS2.

23 Extendable Unied Land Use modelling platform (XULU)is a platform for running generic spatial

models written in Java (Thamm et al., 2007). A clone of the CLUE-S LUCC model has been

developed using XULU. The original CLUE-S model was developed at the University of Wageningen

(Schmitz, 2005). XULU Project webpage: http://www.xulu-project.org/


24 OSI Website: http://www.opensource.org/
4 THE DIGITAL GEOGRAPHIC REPRESENTATION FRAMEWORK 47

The open source policy of free redistribution, as required by the OSI open source

standard, is very important. It allows the DGRF design to incorporate and adapt ex-

ternal open source software. Depending on the particular open source license, use and

redistribution can implicate that the resulting application also has to be released under

an open source license (Laurent, 2004).

To foster easy distribution of the atlas package, the AVT is licensed as open source.

This guarantees that no license fees incur, because open source applications belong to

the public domain and thence can not be sold or owned (Grassmuck, 2004). This gives

the atlas creators the freedom to sell their product (but not the AVT) or to give it away

for free.

Peterson (2007b, pp.43) sees a new model for cartographic development arise in the

interplay of many dierent actors - from developers of basic geo-spatial libraries and

bug xers to researchers and non-prot organizations. All participants create small

contributions which can by reused and adapted to create powerful solutions. The DGRF

follows this new model by licensing the AVT as open source and by applying open

standards where ever possible.


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 48

Figure 11: AtlasCreator-AtlasViewer Framework

5 AtlasCreator-AtlasViewer Implementation

The framework designed is a conceptual approach to facilitate the production workow of

multimedia atlases. To prove the feasibility of the concept, the AtlasCreator-AtlasViewer

framework (AC-AV) has been implemented as a prototype. The AC-AV tries to provide

appropriate realizations for all DGRF requirements described in the previous chapter.

The AC-AV consists of two applications: The correspondent to the Atlas Visualiza-

tion Tool is the AtlasViewer (AV) application, the implementation of the Atlas Authoring

Tool is the AtlasCreator (AC) application. For the atlas package format a new Exten-

sible Markup Language (XML), called Atlas Markup Language (AtlasML), has been

developed. The structure of the AC-AV is shown in gure 11.

To test the AC-AV prototype, the IMPETUS project plans to create the successor of

the IIDA1 (see 3.2) with the AC-AV: the IMPETUS Interactive Digital Atlas (version 2.0)

(IIDA2). One implication of the IMPETUS test case has been that creators, consumers,

contents, and functionality are dened precisely as shown in table 4.

The IMPETUS project has continuously supported the development of the AC-AV

with realistic samples of geospatial and multimedia content. Despite these IMPETUS

related requirements, the AC-AV is programmed as a general tool. Content and logic

are strictly separated: The prototype can be used for dierent atlas topics, on dierent

scales, in dierent languages, and with adapted interactivity - without any changes to

the applications.
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 49

Use case Creators Consumers Functionality

IMPETUS IMPETUS scien- Stakeholders in Presentation of interdisci-

Interactive tists; Benin and Mo- plinary research ndings;

Digital Atlas creation rocco; Basic analysis functions;

Atlas team at the ZFL Laymen or novice Facilitate re-usability of data

(version 2.0) GIS users

Table 4: Creators, consumers, and required functionality for IMPETUS.

Chapter 4 discussed the design, the following sections describe its implementation.

Java as a basic requirement The AC-AV has been programmed in the Java pro-

gramming language. Java has been developed by Sun Microsystems and rst released

in 1995. Java applications are typically compiled to a bytecode, which can run on any

Java Virtual Machine (JVM) regardless of underlying computer architecture (Ullenboom,

2007). Thus the AC-AV can run on all platforms where a recent JVM is installed.

Java has been selected for three reasons:

1. Following the Sun Microsystems advertisement slogan Write once, run anywhere,

Java is a cross-platform system which is not dependent on the underlying operating

system (Ullenboom, 2007).

2. Java is not just a programming language. It also contains a plethora of ready-to-

use tools and libraries, especially for XML and Internet applications (Ullenboom,

2007).

3. Java is the author's programming language of choice.

Both AtlasCreator and AtlasViewer need Java version 6 or later installed. This is the

only requirement to use the AC-AV framework. Java 6 may be downloaded for free under

the terms of the Sun Microsystems, Inc. Binary Code License Agreement
25 from the Sun

Microsystems Java website


26 . Paragraph B of the license grants you a non-exclusive,

non-transferable, limited license without fees to reproduce and distribute the Software,

provided that (i) you distribute the Software complete and unmodied and only bundled

25 Sun Microsystems, Inc. Binary Code License Agreement online: http://java.sun.com/javase/6/


jre-6u4-license.txt
26 Sun Microsystems Java 6 webpage: http://java.sun.com/javase/6/
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 50

as part of, and for the sole purpose of running, your Programs . Hence it is possible to

distribute the Java 6 software with the atlas on the same CD. Oine atlas users which

do not yet have Java 6 installed can install it directly from the same medium that the

atlas is distributed on.

5.1 The AtlasViewer application


Two approaches exist for the development of software for multimedia atlas products:

Multimedia in GIS and GIS in Multimedia (Bär & Sieber, 1999). Both digital versions of

the Atlas of Switzerland have been developed following the GIS in Multimedia approach.

Schneider (1999, p.2) describes the approach as more costly but also more exible

in the design of the GUI. The counterpart approach - Multimedia in GIS - bases the

development of the atlas software on existing GIS software


27 . The GIS functionality is

then extended to integrate non-spatial thematic contents.

The AV development followed the second approach and based its development on an

existing GIS toolkit: the GeoTools GIS library


28 . GeoTools is an open source Java GIS

library which provides functions and methods to handle geo-spatial data. The library

implements OGC specications as they are developed. GeoTools itself does not provide

any GUI and is not intended as an end-user application. The library is distributed under

the terms of the Library GNU Public License (LGPL)


29 (Macgill, 2005).

The AV software has been created as a generic Atlas Visualization Tool. An overview

of the nal GUI can be seen in gure 12.

5.1.1 Visualization
The visualization components developed for the AV are separated into the standard geo-

visualization component (5.1.1.1) and an extension to handle eciently big raster images

(5.1.1.2).

5.1.1.1 Geo-visualization component The geo-visualization component of the AV

has been developed on the basis of the XULU plug-in  Geo-Visualisierung described in

Schmitz (2005), which also is based on the GeoTools Java library. Further enhancements

27 Schneider (1999, p.2) binds this approach to commercial GIS software, but this denition is probably

based on the ignorance of powerful non-commercial GIS libraries and tools.

28 GeoTools project webpage: http://geotools.codehaus.org/


29 Lesser General Public License online: http://www.gnu.org/licenses/lgpl.txt
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 51

Figure 12: AtlasViewer - Example for map visualization.

of the geo-visualization component during the development of the AV have been ported

back to the XULU project.

The visualization component can render a stack of raster and vector layers. Basic

map elements like scale bar and grid exist as shown in gure 13. If the CRSs of the layers

are inhomogeneous, all layers are automatically transformed to the CRS of the bottom

layer.

The component allows labeling of polygon, line, and point objects. Positioning and

selective generalization of labels are handled by the underlying GeoTools library.

The geo-visualization component fully renders the map whenever the view-port

changes. This implies that the GeoTools algorithm reallocates all labels - also when the

view-port has just been moved by a single pixel.

5.1.1.2 Rendering of big raster image les An important requirement for the

AV is the possibility to handle big raster les eciently. This is especially motivated by

the requirements for the IIDA2 to publish remote-sensing products like high resolution

satellite images and large area land-use classications


30 .

30 For example, the Zentrum für Fernerkundung der Landoberäche (en.: Center for Remote Sensing of

Land Surfaces) (ZFL) created a LANDSAT satellite image of Benin with a resolution of 6692x13638
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 52

Figure 13: AtlasViewer - Geo-visualization component.

The default algorithm to display an image - as used by the XULU geo-visualization

component - is to rst fully load the image into the computer's memory and then re-

sample it to the resolution of the AV's view-port. For large rasters this procedure is very

memory- and time-consuming. This has led to the conclusion that a special procedure

for big images had to be developed.

The oversize in memory and the computational eort of down-sampling are the two

bottlenecks in real-time visualization of big raster images with the standard procedure.

The problems can be avoided by using image tiling in combination with an image pyramid

and image compression algorithms.

Image tiling The primary problem with big raster images - they can not be fully

loaded into memory - is tackled by the concept of tiling. Tiling means to separate the

image into many smaller images (Shirley, 2005, p.274). A single image le of 1000px²

can thus be separated into hundred les of 100px² each. By linking the tiles to a spatial

index only the les which intersect with the view-port must be uncompressed and loaded

into memory. Thereby it is possible to avoid the need to load the whole raster coverage

into the computer's main memory.

Image pyramid The second problem with large raster images is their versatility:

Due to the high resolution and extend they can be analyzed on a wide range of scales.

To show a small scale map of a high resolution raster, it has to be down-sampled which

pixels and three bands of 8bit resolution. This is equivalent to 279Mb of geodata and can not be

visualized with the standard algorithm.


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 53

can be very computationally intensive. Image pyramids are a solution to avoid excessive

re-sampling delays during real-time visualization.

The image pyramid follows a Time-Memory Trade O


31 by transferring the re-

sampling eort from the real-time visualization process (in the AV), to a one-time work

(in the AC) resulting in an increased storage size.

Instead of down-sampling the original raster image on demand (e.g. whenever the

user is changing the scale) the raster is being down-sampled in the creation process once

and then stored and distributed in multiple resolutions (Ogden et al., 1985). When

the AV software needs to visualize the raster in a special resolution, the nearest higher

resolution is chosen and down-sampling is performed on that data-set.

The number of resolution levels of an image pyramid has to be carefully chosen. The

more resolutions are pre-calculated, the less down-sampling time is needed to present the

raster in a requested resolution. At the same time the amount of storage space increases.

Every image pyramid is grounded on the original (highest resolution) raster extension.

Hence the base area of the pyramid equals the full extent and resolution of the original

raster image. At the peak of the image pyramid, there is only one pixel that visualizes

the most aggregated view at the smallest displayable scale.

Tiled Image Pyramid (TIP) layers The concepts of image tiling and image

pyramid have been combined to Tiled Image Pyramid (TIP) layers which can be vi-

sualized by the AV. A TIP layer is an image pyramid where every resolution level is

tiled.

Figure 14 shows the concept of a TIP with three resolution levels: 100%, 50% and

10%. If a resolution of 40% is requested for geo-visualization, the down-sampling will be

carried out on the 50% resolution raster graphic and thereby reduces the amount of data

and computational delay involved signicantly.

The geo-visualization component (5.1.1.1) has been extended to visualize TIP layers.

This enables the AV to geo-visualize rasters of virtually unlimited size


32 in reasonable

31 Time-Memory Trade O is a general concept to trade the calculation time of an algorithm against

it's memory (or disk storage) consumption - and vice versa. The toehold to this concept is to pre-

calculate as much as possible and then distribute the new algorithm together with a larger library

of prepared and pre-calculated results (Stamp, 2003).

32  Virtually unlimited size means, that the geo-visualization eort of large raster images does not

correlate with their size. Only the storage size of the atlas package limits the raster's size.
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 54

Figure 14: Tiled Image Pyramid (TIP) for ecient visualization of big rasters

time and with very low memory requirements


33 . Figure 17 shows the AV visualizing a

TIP layer of approximately sixty mega-pixel (7500px²).

5.1.2 Interactivity
All dialogs of the GUI are interactive. However dialogs are a constrained form of in-

teractivity, because specic rules exist about which interaction with a dialog is valid

an accepted by the software. The following subsections describes the interactive GUI

components which have been developed for the AV.

5.1.2.1 Menu structure The only native entries in the menu bar of the AV are

File and Help. All other menus are dened by the atlas author via the AC tool. Menu

items can contain links to data entries of the atlas, that is to maps, layers, videos, and

documents. For every type of menu item a descriptive icon is shown. When a menu item

is clicked, the AV acts as follows:

Selecting a map item from the menu replaces any active map in the geo-visualization

component and shows it.

33 TIP layers have been successfully tested using a JVM with only 150Mb memory assigned.
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 55

Figure 15: AtlasViewer - Interactive legend

Selecting a layer item from the menu will add the selected layer to the loaded map.

If the layer already exists in the map, the action will be rejected with a explanatory

message.

Selecting video, sound, or documents opens the media in a new window without

changing the map.

5.1.2.2 Interactive legend The AV has been equipped with an interactive legend

which automatically updates in context of the map shown. The concept of  Smart

Legends (Wiesmann, 2007) has been applied. The legend is the main interface to ma-

nipulate the visualization of map and layers. Figure 15 shows the legend and enumerates

its interactive functionality.

1. Layer information

For every layer the author can attach multi-lingual descriptions in form of HTML

pages. If a description is available for the active language the legend will automati-

cally show a blue i icon. If the user clicks that icon, a pop-up window with layer

specic meta information appears. This pop-up window at the same time acts like
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 56

a basic web browser


34 . The HTML pages may contain images, tables, and links to

other internal (part of the atlas) or external (on the Internet) links.

2. Layer tool menu

The tool icon opens a layer specic menu. All actions on the tool menu relate only

to the corresponding layer. The attribute geo-object lter (5.1.2.5) and the export

function (5.1.2.6) can be accessed from there. The menu item transparent allows

to change the transparency of the layer. Clicking the remove item removes the

layer from the legend (and thereby from the active map view).

3. Hide layer in legend

The rightmost icon can be used to toggle visibility of a specic legend entry. The

hidden layer is still visible in the map but takes up less space in the legend. It can

be used to set up a clearer legend.

4. Visibility indicator

The eye icon is used to indicate the layer's state of visibility in the map. When the

eye is closed the layer is not shown in the map. The eye is half closed when the

layer has been set to 50% transparency. By clicking on the icon the layer can be

hidden and un-hidden.

The order of layers can be changed by using drag-and-drop. Therefor the user clicks on a

layer in the legend, drags it to the desired position, and releases the mouse button. The

map view will be automatically updated to the new order of layers.

Double clicking on a layer in the legend makes the geo-visualization component zoom

to the layer's extends.

Removing layers from the legend does not change the atlas content. The original map

can always be reloaded by selecting it from the menu.

To add layers to a map, they have to be selected from the menu. In combination

with the menu, the interactive legend provides all functionality needed by atlas users to

compile customized maps


35 .

34 A web browser is a software which can render HTML documents and allows interaction with them

by clicking links, which lead to other HTML documents.

35 It is a good idea to deliver an extra  empty map with the atlas which just has a few layers included

to provide basic orientation. The user can select that  empty map from the menu and start to add

layers as he/she wishes.


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 57

Figure 16: AtlasViewer - Map tools.

Figure 17: AtlasViewer - Info-tool querying a land-use raster

5.1.2.3 Map tools In the top left corner four radio buttons
36 hover above the map

(see gure 17). They are used to switch the functionality of the mouse. The blue

shimmer indicates which tool is active at the moment. Tools are (from left to right) Info-

tool (5.1.2.4), panning, zoom-in and zoom-out. The other three buttons allow spatial

navigation with the mouse as described in 4.3.2.1.

When a map tool is selected the mouse cursor changes accordingly. A magnifying

glass with a minus and a plus symbol has been chosen for the zoom tools. The panning

tool is indicated with an open hand that changes to a closed st while the user  grabs

the map and moves it.

5.1.2.4 Info-tool The Info-tool allows to query geo-objects by clicking them in the

map. The algorithm always reports on the uppermost geo object which is hit under the

36 Radio or option buttons allow to choose exactly one option from a predened set of options. When

one button is selected, the other buttons are automatically deselected.


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 58

Figure 18: AtlasViewer - Info-tool querying vector objects

position of the mouse. It searches in a radius of ten screen pixels to make it easier to hit

point and line objects.

The Info-tool always shows the cursor's position in map coordinates in the Coordinates
section of the pop-up window. All value elds in the Info-tool are accessible to standard

copy and paste actions. Other parts of the Info-tool window are more dynamic:

Figure 17 shows the Info-tool in use with a raster layer (land-use classes). For raster

layers the Info-tool presents the raster value (2) and its translation according to the

legend ( gallery forest).

Figure 18 shows two pop-up windows for vector layers. The Attributes block shows
properties of the selected geo-object. Which attributes are shown in the Info-tool can be

dened in the AC (see 5.2.2.4). Besides text and images, attributes values may also be

visualized as buttons that open PDFs or as an embedded player to playback a video or

sound clips.

5.1.2.5 Geo-object lter The interactive legend (5.1.2.2) allows the user to open

a lter dialog for any layer as seen in gure 19. The lter tool was developed as a part

of the XULU project and has later been embedded into the AV. It evaluates a lter
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 59

Figure 19: AtlasViewer Geo-object lter tool.

expression against all geo-objects of a given layer. The expression may contain attribute

names (as $ATTNAME variables) and basic algebra (selectable via the pull-down menu).

The schema
37 of the layer's database is shown in the top part. This gives an overview

of the variables and data types available in the layer. Variable names can be directly

copied into the lter formula by a double click.

The following example of a lter expression will only show settlements which grew

by more than 3% from 2003 to 2004:

1 ( $ POP2003−$POP2004 ) > ( $ POP2003 / 1 0 0 * 3 )

The test filter button allows to test the lter expression and preview the selected
object in a preview window.

The lter tool has not been fully implemented as described in 4.3.2.4, in the sense

that the AC-AV does not yet allow the atlas author to predene lter expressions and

encapsulate them in human-readable questions. As a prototype however, the tool fullls

the lter functionality.

37 The schema of a database denes the structure, e.g. the columns and data types which can be saved

in the columns. The schema does not contain the underlying data (Kemper & Eickler, 2004, p.22).
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 60

5.1.2.6 Export Besides an extended screenshot functionality (File -> Make smart
screenshot), the interactive legend allows to export layers via the tools menu. The user

will be asked for a target directory. All layers will be exported the way they have been

imported - including CRS, Styled Layer Descriptor (SLD), or other auxiliary les. The

export function is only available if it has been enabled for the layer in the AC.

5.1.2.7 PDF document visualization The AV requires a third-party PDF viewer

to be installed on the user's computer. When requested to open a document, the AV rst

extracts it from the atlas package to a temporary directory on the computer. Thereby

the PDF le is given a meaningful name which is derived from the internationalized title

of the document. This name will likely be presented in the title bar of the PDF viewer.

After the PDF is extracted the AV follows dierent strategies for Microsoft Windows,

GNU/Linux and Mac OS:

On Microsoft Windows the PDF is opened by calling the application that is linked

with the .pdf le extension.

1 cmd . e x e /c start c : \ temp \ W a s s e r h a u s h a l t . p d f

On Mac OS the NeXTSTEP open command is used to open the PDF with its associated
viewer. This is recommended for maximum compatibility by Marinacci & Adamson

(2005, ch.11).

1 open / tmp / W a s s e r h a u s h a l t . p d f

On GNU/Linux based distributions there is no standardized system to associate le

types with applications. Hence the AV has to determine which PDF viewer is installed.

The following sequence of PDF viewers is tested by the AV: Evince


38 (Gnome default),

KPdf
39 (KDE default), Adobe Acrobat for Linux40 , ePDFView41 , Xpdf42 . If any of the

PDF viewers is found it will be used to show all subsequent documents.

Whenever the AV is started or closed it cleans all related les from the temporary

directory. This prevents the temporary copies of documents and videos to occupy too

much disk space.

38 Evince project webpage at gnome.org: http://www.gnome.org/projects/evince/


39 KPdf project webpage at KDE.org: http://kpdf.kde.org/
40 Adobe Systems Acrobat Reader for Linux: http://www.adobe.com/support/downloads/product.
jsp?product=10&platform=unix
41 ePDFView project website http://trac.emma-soft.com/epdfview/
42 xpdf project website http://www.foolabs.com/xpdf/
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 61

5.1.2.8 Video and sound playback The Sun Microsystems Java Media Framework

(JMF)
43 has been embedded into the AV and provides a Java based video and sound

player component. It includes a Cinepak codec which needs not be previously installed

on the OS and thus ensures playback without any codec requirements. JMF version

2.1.1e has been embedded in the AV.

The AC ensures that all video imported into the atlas is compatible with playback

by the JMF component (5.2.1.4).

5.1.3 Subsumption
The AtlasViewer is a Java based implementation of the Atlas Visualization Tool designed

in section 4.3. It is a user friendly, GIS-like explorer for geoproducts embedded in mul-

timedia. A Multimedia in GIS approach has been applied and the application has been

developed on the basis of the open source GeoTools library. The program is able to dis-

play videos, images, documents, and sounds in addition to geospatial portrayal. Ecient

navigation through large raster imagery has been make possible on standard computers.

A variety of tools allows the user to explore the geo-spatial and multimedia contents.

The AV is generic: upon start up it analyses the delivered atlas package (see 5.3) and

adapts itself to it.

5.2 The AtlasCreator application


The AC is a Java based implementation of the Atlas Authoring Tool designed in section

4.1. Its GUI provides the central interface for the atlas authors. It helps them to import

dierent content elements (5.2.1) and provides ways to collocate them to a multi-lingual

atlas product (5.2.2). Multiple aspects of the AV GUI - e.g. legend descriptions, menu

items, colors, and messages - can be adapted to the purpose of the multimedia atlas.

5.2.1 Import
Before data can be imported into the AC an atlas project has to be loaded or created.

A new project is created via the File -> New menu item. After choosing a directory,

the AC asks for a list of languages in which the new atlas shall be created. Subsequent

questions for title, description, creator or vendor, and copyright notice must be answered

multiple times for all stated languages.

43 Java Media Framework webpage: http://java.sun.com/products/java-media/jmf/


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 62

Figure 20: AtlasCreator - Authoring the IIDA2.

The general import routine for the AC has been implemented via drag-and-drop. The

author clicks on the le to import and drags it over the data pool list (see gure 20).

When the mouse button is released it will automatically determine the le type and start

the import.

5.2.1.1 Geo-spatial import Import of geo-spatial data sets has been implemented

for basic raster and vector le types (4.1.1).

It has not been possible to implement direct import from OGC SDIs within the time

frame of this thesis. Also automatic generation of TIP layers for large raster les has not

yet been implemented in time. For the IIDA2 the large raster layers have been manually

converted to TIP layers. The manual creation of TIP layers is described in appendix A.3.

After creation, TIP layers can be imported automatically like other geo-spatial data les.

5.2.1.2 Handling of Coordinate Reference Systems As described in section

2.2.1, transformation of coordinates between two CRSs requires the exact denition of the

CRSs and all their parameters. The AC supports two ways to formalize this information:

EPSG and WKT.

EPSG codes The OGP Surveying & Positioning Committee (formerly European

Petroleum Survey Group (EPSG)) aims at standardization of common projection de-


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 63

nitions. They maintain a freely available database of so-called EPSG-codes. Using that

database, a single EPSG-code can represent a fully dened CRS (Neteler & Mitasova,

2008, p.38). The AC includes version 6.14 (September 2007) of the EPSG-database and

can interpret EPSG codes as full CRS denitions .

When importing a geo-spatial data le into the AC its CRS can be dened by using

a single EPSG-code. The code has to be written into a .prj projection le as shown:

1 EPSG:32543

The .prj le has to have the same base le name as the corresponding data le.
44

If this naming convention is heeded, the AC automatically recognizes the CRS and can

transform the coordinates for visualization in a dierent CRS.

Well-known text Well-known Text (WKT) is a text markup language for repre-

senting vector geometries and CRSs. The format is regulated by the OGC and described

in the Coordinate Transformation Service specications (OGC, 2001).

As an alternative to the use of EPSG-codes, the AC can additionally read the WKT

standard. This allows for a more exible denition of CRSs which are not part of the

EPSG-database. This was the case for the CRS used by the IMPETUS project for their

Morocco test site. By creating a WKT denition it has been possible to import and

transform local coordinates into the AC. The following example shows the denition of

the Lambert IMPETUS Morocco CRS:

1 PROJCS [ " Lambert_IMPETUS_Marokko " ,

2 GEOGCS [ " GCS_Merchich_Degree " ,

3 DATUM[ " D_Merchich " ,

4 SPHEROID [ " C l a r k e _ 1 8 8 0 _ I G N " , 6 3 7 8 2 4 9 . 2 , 2 9 3 . 4 6 6 0 2 ] ,

5 TOWGS84 [ 3 1 , 146 , 47 , 5, 3, 3, 9]] ,

6 PRIMEM [ " G r e e n w i c h " , 0 . 0 ] ,

7 UNIT [ " D e g r e e " , 0 . 0 1 7 4 5 3 2 9 2 5 1 9 9 4 3 3 ] ] ,

8 PROJECTION [ " L a m b e r t _ C o n f o r m a l _ C o n i c " ] ,

9 PARAMETER [ " F a l s e _ E a s t i n g " , 5 0 0 0 0 0 . 0 ] ,

10 PARAMETER [ " F a l s e _ N o r t h i n g " , 3 0 0 0 0 0 . 0 ] ,

11 PARAMETER [ " C e n t r a l _ M e r I I D A n " , −5.4] ,


12 PARAMETER [ " S t a n d a r d _ P a r a l l e l _ 1 " , 3 1 . 7 2 3 9 2 5 6 4 7 2 ] ,

13 PARAMETER [ " S t a n d a r d _ P a r a l l e l _ 2 " , 3 4 . 8 6 6 4 5 7 6 5 5 8 ] ,

14 PARAMETER [ " L a t i t u d e _ O f _ O r i g i n " , 3 3 . 3 ] ,

15 UNIT [ " M e t e r " , 1 . 0 ]

16 ]

44 For example, if the data le is called landsat122.gif, the .prj le must be called landsat122.prj.
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 64

5.2.1.3 Cartographic representation rules import As described in section 2.4.2,

geo-spatial data only describes what to display - not how to display it. Additional to the

import of geo-spatial data, the AC also has to import cartographic representation rules

to determine the appearance of the maps.

OGC Styled Layer Descriptor To formalize cartographic representation rules

the Open Geospatial Consortium (OGC) developed an XML based computer language:

the Styled Layer Descriptor (SLD) format.


45

SLD addresses the need for controlling the graphical representation of spatial infor-

mation decoupled from a specic software provider. SLD allows to formalize complex

cartographic representation rules (OGC, 2002).

All web service products related to the OGC Spatial Data Infrastructure standard-

izations support SLD for the description of cartographic representation rules. Due to

its openness SLD is especially well supported in Free and Open-Source Software for

GIS (FOSSGIS) (Weiser, 2005, p.17).

A SLD document is a collection of rules which can be applied to a set of geo-spatial

features. As an XML based language SLD provides some degree of human-readability.

Its text-based rules can be viewed and edited with standard text editors. Rules contain

lters and symbolizers. Filters control which objects shall be styled and symbolizers

determine how these objects will be displayed (Macgill, 2005).

Filters can be linked to numerical and logical operations on attribute values. The

following SLD lter example will aect only those geo-objects of which the attribute

PERSONS lies between 2 million and 4 million:

1 < o g c : F i l t e r>

2 < o g c : P r o p e r t y I s B e t w e e n>

3 <o g c : P r o p e r t y N a m e>PERSONS</ o g c : P r o p e r t y N a m e>

4 <o g c : L o w e r B o u n d a r y>

5 < o g c : L i t e r a l >2 0 0 0 0 0 0</ o g c : L i t e r a l >

6 </ o g c : L o w e r B o u n d a r y>

7 <o g c : U p p e r B o u n d a r y>

8 < o g c : L i t e r a l >4 0 0 0 0 0 0</ o g c : L i t e r a l >

9 </ o g c : U p p e r B o u n d a r y>

10 </ o g c : P r o p e r t y I s B e t w e e n>

11 </ o g c : F i l t e r >

45 SLD XML was standardized in 2002 and is specied in the SLD Implementation Specication 02-070
(OGC, 2002).
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 65

The order of the rules is important as it denes the order of rendering. Rules that

apply rst will be rendered rst and following rules will be rendered on top, e.g. they

can  over-paint the symbols of the earlier rules (Weiser, 2005, pp.27).

An important aspect for interactive maps - where the scale can be dynamically

changed by the map user - is the possibility to dene scale dependent rules which allow

automating selective generalization of maps (OGC, 2002).

According to Dietze & Zipf (2007, p.2), SLD has only limited capabilities for thematic

mapping. Only simple choropleth maps and point maps can be described. Anyhow, for

the AC-AV the OGC standard has been preferred to the development of a self-made (non

standardized) solution due to its compatability with a growing pool of GIS applications.

Styled Layer Descriptor document import Import for SLD documents (as

described in 4.1.2) has been developed for the AC on the basis of GeoTools. The SLD

document for a layer is expected to have the same le base name, but with a .sld
extension. When importing a geo-spatial data le a corresponding SLD document is

automatically imported. If no .sld le is found the AC provides default SLDs for

points, lines, and polygons.

For the production of the IIDA2, the SLD documents have been created with the

help of the open source tool ArcMap2SLD


46 described in Weiser (2005) and Weiser &

Zipf (2005). The tool allows ESRI ArcGIS color and symbol denitions to be converted

to SLD documents. Thereby the ESRI ArcMap


47 tool is used as a visual editor for the

cartographic representation rules.

Unfortunately, ArcMap2SLD cannot convert the full range of ArcMap denitions. In

some cases the output did not fully comply with the OGC standard
48 . Most of the SLD

documents had to be manually revised before they could be imported into the AC.

5.2.1.4 Video import Delivering videos as part of the multimedia atlas has turned

out to be more dicult than expected due to codec restrictions. Codecs are algorithms

to encode and decode videos. When a video is played, the player software usually asks

the OS to decode the video with the appropriate codec. Codecs are not part of video

46 Arcmap2SLD project webpage: http://wald.intevation.org/projects/arcmap2sld/


47 ArcMap is the tool of ArcGIS, used to digitize, edit and symbolize data to create maps for the digital

or analogue export.

48 Bug reports have been submitted to the ArcMap2SLD project website.


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 66

players, but installed on the OS. Many dierent codecs exist: some are open source,

others are copyrighted by OS vendors like Microsoft or Apple.

To create a platform independent multimedia atlas product, the videos have to be

encoded with a codec that is available on all targeted OSs. Unfortunately the list of

default codecs installed on Microsoft Windows, Mac OS and GNU/Linux does not inter-

sect. Hence the atlas could not deliver videos which will run on all three systems without

requiring further codec installations


49 .

A solution was found by using the Sun Microsystems Java Media Framework (JMF).

Contrary to the general procedure (which asks the OS to decode the video), the JMF

comes with an internal, Java-based player for the Cinepak video codec. Hence, using only

Cinepak coded videos and JMF as the AV's player component, no further requirements

are needed.

Cinepak was designed in 1992 to playback 320x240 resolution video at 1x single speed

CD-ROM transfer rates (150 KByte/s). In terms of compression ratio the codec is rather

antiquated
50 (Ferguson, 2001), but its compatibility makes it the default format for the

AC-AV.

When importing a video the AC determines whether it is compatible with the JMF.

Figure 21 on the following page shows the JMF video player component and how it is

used by the AC to test all imported videos.

Automatic conversion of incompatible video clips during import has not been imple-

mented in this prototype of the AC.

The mencoder project


51 provides an open source video converter. All videos for the

IIDA2 have been manually converted using the following mencoder command (SOURCE

and TARGET have to be replaced with the input and output video le name):

1 mencoder −o v c vfw −x v f w o p t s c o d e c= i c c v i d . d l l −o a c pcm SOURCE −o TARGET

After conversion, the videos can be imported into the AC by drag-and-drop.

To embed small videos into HTML descriptions, the GIF89a image format is addi-

tionally supported. As stated by Compuserve (1990), the format is not intended for

49 On all OSs, extra codecs have to be installed with  administrator rights. Thus it was not an option

to ship an open source codec with the AV. As dened in 4.3.3.2, the user shall not be hassled to use

 administrator rights and extra installation routines to use the atlas product.

50 A disadvantage of the Cinepak format is its size: A 640 x 480 video of 20 seconds uses 32Mb. It can

hence only be used for short, low resolution scenes. On the other hand, Cinepak codec was designed

for 386 computers and requires very little CPU power (Ferguson, 2001).

51 Mplayer and Mencoder project webpage http://www.mplayerhq.hu/design7/news.html


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 67

Figure 21: AtlasCreator - Testing compatibility during video import

animation even though it can be realized in a limited way. GIF animations should not

exceed few seconds of duration.

5.2.2 Content Management


When the import of the building blocks for the multimedia atlas is complete, the AC

provides the author with tools and concepts to structure the content and to compile new

elements like maps and menus.

5.2.2.1 Content model As required by the DGRF (see 4.1.4) the AC must provide

ways for the contents of the atlas to be structured generically. The content model design

for the AC contains three classes of objects: Data pool entries, maps, and groups. The

possible references between them are shown in gure 22.

Data pool entries The data pool manages geo-spatial data sets and multimedia

supplements that are imported into the AC. There is one data pool per atlas. Simplied,
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 68

Figure 22: Model for structuring atlas content

Figure 23: Topology of data pool entries

the data pool can be seen as a folder on the hard drive where all imported data are copied

to.

Data pool entry is the top class in a typology of data pool entry types shown in

gure 23. Every class has properties which are handed down to its successors.
52 The

data pool entries represent the building blocks of the atlas content.

52 For example, every data pool entry has internationalized elds for name and description. The layer

class additionally has a legend attribute. A vector layer is a special kind of layer that inherits the

properties name, description and legend from the parent classes.


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 69

Maps The geo-spatial data sets in the data pool only represent single layers. Van

Elzakker (2004, p.138) argues, that a novice user must be able to choose from a set of

ready-made maps.
53 Otherwise he/she might create  bad maps and misinterpret the

situation, because it requires cartographic knowledge to assemble layers to meaningful

maps. Thus the AC allows the creator to prepare maps from layers in the data pool.

These maps are stacked in the map pool. There is one map pool per atlas.

Maps not only contain references to layers. A map can also be related to media

entries in the data pool (e.g. a pedological PDF document to accompany a soil map).

Hence a map can also have references to media entries (see gure 22).

Accompanying HTML descriptions for a map are not stored in the data pool but as

direct attributes of the map itself.


54

Groups The AC provides a way to structure its content thematically (e.g. see

AoS2 theme groups: table 1 on page 24). Manifold views on the content are imaginable:

Structured by creator (organization -> author), geographically (country -> region ->

location), by media type (maps, videos, images).

In the AC content model, a group represents a basic unit to create arbitrary hierar-

chies. A group can hold multiple references to elements in the map and data pool. A

group can also reference to other groups which can then be seen as subgroups. This allows

to create arbitrary tree structures to represent a great variety of thematic topologies.

Every atlas has one root group. All other groups are subgroups.

This prototype simplies


55 the group concept and uses the root group directly to

describe the menu structure of the atlas. Hence sub-groups of the root group represent

entries in the menu bar.

5.2.2.2 Creating Maps When multimedia les, geo-spatial data, SLDs, and CRS

information have been imported the author can start preparing multimedia maps for

atlas product.

53 Nevertheless, the AV additionally allows brave users to modify and create their own maps.

54 Specic HTML descriptions for a map are stored as direct attributes of the map because they are

always bounded to exactly one map (1:1 relationship). There exists no reason to manage a reference

to the data pool. All other data pool entries can be referenced by many maps (1:n relationship).

55 This limitation of the prototype is a hindrance due to the lack of development time. The group

concept allows to structure the content in many more generic ways: It can dene one branch for the

menu, another to structure of the index, etc.


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 70

Figure 24: AtlasCreator - Map composer

The GUI of the AC is divided into three areas. The data pool area on the left shows

a list of all data pool entries. The right is separated into the map pool and a menu

structure tree. To create references as shown in gure 22, the mouse can be used to

drag-and-drop an object from one window to another. Figure 20 shows the AC GUI with

some IIDA2 samples.

Items in the map pool list may be managed with the Add, Delete, and Preferences
buttons. The third allows to translate the map's meta information (see gure 25). The

green box surrounding one of the maps identies the map which the atlas product will

start with when delivered.

When a map is double clicked the Map Composer window opens for that map as shown
in gure 24. The map composer window is an editable preview of the map. To add a

new layer it has to be dragged from the data pool and dropped into the legend of the

map composer with the mouse.

The button Save map area allows to dene the extend of the map which is shown

when it is opened by a user in the AV. When a folder containing HTML les is dropped

at the map composer it imports it as a map description page.

5.2.2.3 Creating Menus The menu tree on the right reects the menu bar that will

appear in the atlas product. Menus and sub menus can be created and deleted with
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 71

Figure 25: AtlasCreator - Translation of data pool entries and maps

Figure 26: AtlasCreator - Translation of database attributes

the Create new group and delete buttons. Edit allows to translate the title and the

description of the menu entry.

To add data pool entries or maps to the menu drag-and-drop can again be used. This

permits to reference any content element in the atlas via the menu. The usage of the

menu in the AV has been described in section 5.1.2.1.

5.2.2.4 Metadata internationalization The AC assists the author with translat-

ing metadata for all entities of the content model: data pool entries, maps, and groups.

Every item in the data pool list, the map pool list, and in the menu structure tree can

be translated. A GUI similar to gure 25 will appear. The prompted languages de-

pend on the languages dened for the atlas project. Red input elds warn the author of

untranslated labels.

Selecting a vector layer from the data pool activates the Attribs button. It allows to

adapt the appearance of the Info-tool (5.1.2.4). The attribute translation GUI (gure 26)
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 72

Figure 27: AtlasCreator - Translation of raster legends

generates a table with the database schema of the vector layer. The rst column shows

the original names of the columns in the database. The second column lists the data

types of the columns. The third column visible allows to select the attributes, that

are to appear in the Info-tool. The unit column allows to dene a text to be appended

to the attribute's values. Following pairs of name and description represent dierent

translations for that attribute's name. The description eld is used as a tool-tip
56 for

the attribute in the Info-tool.

To translate the entries of raster based legends, the Legend button in the data pool

window must be used. This tool allows to add and delete value-translation pairs. For

every pair an entry in the legend scale is created. The button gaps on/off switches

between the correct legend type for metric and interval scales. By not translating every

pair and using gaps off mode, the GUI can also be used to create non textual scales as

seen in gure 12 on page 51.

5.2.2.5 Publishing the atlas Selecting File->Export makes the AC export the

atlas as a transferable atlas package. No user interaction is required as the packaging in

fully automatized. The atlas package (described in section 5.3) is a folder that can be

burned on a medium or copied onto web-server.

56 In GUI design, a tool-tip is a short text which appears when the mouse lingers over a component for

a while. The tool-tip vanishes as soon as the mouse is being moved (Marinacci & Adamson, 2005,

p.225).
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 73

5.2.3 Subsumption
The AtlasCreator is a Java based implementation of an Atlas Authoring Tool as designed

in section 4.1. It allows to import data and metadata as building blocks of the atlas.

The content can be structured, maps can be compiled, and it assists the author in the

translation of the metadata. It is the central tool to manage the content of the atlas and

adapt its appearance.

5.3 Atlas Package


The atlas package is the third module of the AtlasCreator-AtlasViewer framework. The

package must store all data, settings, parameters, and translations - in short: the contents

- that have been prepared with the AC. The following parts explain how the concepts of

section 4.2 have been put into practice for the atlas package format.

5.3.1 Integrity and Compression


The Java Archive (JAR) is the Java standard le format for bundling programs. JAR

les can be created with an option to compress the content in ZIP format
57 . Additionally,

Java provides tools to sign and validate JARs to guarantee that the archive has not been

changed (Ullenboom, 2007).

When the author decides to export the atlas, the AC builds compressed JARs for

every data pool entry and signs them with a digital security certicate. Before the

AV starts, the signatures of all archives are checked to ensure that no les have been

manipulated or damaged during transfer (scratches on CD, Internet connection errors,

etc.).

Depending on the type of the certicate, not only the integrity of the archives can be

validated, but also the true authorship can be proven by online atlas users.

For self made certicates, only the integrity of the archives can be validated. If the

atlas is used online, the user will be shown a warning that the author of the application

can not be veried. To unfold maximum security of signed JARs, a digital certicate has

to be obtained from an international security authority


58 . With an ocial certicate,

the warning will not appear.

57 ZIP compression described at Wikipedia.org: http://en.wikipedia.org/w/index.php?title=ZIP_


%28file_format%29&oldid=185025484
58 .X509 certicates are for example VeriSign (commercial) http://www.verisign.com/ssl/
Issuers of

buy-ssl-certificates/ or CACert.org (community driven): http://www.cacert.org/.


5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 74

5.3.2 Flexibility
As discussed in 4.2.1, the package format must be exible enough to be used oine or

online.

5.3.2.1 Java Web Start The Java Web Start


59 technology has been selected to

achieve the requested exibility of usage. It allows to run a standard Java application

via the Internet by using the Java Network Launching Protocol (JNLP).

To enable Java Web Start (JWS) for the AV, a customized JNLP conguration le

has to be created whenever the atlas is changed. The AC automatically generates JNLP

conguration les when exporting an atlas. The following listing shows a shortened

version of a JNLP le generated for the IIDA2.

1 <? xml v e r s i o n =" 1 . 0 " e n c o d i n g="UTF−8" ?>

2 <j n l p c o d e b a s e=" h t t p : / /www . w i k i s q u a r e . d e / a t l a s / " h r e f =" a t l a s V i e w e r . j n l p "

s p e c=" 1 . 0 + ">

3 <i n f o r m a t i o n l o c a l e =" d e ">

4 < t i t l e>I n t e r a c t v e r Digitaler IMPETUS Atlas 2.0 beta 2007 − 11 − 28</ t i t l e >

5 < d e s c r i p t i o n>IMPETUS</ d e s c r i p t i o n>

6 <o f f l i n e − a l l o w e d />
7 <v e n d o r>IMPETUS</ v e n d o r>

8 <i c o n h r e f =" s p l a s h s c r e e n . p n g " k i n d=" s p l a s h " />

9 </ i n f o r m a t i o n>

10 <i n f o r m a t i o n l o c a l e =" f r ">

11 < t i t l e>A t l a s numerique IMPETUS version 2.0 beta 2007 − 11 − 28</ t i t l e >

12 < d e s c r i p t i o n>IMPETUS</ d e s c r i p t i o n>

13 <o f f l i n e − a l l o w e d />
14 <v e n d o r>IMPETUS</ v e n d o r>

15 <i c o n h r e f =" s p l a s h s c r e e n . p n g " k i n d=" s p l a s h " />

16 </ i n f o r m a t i o n>

17 < s e c u r i t y> <a l l − p e r m i s s i o n s /> </ s e c u r i t y>

18 < r e s o u r c e s>

19 <j 2 s e h r e f =" h t t p : / / j a v a . s u n . com / p r o d u c t s / a u t o d l / j 2 s e " j a v a −vm− a r g s="−

Xmx250m" v e r s i o n =" 1 . 6 + " />


20 <j a r d o w n l o a d=" e a g e r " h r e f =" a v . j a r " p a r t=" m a i n " />

21 <j a r d o w n l o a d=" e a g e r " h r e f =" l i b / g p s y l o n _ b i n −0.5.3/ gpsinput −0.5.3. jar"

p a r t=" m a i n " />

22 <j a r d o w n l o a d=" e a g e r " h r e f =" l i b /JMF − 2 . 1 . 1 e / j m f . j a r " p a r t=" m a i n " />

23 ...

24 < !−− D at ap o ol e nt ry Drohnenlandung−−>


59 Java Web Start Technology webpage: http://java.sun.com/products/javawebstart/
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 75

25 <j a r d o w n l o a d=" e a g e r " h r e f =" v i d e o _ l a n d u n g . j a r " p a r t=" v i d e o _ l a n d u n g " />

26 <p a c k a g e name=" s k r u e g e r . a t l a s . * " p a r t=" v i d e o _ l a n d u n g " r e c u r s i v e =" t r u e " /

>

27 ...

28 < !−− D at a po ol e nt ry Brunnen−−>


29 <j a r d o w n l o a d=" e a g e r " h r e f =" v e c t o r _ b r u n n e n _ p i c . j a r " p a r t="

v e c t o r _ b r u n n e n _ p i c " />

30 <p a c k a g e name=" s k r u e g e r . a t l a s . * " p a r t=" v e c t o r _ b r u n n e n _ p i c " r e c u r s i v e ="

t r u e " />

31

32 </ r e s o u r c e s>

33 <a p p l i c a t i o n −d e s c main − c l a s s =" s k r u e g e r . a t l a s . A t l a s V i e w e r " />

34 </ j n l p>

Recognizable are the multi-lingual sections <information> ... <information/>


(lines 3 and 10) which describe general parameters of the atlas. The <resources> ...
</resources> block (starting line 18) is used to reference all JARs which the application
depends on. Besides the AV itself and other Java libraries, this section also lists all data

pool entries as JAR resources.

JWS not only manages downloading and validation of the required resources but also

keeps a cache on the user's computer. The rst time the atlas is started via JWS it

will download all required JARs. During subsequent starts of the atlas JWS will only

download a JAR if it has been altered on the server.

The server does not need to have any special software running to provide a JWS

application. A link to the JNLP le is sucient for a Java enabled browser to pass

handling of the application to JWS.

5.3.2.2 Execution of the atlas

Oine When the atlas has been obtained on CD-ROM, a start.bat (for Microsoft
Windows) and start.sh (for GNU/Linux and Mac OS) exists in the main folder. Both

les are generated by the AC when packaging the atlas. Executing the appropriate le

will start the atlas. Depending on the settings in Microsoft Windows the atlas will

autostart when the CD-ROM is inserted.

Online When the atlas is accessed online the user must only click a link with a

Java enabled browser. Having agreed to trust the application (the atlas application will
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 76

afterwards have full access to the hard drive) the atlas will open in a separate application

window as if it had been started locally.

5.3.3 atlas.xml conguration le


JWS and JAR are technologies for packaging the atlas and its content. To represent

the content relationships, parameters, translations, and settings another approach is

necessary.

Basically, the AC just has to write all parameters into a settings le, pack the le

into an atlas package and let the AV interpret it. The interesting question is how the

information is stored in the le.

Requirements of openness and re-usability (4.4) demand the AC-AV design not to

store these information in an arbitrary, secret way. It rather has to be clearly and openly

dened so that other FOSSGIS are encouraged to use it.

5.3.3.1 XML and XML Schema Extensible Markup Language (XML) is a general-

purpose markup language standardized by the World Wide Web Consortium (W3C)
60 .

Its primary purpose is to facilitate the sharing of data across dierent information systems

(Eckstein & Eckstein, 2004).

For all the dierent scenarios where XML is applied, domain specic XML languages

are dened. For example, this is the case for SLD, Geographic Markup Language (GML),

or XHTML. For all these languages, explicit language denitions have been developed

which explain valid syntax and semantics for the specic language. These denitions are

usually written in the XML Schema language


61 .

Many advantages can be found when such standardized language denitions are avail-

able. For example the XML Schema le for SLD


62 can be used to validate an arbitrary

SLD document. Validation of XML les against XML Schema denitions can be handled

by freely available Java tools. The AC and AV use such a validation tool to ensure that

SLD documents contain only valid SLD language (in respect to the ocial OGC Schema

denition) before they are interpreted.

60 XML specications at W3C.org: http://www.w3.org/XML/


61 XML Schema language is used to express a set of rules to which an XML document must conform

in order to be considered valid according to that schema (Eckstein & Eckstein, 2004).

62 SLD Schema denitions online: http://arcmap2sld.geoinform.fh-mainz.de/Dateianlagen/SLD_


Schemadateien.rar
5 ATLASCREATOR-ATLASVIEWER IMPLEMENTATION 77

5.3.3.2 Atlas Markup Language For every atlas package, the AC saves the atlas

conguration to a le called atlas.xml. An explicit language has been dened for it

using XML Schema. The language is called AtlasML. Full denition in XML Schema is

available at http://localhost:7272/skrueger/atlas/resource/AtlasML.xsd as soon

as the AC or the AV have been started on the local computer.

The AV uses the XML Schema denition to validate its conguration le before it

is being interpreted. Hence the chances are lower that the AV will behave unexpectedly

due to errors in the conguration le. The validation process will rather point to the

dubiety on start up and exit cleanly.

Another motivation for the accurate denition of the atlas conguration format has

been the possibility for third party programs to create or read the format.

5.3.4 Subsumption
Once the atlas product has been exported it is required to be ready for transfer and

execution on any standard computer. The technologies described above fulll these

requirements and allow the same package to be either burned on a CD-ROM or provided

to users via the Internet.

The package format contains a well described atlas conguration le (atlas.xml)

which gives other FOSSGIS the opportunity to step in on behalf of the AC or the AV.

5.4 Preview of IMPETUS Interactive Digital Atlas (version 2.0)


A preview of the IIDA2 has been created as a proof-of-concept using the AC-AV frame-

work. An online version can be accessed at http://www.wikisquare.de/da-atlas-demo.


It is required to have Java Runtime Environment (JRE) 6 or later installed. A download

link is included on the page. The size of the atlas is 250Mb and the whole atlas will be

downloaded and veried at start up. Depending on the Internet connection this can take

a long time but will only be done once. The next time the atlas is started it will only

download changes of the content.

Table 5 describes a few selected entries of the atlas. It has to be emphasize that the

preview of the IIDA2 has its focus on the abilities of the AV software and not on the

content.
6 DISCUSSION AND CONCLUSION 78

Menu entry Content

Maps-> Benin-> An overview map of Benin. While zooming into the map,

Overview of smaller towns and villages appear. This has been realized with

Benin scale dependent SLD rules (5.2.1.3).

Maps-> Benin-> A LANDSAT image of Benin with an extend of 6692x13638

Landsat pixels has been converted to a TIP layer to be visualizable in

the AtlasViewer (5.1.1.2).

Maps-> Benin-> The blue/white points on the map describe wells and show

Wells and photographs and details when clicked with the Info-tool (18).

Waterspots
Maps-> Benin-> A land use classication of the Haute Vallée de l'Ouémé (en.:

land use Upper catchment of the Ouémé river). The values of the TIP

classification raster layer can be queried with the Info-tool (5.1.2.4).

Extra-> Drone A video with sound. Playback is implemented using the Sun

landing Microsystems JMF (5.1.2.8, 5.2.1.4).

Extra-> Diploma A PDF document of this diploma thesis. The document will be

thesis opened with a local PDF viewer (5.1.2.7).

Table 5: Selected contents of the IIDA2 demo.

6 Discussion and Conclusion

6.1 Evolution of the DGRF design


The digital distribution of geospatial information via multimedia atlases has many ad-

vantages compared to the traditional paper map. The fundamental concept of communi-

cation via paper maps (2.5) though is still inherent and can even be recognized in design

of the DGRF.

Paper map communication ow is as follows: The map maker intends to describe a

spatial situation to an audience. He or she therefore encodes the information in symbols

which can be xed on a map (gure 2). The map can be given to a reader who interprets

the symbols and constructs her or his image of the situation (gure 3).

Basic aspects of paper map communication have had an impact on the design of the

DGRF framework. For example, the main actors of map communication are the map

makers and the map readers. This is reected in the separation of the two programs:

Atlas Authoring Tool (4.1) and Atlas Visualization Tool (4.3). One specic application
6 DISCUSSION AND CONCLUSION 79

for the people that create atlases and one exclusively for the users. The creation of user-

specic programs allowed for optimally adapting them to their dierent requirements.

The design of the atlas package format (4.2) evolved mainly from its function as a

transport medium in atlas communication. The transportation medium properties vary

greatly between paper maps and multimedia atlases. A paper map is a medium which

transports and visualizes the information at the same time. For digital atlas products,

these functions can be separated into geospatial data and cartographic representation

rules (2.4.2).

The media further dier in their versatility and possibilities. Reproduction of paper

maps via printing is costly whereas a digital medium can be replicated cheaper and is al-

ways as good as the original. Not to mention it can be transferred nearly instantaneously

through global communication networks.

Paper maps are static visual representations which the map reader can interpret

directly with his eyes. A multimedia atlas can be animated with sound and video but it

can not be accessed without a computer and software to visualize the contents.

The additional functionality of the digital medium results in an end product that is

a much more powerful commication tool, but its creation is more complicated. However,

once it has been created it is cheaper and easier to reproduce.

The DGRF tries to encapsulate this complexity (see gure 9). The publishers of

a multimedia atlas can do all authoring and publishing tasks with one tool which is

tailor-made for people with GIS experience, but which have no special knowledge of

XML, Internet services, or atlas programming. Packaging the multimedia atlas to the

transferable atlas package format is done with one click in the authoring tool. The

visualization tool for the atlas users is primarily focused on usability and does not require

any previous GIS knowledge.

6.2 Implementation of the AtlasViewer - AtlasCreator framework


The AC-AV framework is a Java and XML based implementation of the DGRF concept.

The abstract tools of the DGRF have been created as the AtlasCreator program for

the atlas makers, and the AtlasViewer program for the atlas users. The implementation

followed a Multimedia in GIS approach (5.1). For basic GIS functionality the open source

GeoTools GIS toolkit has been chosen. GUIs and multimedia functionality have been

added by using the Java Swing library and the Sun Microsystems Java Media Framework.

In total, ten external open source projects have been selected to realize the AC-AV
6 DISCUSSION AND CONCLUSION 80

framework. Their authorship for imported libraries is honored in the license.txt le

included with every atlas package.

In the following section the key features of the AC-AV framework in relation to the

requirements of the DGRF will be discussed critically.

6.2.0.1 Online/Oine distribution In order to also reach users in less developed

countries where ICT is sparse, the design incorporates a exible distribution format that

allows online and oine use. This has been accomplished with JAR and Java Web Start

technology. When the creator presses the export button in the AtlasCreator, a hybrid

package is created. It can be burned on a CD-ROM or copied onto a web server without

any changes. When published online, the web server being utilized is not required to

have any installations, like PHP, MySQL, etc. Any free HTTP server like Lighttpd
63 or

Apache
64 is sucient.

The AtlasViewer 2007 downloads the whole atlas when it is started the rst time.

Successive uses are quick, but a smarter management with dowloading parts on demand

is still needed. This feature has been implemented but temporarily disabled due to a

bug. This feature is the only aw of the current online version and it should be fully

implemented in a future version.

Interoperability To enable communication with as many users as possible the tools

have been designed to be platform independent. The AV and AC tools run with full func-

tionality on Microsoft Windows, GNU/Linux, or Mac OS. This has been accomplished

by using Java as the programming language.

Platform independent video and sound playback have been the most challenging

parts. The current solution of using the Cinepak codec in combination with the Sun

Microsystems JMF is fully functional but requires immoderate amounts of disk space. A

later version will evaluate a transition to the recently released JVLC


65 multimedia player

which allows stronger compression algorithms.

Large raster visualization The geo-visualization component fullls all requirements

of the DGRF. It is based on a shared XULU component and has been extended with

support for tiled image pyramid (TIP) layers. TIP layers can have extensions of tens

63 Lighttpd HTTP server at Wikipedia.org: http://en.wikipedia.org/wiki/Lighttpd


64 Apache HTTP server at Wikipedia.org: http://en.wikipedia.org/wiki/Apache_HTTP_Server
65 JVLC Java Multimedia Library webpage: http://trac.videolan.org/jvlc
6 DISCUSSION AND CONCLUSION 81

of thousands of pixels and the AtlasViewer is still able to perform spatial navigation
66

on standard computers at satisfactory speed. The algorithms and optimizations for

the TIP layers have been provided to the GeoTools community and will most likely be

incorporated into the GeoTools code base.

Optimally, the map maker would import a large raster into the AtlasCreator, and it

would be automatically converted to a tiled image pyramid. In AtlasCreator 2007, large

rasters have to be converted manually before they can be imported (see A.3). This is an

interim solution and should be improved in a later version.

Cartography representation rules The AtlasViewer imports, interprets, and visu-

alizes the full scope of the OGC Styled Layer Descriptor language. This allows for the

creation of powerful scale-dependent attribute-ltered cartographic representation rules

for vector and raster data. The SLD format is an open standard that is used by all

relevant OGC web server implementations. The conversion of ArcMap projects to SLD

representation rules has been possible thanks to the ArcMap2SLD converter developed

by Weiser, 2005.

Developing a complete SLD editor requires tedious and cumbersome GUI programing.

While a grahical SLD editor was planned as part of the DGRF design, only SLD raster

layers can be edited in AtlasCreator 2007. Further development of the AtlasCreator will

focus on a complete editing package.

Standardization and openness Concerning the requirements of openness and stan-

dardization, the AC-AV framework has been programmed in accordance to many OGC

and other de facto standards. EPSG and OGC WKT have been utilized as alternative

representation forms for CRSs, OGC SLD for cartographic representation rules, Geo-

TIFF, OGC GML, ESRI Shapele, and Arc/Info ASCII Grid are accepted as geodata

import formats.

When the AC exports a multimedia atlas it is packaged into the atlas package format

(5.3) which contains all the data and settings for the atlas. All meta information is repre-

sented in a newly developed XML-based language: Atlas Markup Language (AtlasML).

The syntax of AtlasML has been dened with the W3C XML Schema. The language

denition le is openly available as part of the AV and allows other developers to work

with it.

66 Spatial navigation includes panning over the map, as well as changing the scale (zoom).
6 DISCUSSION AND CONCLUSION 82

Internationalization The ability to author multi-lingual releases has been another

important aspect of the DGRF design. The IMPETUS Interactive Digital Atlas version

2.0 will be produced in French, German, and English. The AC-AV framework allows

one to create a single multi-lingual atlas which adapts to the language of the computer

it is started on. The requirement to translate atlas metadata has been implemented as

functions of the AC. The AC enables the author to translate legends, data base columns

and menus entries, as well as the descriptions and titles for every data pool entry. The AC

software produces a warning when a multi-lingual atlas with an uncompleted translation

is about to be exported.

6.3 Conclusion
The design and implementation discussed in this work aimed at developing an applicable

framework for communicating contextualized geospatial information in form of interactive

multi-lingual multimedia atlases from scientic creators to the general public that would

be easy-to-use and inexpensive. A low cost and generic solution that can be used for

publishing online and oine digital atlases is considered necessary because interest and

demand for perceptive geoproducts is growing especially in third world countries. Java

and Open Standard Technologies were chosen as the main tools for developing such a

framework because they can be used at no cost and provide libraries for GIS and Internet

development. A pair of software applications, the AtlasCreator and the AtlasViewer, have

been developed to encapsulate the complexity of creating, publishing and viewing such

products. The use of the open source GeoTools toolkit for a Multimedia in GIS approach

proved to be very eective. The prototype implementation, tested in Benin with local

stakeholders and also available online at http://www.wikisquare.de/da-atlas-demo,


received positive feedback.

The experience described here illustrates that it is possible to facilitate the com-

munication of scientic geospatial products and data unto a general public uneducated

in GIS use, even if they don't have access to fast Internet. It is also possible that, in

the future, other institutions communicate their research ndings and geoproducts to a

greater audience by using the AtlasCreator-AtlasViewer framework developed as part of

this research.
REFERENCES 83

References

Andrienko, G., Andrienko, N., & Voss, H. (2003). GIS for Everyone: The CommonGIS

Project and Beyond. In M. Peterson (ed.), Maps and the Internet, Chapter 8, pp.

131146. Elsevier Science, Oxford.

Bartelme, N. (2005). Geoinformatik: Modelle, Strukturen, Funktionen. Springer, Heidel-

berg.

Berry, J. K. (1993). Beyond Mapping. Concepts, Algorithms, and Issues in GIS. H.

Dennison Parker, Fort Collins.

Bollmann, J. & Koch, W. G. (eds.) (2001). Lexikon der Kartographie und Geomatik in

zwei Bänden. Band 1 A bis Karti. Spektrum Akademischer Verlag, Heidelberg/Berlin.

Bundesamt für Landestopograe, W. (2004). Atlas der Schweiz 2.0 (2004). CD-ROM mit

Begleitbuch. Version 2.0. CD-ROM, DVD.

Bär, H. R. & Sieber, R. (1997). Atlas der Schweiz  Multimedia Version: Abaptierte

GIS-Techniken und qualitative Bildschirmgraphik. In D. Grünreich (ed.), GIS und

Kartographie im multimedialen Umfeld. Kirschbaum Verlag GmbH, Bonn.

Bär, H.-R. & Sieber, R. (1999). Towards High Standard Interactive Atlases. The "GIS in

Multimedia" approach. In Proceeding of th 19th ICA/ACI International Cartographic

Conference ICC, p. 235241, Ottawa.

Cartwright, W., Crampton, J., Gartner, G., Miller, S., Mitchell, K., Siekerska, E., &

Wood., J. (2001). Geospatial Information Visualization User Interface Issues. Carthog-

raphy and Geographic Information Science, 28(1), 119.


Cartwrigth, W. (2007). Development of Multimedia. In W. Cartwrigth, M. Peterson,

& G. Ganter (eds.), Multimedia Cartography, Chapter 2, pp. 1134. Springer, Berlin,

2nd edition.

Casati, R., Smith, B., & Varzi, A. C. (1998). Ontological Tools for Geographic Repre-

sentation. In N. Guarino (ed.), Formal Ontology in Information Systems, pp. 7785.

IOS Press, Amsterdam.

Collins, L. (2007). Cutting the cost of computing - Hardware and softwaremakers are

eager to help emerging economies join the information revolution. Engineering &

Technology, 2(2), 3437.


REFERENCES 84

Compuserve, I. (1990). Graphics interchange format (sm), version 89a (modied). Tech-

nical report, Compuserve, Columbus, Ohio.

Conner, J. O. (2007). Java Internationalization: An Overview. Webpage. Available on-

line at http://java.sun.com/developer/technicalArticles/Intl/IntlIntro; vis-

ited on Jun. 2th 2007.

Crane, N. (2002). The man who mapped the world. Weidenfeld & Nicolson.

Cron, J. (2006a). Graphische Benutzeroberächen interaktiver Atlanten. Konzept zur

Strukturierung und praktischen Umsetzung der Funktionalität. Diplomarbeit, Fachbere-

ich Vermessungswesen/Kartographie. Hochschule für Technik und Wirtschaft Dresden

(FH).

Cron, J. (2006b). Kriterienkatalog. Umsetzung von Funktionen interaktiver Atlanten.

Diplomarbeit, Fachbereich Vermessungswesen/Kartographie. Hochschule für Technik

und Wirtschaft Dresden (FH).

DiBiase, D. (1990). Visualization in earth sciences. Earth & Mineral Sciences, Bulletin

of the College of Earth and Mineral Sciences, 59, 1318.


Dietze, L. & Zipf, A. (2007). Extending OGC Styled Layer (SLD) Descriptor for Thematic

Cartography. In 4th Int. Symp. on LBS and Telecartography 2007.

Eckstein, R. & Eckstein, S. (2004). XML und Datenmodellierung. XML-Schema und

RDF zur Modellierung von Daten und Metadaten einsetzen. Dpunkt, Heidelberg.

Erle, S., Gibson, R., & Walsh, J. (2005). Mapping Hacks. Tips & Tools for Electronic

Cartography. O'Reilly.

Ferguson, T. (2001). Cinepak (CVID) stream format for AVI and QT. Webpage. Avail-

able online at: http://www.csse.monash.edu.au/%7etimf/videocodec/cinepak.


txt; visited Jul. 11th 2007.

Francula, N. (2005). Cartography in the 21st Centruy. Cartography and Geoinformation.

Journal of the Croatian Cartographic Society., 4(4), 415.


Gietler, L., B. H. (2003). Web Mapping a la Cuisine. In Angewandte Geographische

Informationsverarbeitung XV. Beiträge zum 15. AGIT-Symposium., pp. 9197.


REFERENCES 85

Golledge, R. G., Rice, M. T., & Jacobson, R. D. (2006). Multimodal Interface for Rep-

resenting and Accessing Geospatial Information. In J. S. Sanjay Rana (ed.), Frontiers

of Geographic Information Technology, Chapter 9. Springer, Heidelberg.

Goodchild, M. F., Yuan, M., & Cova, T. J. (2007). Towards a general theory of geographic

representation in GIS. International Journal of Geographical Information Science,

21(3), 239260.
Grassmuck, V. (2004). Freie Software. Zwischen Privat und Gemeineigentum. Bundeszen-

trale für politische Bildung, Bonn.

Haggett, P. (2001). Geography. A global synthesis. Pearson Hall, Upper Saddle River.

Hake, G., Grünreich, D., & Meng, L. (2002). Kartographie: Visualisierung raum-

zeitlicher Informationen. Walter de Gruyter, Berlin.

Hardy, P., Briat, M.-O., Eicher, C., & Kressmann, T. (2004). Database-driven Cartog-

raphy from a DLM, with Multiple Representations and Human Overrides. In ICA

Workshop on 'Generalisation and Multiple Representation'.

Herzog, A. (2003). Developing Cartographic Applets for the Internet. In M. Peterson

(ed.), Maps and the Internet. Elsevier, Oxford.

Hrebícek, J. & Konecný, M. (2007). Introduction to Ubiquitous carthography and Dy-

namic geovisualisation with Implications for Disaster and Crisis management. In The

Geospatial Web, Chapter 20, pp. 209214. Springer, Heidelberg.

Huber, S. & Schmid, C. (2003). 2nd Atlas of Switzerland - Interactive Concepts, Func-

tionality and Techniques. In Proc. of the 21th Int. Conference of the ICA, Durban,

South Africa, pp. 13981405.

Huber, S., Sieber, R., & Wipf, A. (2003). Multimedia in der Gebirgskartographie

- 3D-Anwendungen aus dem Atlas der Schweiz - interaktiv 2. Kartographische

Nachrichten, 5.
Hurni, L. & Räber, S. (2004). Atlas of Switzerland 2.0. In Proc. of the 4th ICA Mountain

Cartography Workshop, Núria, Spain, 2004.

Husdal, J. (2001). Can it really be that dangerous? Issues in visualization of risk and

vulnerability. Webpage. Available online at http://www.husdal.com/blog/2001/10/


can-it-really-b.html; visited on Dec. 15th 2007.
REFERENCES 86

IMPETUS (2003). IMPETUS Westafrica. Case studies for selected catchments in dier-

enct climate zones. Final Report 01.05.2000-30.04.2003. IMPETUS, Cologne.

IMPETUS (2006). IMPETUS Westafrika: Fallstudien für ausgewählte Flusseinzugsgebi-

ete in unterschiedlichen Klimazonen. Sechster Zwischenbericht: 1.1.2005 - 31.12.2005.

Issing, L. & Klimsa, P. (1997). Information und Lernen mit Multimedia. Beltz, Weinheim,

2nd edition.

ITU/UNCTAD (2007). World Information Society Report 2007. Beyond WSIS. Interna-

tional Telecommunication Union, Geneve.

Jobst, M. (2007). Standards and Open Source for Cartographic Multimedia Applications.

In W. Cartwrigth, M. P. Peterson, & G. Gartner (eds.), Multimedia Cartography,

Chapter 29, pp. 415426. Springer, Heidelberg, 2nd edition.

Keates, J. S. (1982). Understanding maps. Longman, London.

Kemper, A. & Eickler, A. (2004). Datenbanksysteme - Eine Einführung, 5. Auage.

Oldenbourg.

Kiehle, C., Greve, K., & Heier, C. (2006). Standardized Geoprocessing - Taking Spatial

Data Infrastructures one Step Further. In Processdings of the 9th AGILE Conference

on Geographic Information Science, Visegrád, Hungary, 2006.

Kraak, M.-J. (1999). Visualising spatial distributions. In Geographical Information Sys-

tem. Principles and Technical Issues., Chapter 11, pp. 157173. Wiley, New York.

Kraak, M.-J. (2002). Current trends in visualisation of geospatial data with special

reference to cartography. Indian Cartographer, SDI-01, 319324.


Lake, R. (2007). Ron Lake's blog on geoweb and markup.

Lake, R. & Farley, J. (2007). Infrastructure for the Geospatial Web. In The Geospatial

Web, Chapter 2, pp. 1526. Springer, Heidelberg.

Laurent, A. M. S. (2004). Understanding Open Source and Free Software Licensing.

OReilly.

Longley, P. A., Goodchild, M. F., Maguire, D. J., & Rhind, D. W. (eds.) (1999). Geo-

graphical Information Systems. Principles and Technical Issues. Wiley, Weinheim, 2nd

edition.
REFERENCES 87

MacEachren, A. M. (1994). Visualization in Modern Cartography (Modern Cartography,

Vol 2). Pergamon Press Inc.

MacEachren, A. M. (1998). Visualization - Cartography for the 21st century. In Pro-

ceedings of the Polish Spatial Information Association conference, May, 1998, Warsaw

Poland.

MacEachren, A. M. & Kraak, M.-J. (1997). Exploratory cartographic visualization:

advancing the agenda. Computers and geosciences (special issue), 23(4), 335343.
MacEachren, A. M., Brewer, C. A., & Pickle, L. (1998). Visualizing Georeferenced Data:

Representing Reliability of Health Statistics. Environment & Planning A, 30/9, 1547


1561.

Macgill, J. (2005). An Introduction to the GeoTools Library and the Family of Re-

lated Projects. Presentation. Available online at http://mapserver.gis.umn.edu/


community/conferences/MUM3/eogeo/macgill; visited on Jan 10th 2007.

Marinacci, J. & Adamson, C. (2005). Swing Hacks. O'Reilly Media.

Minami, M. & ESRI (2000). Using ArcMap. GIS by ESRI. Environmental Systems

Research Institute, Redlands, Calif.

Morita, T. (2004). Ubiquitous Mapping in Tokyo. In ICA UPIMap 2004, Tokyo, Japan,

Tokio.

Muehrcke, P. C. (2001). Map Use: Reading, Analysis, Interpretation. J.P. Publications,

Madison, 4th edition.

Neteler, M. & Mitasova, H. (2008). Open Source Gis. A GRASS Approach. Springer,

Heidelberg, 3rd edition.

OGC (2001). Coordinate Transformation Service Implementation Specication. Techni-

cal report, Open Geospatial Consortium, Inc.

OGC (2002). Styled Layer Descriptor Implementation Specication. Technical report,

Open Geospatial Consortium, Inc.

Ogden, J., Adelson, E. H., Bergen, J. R., & Burt, P. J. (1985). Pyramid-based computer

graphics. RCA Engineer, 5(30), 415.


REFERENCES 88

Ormeling, F. (1993). Ariadne's thread - Structure in multimedia atlases. In Proceedings

ICA conference Köln 1993., pp. 10931100.

Ormeling, F. (1995). Atlas information systems. In Proceedings 17th International Car-

tographic Conference, Barcelona 1995., pp. 21272133, Barcelona. Institut Cartograc

de Catalunya.

Ormeling, F. (1998). Atlas Terminology and Atlas Concepts. In Ormeling, Tichelaar,

& Villanueva (eds.), Proceedings of the Seminar on Modern Cartography: Maps for

Navigating the Information Highway. Bandung, Indonesia, 1997, pp. 513, Enschede.

Ormeling, F. (2007). Map Concepts in Multimedia Products. In W. Cartwrigth, M. P. Pe-

terson, & G. Gartner (eds.), Multimedia Cartography, Chapter 8, pp. 105116. Springer,

2nd edition.

Peterson, M. (2003). Maps and the Internet: An Introduction. In M. Peterson (ed.),

Maps and the Internet, Chapter 1, pp. 116. Elsevier Science, Oxford.

Peterson, M. P. (2007a). Elements of Multimedia Cartography. In W. Cartwrigth,

M. P. Peterson, & G. Gartner (eds.), Multimedia Cartography, Chapter 5, pp. 6373.

Springer, 2nd edition.

Peterson, M. P. (2007b). The Internet and Multimedia Cartography. In W. Cartwrigth,

M. P. Peterson, & G. Gartner (eds.), Multimedia Cartography, Chapter 3, pp. 3550.

Springer, 2nd edition.

Robinson, A. H., Morrison, J. L., Muehrcke, P. C., Kimerling, A. J., & Guptill, S. C.

(1995). Elements of Cartography. Wiley, New York, 6th edition.

Romagnoli, D. & Giannecchini, S. (2006). Coordinate Reference Systems, an extensive

introduction. Available online at http://simboss.files.wordpress.com/2006/03/


CRS.pdf; visited on Nov. 10th 2007.

Schmitz, M. (2005). Entwicklung einer generischen Plattform zur Implementierung von

Simulationsmodellen am Beispiel der Landnutzungsmodellierungw3s. Diplomarbeit,

Rheinische Friedrich-Wilhelms-Universität zu Bonn.

Schneider, B. (1999). Integration of analytical GIS-functions in Multimedia Atlas Infor-

mation Systems. In 19th International Cartographic Conference, Ottawa, Canada.


REFERENCES 89

Shirley, P. (2005). Fundamentals of Computer Graphics. AK Peters, Ltd., 2nd edition.

Sieber, R. & Huber, S. (2007). Atlas of Switzerland 2  A highly interactive thematic

national atlas. In W. Cartwrigth, M. Peterson, & G. Ganter (eds.), Multimedia Car-

tography, Chapter 12, pp. 161182. Springer, Berlin, 2nd edition.

Stahr, K., Herrmann, L., & Vennemann, K. (1999). Atlas of Natural and Agro-

nomic Resources of Niger and Benin. Webpage. Available online at http://www.


uni-hohenheim.de/~atlas308; visited on June 1th 2007.

Stamp, M. (2003). Once upon a Time-Memory Trade O. Technical report, San Jose

State University, California, USA.

Tanenbaum, A. S. (2002). Moderne Betriebssysteme. Prentice Hall/Pearson Studium,

2nd edition.

Thamm, H.-P., Schütz, O., Recha, P., Judex, M., Menz, G., Christoph, M., Speth, P.,

Diekkrüger, B., Jaeger, A., & Homann, R. (2005a). Appropriate Distribution of

Research Results: The IMPETUS Atlas. Poster.

Thamm, H.-P., Schütz, O., & Christoph, M. (2005b). IMPETUS Atlas - Benin - Research

Results. University of Bonn, Remote Sensing Research Group.

Thamm, H.-P., Recha, P., Judex, M., & Schütz, O. (2006). Ein ezientes System für die

Darstellung von Forschungsergebnissen  Der Interaktive Digitale IMPETUS Atlas.

In AGIT 2006  Symposium und Fachmesse für angewandte Geoinformatik, 04.07.-

06.07.2006, Salzburg.

Thamm, H.-P., Bode, T., Schmitz, M., & Cremers, A. (2007). XULU  Eine JAVA-

basierte Plattform zur Implementierung von Simulationsmodellen. In Beiträge zum

AGIT-Symposium.

Trévisan, J. (2004). From DLM to Multi-Representation DCM - Modelling and Applica-

tion on Buildings. In ICA Workshop on 'Generalisation and Multiple Representation'.

Ullenboom, C. (2007). Java ist auch eine Insel. Programmieren mit der Java Standard

Edition Version 6. Galileo Press, 7th edition.

van Elzakker, C. P. J. M. (2004). The use of maps in the exploration of geographic data.

Koninklijk Nederlands Aardrijkskundig Genootschap, Faculteit Geowetenschappen,


REFERENCES 90

Universiteit Utrecht; International Institute for Geo-Information Science and Earth

Observation.

W3Schools (2007). OS Platform Statistics. Webpage. Available online at http://www.


w3schools.com/browsers/browsers_os.asp; visited on Dec. 20th 2007.

Weibel, R. & Dutton, G. (1999). Generalising spatial data and dealing with multiple rep-

resentations. In Geographical Information Systems. Principles and Technical Issues.,

Chapter 10, pp. 125156. Wiley, Cambridge.

Weiser, A. (2005). Automatisierte Generierung von Styled Layer Descriptor-Dateien aus

ESRI ArcGIS-Projekten zur Publikation mit OGC-konformen Mapservern. Diplomar-

beit, FH Mainz, University of Applied Sciences, Mainz.

Weiser, A. & Zipf, A. (2005). A visual editor for OGC SLD les for automating the

conguration of WMS and mobile map applications. In W. C. Georg Gartner &

M. P. Peterson (eds.), 3rd Symposium on Location Based Services and TeleCartography.

Vienna. Austria. 2005., Lecture Notes in Geoinformation and Cartography, pp. 265

278. Springer.

Welch, T. A. (1984). A technique for high-performance data compression. Computer,

17(6), 819.
Wiesmann, S. (2007). Smart Legend. Aufbau und Interaktivität digitaler Kartenlegenden.

Diplomarbeit, Geographisches Institut der Universität Zürich., Zürich.


A APPENDIX 91

A Appendix

A.1 Preparing data to be imported into the AtlasCreator


When you plan to publish your data with the Atlas Creator, you have to

1. Use lower case le names

The Microsoft Windows operating system doesn't distinguish between upper-

and lower-case le-names. On Microsoft Windows the le-names Trees.shp and

trees.SHP point to the same le and two les with these names can not exist in

one directory. GNU/Linux and Mac OS are case-sensitive and therefore Trees.shp
and trees.SHP describe two dierent les. This can lead to problems when the

AC is used on Microsoft Windows and later the AV is run on GNU/Linux or Mac

OS. All les should be named lower case only. Avoid image.JPG or Image.jpg -

always use image.jpg.

2. Don't use any special, language-specic characters in le-names

File names should only contain standard ASCII characters. Optimally only digits,

letters a to z and the underscore  _ character. Umlauts like ä, ö, ü
or accented letters like é, ú, and û should be avoided. Technically the

problem of dierent char sets is solved by the denition of the UTF-8 char-set, but

in reality many computers


67 still do not run on UTF-8 and so an ärzte.shp on a

German hard-drive might possibly appear as a érzte.shp or other unpredictable

le-name in another country. References to ärzte.shp might therefor fail if the

atlas is run on a foreign computer.

3. Use one folder per layer

When preparing vector or raster layers, create one folder for each layer. All les

related to one layer shall be saved in the same folder. This includes images, PDFs

or videos linked to any geo-object in the ESRI Shapele.

A.2 Image data compression algorithms


Image data compression algorithms split up into two main classes: lossless and lossy

algorithms. The lossless image data compression algorithms searches for ways to present

67 To be technically correct, it is not the computers that have to be converted to UTF-8, but the le

systems.
A APPENDIX 92

exactly the same data as in the original image, but expressed in a more compact way.

The most famous example of a lossless data compression algorithm is the The Lempel-

Ziv-Welch lossless data compression algorithm (LZW) algorithm, developed in 1984. It is

based on the assumption, that the input data will not make uniformly distributed use of

all values that can be expressed by the image's color depth, as well as the fact, that some

symbols occur in repeating sequences in the original image, e.g. n pixels of the same

color C in a row can be expressed by n*C. Large English texts are typically compressed

by LZW to half of their original size. Compression of graphics with discrete objects in

it (diagrams, charts, etc.) works well with lossless compression algorithms. Real world

photographs (for example a LANDSAT scene) are much harder to compress with lossless

image data compression algorithms. The resulting le might even be bigger than the

original data Welch (1984).

Lossy image compression - as the name hints - looses information while compressing

the image. The trick is that by loosing only a few percent of the actual information, a

compression to only a tenth of the original le-size is possible. The most well known lossy

image data compression algorithm is the JPEG compression. The name JPEG stands for

Joint Photographic Experts Group, the name of the committee that created the standard.

Like for most lossy compression algorithms, a quality/compression-ratio has to be dened

when compressing. This makes it possible to use JPEG to either create tiny (le size)

and fast previews of a photograph, as well as for creating high-original quality images

that still are reasonable smaller (le size) than the original image. How the lossy-image

compression artifacts are perceived by human eye depends highly on the actual image.

Graphics with discrete objects as mentioned above perform very bad. A sharp border

between two monochrome areas is nearly not preservable when using JPEG even with

a high quality/compression ratio. In the case of satellite images, the JPEG algorithm

performs well, meaning that the le size can be signicantly reduced without the human

eye perceiving a loss of quality. If the image is not stored exclusively for human watching,

but to be further processed by digital image algebra, lossy image compression algorithm

should be avoided because the actual pixel values would be changed.

A.3 Preparing TIP layers for import into AC


Big raster les have to be converted to Tiled Image Pyramid (TIP) layers to be eciently

represented in the AV. In AC 2007 version, the process of converting a big raster to a TIP

is not yet automatized. Once a raster has been manually converted, the AtlasCreator
A APPENDIX 93

software recognizes it as a TIP layer and can than handle rasters of virtually unlimited

size. The process of creating a TIP layer consists of six steps:

1. The original raster is being converted to a GeoTIFF.

2. The creator of the atlas product has to decide how many levels of resolution he/she

wants and how they shall be distributed. For the IIDA2 the author has chosen the

following ve levels: 100%, 75%, 50%, 25% and 10%. The original raster was then

down-sampled to the four lower resolutions using a bi-linear sampling algorithm

and each resulting le was copied into its dedicated folder.

3. For each level the raster has to be tiled into many smaller les. Choosing the size

of the tiles has an important impact on the performance. When the user selected

a region by zooming or panning the AtlasViewer software, the algorithm has to

deliver the raster for the selected view-port. Therefor the view-port is intersected

with the tiles. At the border of the view-port, many tiles are only partly visible in

the view-port. Nevertheless, these partly intersecting tiles have to be loaded fully

into the memory. If the size of the tiles is chosen too big, too much memory is

dissipated by loading all the intersecting border tiles fully into memory.

If the tiles are chosen very small, the overhead of intersecting tiles in memory is

minimizing. But another problem arises: Opening a le means an administrative

task for any operating system. Microsoft Windows and GNU/Linux and Mac OS

all have a number of maximum open les. If the operating system manages more

than a few thousand open les at the same time the system will even stop. When

choosing the size of one image tile, one has to be aware that the algorithm will

probably be executed many times per minute and the map might contain more

than just one TIP layer. So the sum of open les on the system due to the TIP

algorithm will multiply.

When making an estimate of the view-port's size, the dimension of a high quality

monitor should be selected. Most display resolutions have an aspect ratio of 4:3 or

16:9. However, the aspect ratio of the view-port in the AV is not xed, because the

window of the software can be arbitrarily re-sized by the user - but by probability it

will be distributed around the aspect ratio of the monitor. Concluding, the aspect

ratio of the tiles should roughly be the same as the aspect ratio of the monitor.

This leads to tiles in the landscape format. The LANDSAT TIP layer in the IIDA2

was created with a tile size of about 320x210px². The exact dimension has to be
A APPENDIX 94

Figure 28: JPEG compression artifacts in a TIP layer at low resolution

adjusted for every resolution level, as an integer divider had to be selected to make

all tiles in one resolution level the same size


68 .

4. To optimize the process of nding the tiles that intersect with the view-port, a

spatial index of all tiles is needed. This spatial index is implemented as an ESRI

Shapele which contains the outline of all tiles as polygons. The only attribute of

these geo-objects is the le name of the tile. One such index is needed per level

of resolution. The GDAL contains the command line tool gdaltindex by Frank

Warmerdam which automatically generates an tile index ESRI Shapele. This

command line tool has to be used on each level of resolution.

5. Tuning the overall size of the layer

The TIP layer - including all it's pre-calculated scales - consumes more disk space

than the original raster image because of the additional down-sampled levels. De-

pending on the count of resolution levels the size of the TIP can be a multiple of

the original. The disk use of the pyramid layers can be modied by applying image

data compression algorithms. The compression strategy included in the IIDA2 is

as follows: At the highest resolution level a lossless LZW compression is applied to

each tile. For all other resolution levels a lossy JPEG compression was applied
69 .

Remote-sensing imagery performs well with JPEG compression and in average the

68 In the 50% resolution level the full raster has an extend of 4121x3388px². The raster was divided

into 13x16 tiles with a dimension of 317x211px² each. This corresponds to an aspect ratio of 1.47.
A APPENDIX 95

size was reduced to one third.

Lossy JPEG compression can't handle discrete edges well. This can be observed

while using the IIDA2. Figure 28 shows defective edges where the national border

of Benin has been clipped. This trace of lossy compression is only inherent to the

compressed resolution levels. At the highest resolution level the artifacts vanish.

6. When steps 1 to 5 have been performed, one nal text based conguration le has

to be created. It lists the directory names and resolutions of all levels. This le can

then be imported into the AC by drag'n'drop. When exporting an atlas created

with the AC, all folders and tiles that are part of the TIP layer are packed into

one JAR.

69 The JPEG compression quality factor was set to 85%. The JPEG quality parameter is nonlinear

parameter.

Вам также может понравиться