Information Systems in Earth Management

From Science to Application

Results from the First Funding Period(2002-2005)

GEOTECHNOLOGIENScience Report

No. 8

Information Systems in Earth Management

ISSN: 1619-7399

In Germany the national programme »Information-Systems in Earth Management«has been initiated in 2002 as part of the R&D-Programme GEOTECHNOLOGIEN.Between 2002 and 2005 six joint projects have been funded with about 4 MillionEuro by the Federal Ministry of Education and Research. All projects were carriedout in close cooperation with various national and international partners from aca-demia and industry.

This report highlights the scientific results from this funding period addressing thefollowing objectives:

- Semantical and geometrical integration of topographical, soil, and geological data - Rule based derivation of geoinformation - Typologisation of marine and geoscientifical information - Investigations and development of mobile geo-services- Coupling information systems and simulation systems for the evaluation of trans-

port processes

The GEOTECHNOLOGIEN programme is funded by the Federal Ministry for

Education and Research (BMBF) and the German Research Council (DFG)

No.

8In

form

atio

nSy

stem

sin

Earth

Man

agem

ent

GEO

TECH

NO

LOG

IEN

Scie

nce

Repo

rt

GEOTECHNOLOGIENScience Report

Information Systems in Earth Management

From Science to Application

Results from the First Funding Period(2002-2005)

No. 8

Number 1

Impressum

SchriftleitungDr. Ludwig Stroink

© Koordinierungsbüro GEOTECHNOLOGIEN, Potsdam 2006ISSN 1619-7399

The Editors and the Publisher can not be held responsible for the opinions expressed and the statements made in the articles published, such responsibility resting with the author.

Die Deutsche Bibliothek – CIP Einheitsaufnahme

GEOTECHNOLOGIEN; Information Systems in Earth Management,From Science to Application – Results from the First Funding Period (2002-2005)Potsdam: Koordinierungsbüro GEOTECHNOLOGIEN, 2006(GEOTECHNOLOGIEN Science Report No. 8)ISSN 1619-7399

Bezug / DistributionKoordinierungsbüro GEOTECHNOLOGIENHeinrich-Mann-Allee 18/1914473 Potsdam, GermanyFon +49 (0)331-620 14 800Fax +49 (0)331-620 14 801www.geotechnologien.degeotech@gfz-potsdam.de

Bildnachweis Titel / Copyright Cover Picture: M. Butenuth

Preface

In Germany the national programme »Infor-mation-Systems in Earth Management« hasbeen initiated in 2002 as part of the R&D-Programme GEOTECHNOLOGIEN. After apublic call, more than 40 project proposalshave been evaluated in an international two-step review procedure, involving severalexperts from four different countries. Finallysix joint projects were recommended and fun-ded by the Federal Ministry of Education andResearch (BMBF) with about 4 Million Eurofor a three year funding period (2002 –2005). The research projects - involving 15partners from academia and industry - cove-red the following key topics:

- Semantical and geometrical integration oftopographical, soil, and geological data

- Rule based derivation of geoinformation - Typologisation of marine and geoscientific

information - Investigations and development of mobile

geo-services- Coupling information systems and simula-

tion systems for the evaluation of trans-port processes

As part of the international Munster GI Days,the closing workshop took place on June 22,2005. More than 60 scientists unveiled theirresults in the presence of the internationalreviewers. The highlights of the research pro-jects are presented in this report.

Information technologies will remain animportant focus of the R&D-Programme GEO-TECHNOLOGIEN. Interoperable informationarchitectures are e.g. an essential link in theearly warning chain. Therefore, their systema-tic further development and implementationwill be a main focus of future research activi-ties in the development of early warningsystems. Corresponding research projects willbe started in early 2007 in the frame theR&D-Programme GEOTECHNOLOGIEN. Theintelligent combination of various researchelements with the goal of a global EarthSystem Management is thus being implemen-ted piece by piece.

Ludwig Stroink

1

Table of Contents

Rule Based Derivation of Groundwater Vulnerability on Different ScalesAzzam R., Kappler W., Kiehle C., Kunkel R., Meiners H.G. Wendland F. . . . . . . . . . . . . . . . . 2 - 11

Overcoming Semantic Heterogeneity in Spatial Data InfrastructuresLutz M., Christ I., Witte J., Klien E., Hübner S. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 - 31

Advancement of Mobile Geoservices: Potential and ExperiencesBreunig M., Bär W., Thomsen A., Häußler J., Kipfer A., Kandawasvika A., Mäs S.,

Reinhardt W., Wang F., Brand S., Staub G., Wiesel J. . . . . . . . . . . . . . . . . . . . . . . . . . . 32 - 51

Development of a data structure and tools for the integration of heterogeneous geospatial data setsButenuth M., Gösseln G. v., Heipke C., Lipeck U., Sester M., Tiedge M. . . . . . . . . . . . . . . . 52 - 73

ISSNEW – Developing an Information and Simulation System to Evaluate Non-point Nutrient Loading into WaterbodiesDannowski R., Arndt O., Schätzl P, Michels I., Steidl J., Hecker J.-M.,

v. Waldow H., Kersebaum K.-C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 - 87

Marine Geo-Information-System for Spatial Analysis and Visualization of Heterogeneous Data (MarGIS)Schlüter M., Schröder W., Vetter L., Jerosch K., Peesch R., Köberle A.,

Morchner C., Fritsche U. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 - 101

Notes

Rule Based Derivation of GroundwaterVulnerability on Different Scales

1. Introduction The project »Development of an informationinfrastructure for the rule based derivation ofgeoinformation from distributive, heterogene-ous geodata inventories on different scaleswith an example regarding the groundwatervulnerability assessment« has been fundedfrom 2002 to 2005 by the BMBF-/DFG-pro-gramme »Geotechnologien – InformationSystems in Earth Management« (project num-ber 03F0372A). The overall goal of the projectwas the development of a Spatial Data Infra-structure (SDI) for the processing of geoinfor-mation in a rule-based manner, independentof scale. This was achieved by providing a»Geo Web Service Groundwater Vulnerability«(GWV) as an integral part of a web-based deci-sion support system. The Geo Web ServiceGWV accesses distributed geodata inventoriesprovided by web service technology and pro-cesses the heterogeneous base data to consi-stent information. The concept for determining groundwater vul-nerability (Hölting et al., 1995) serves as a geo-scientific case study. This case study aims toprovide base data and methods for the deve-lopment of a SDI in order to process distribu-ted geoinformation. Therefore, the project hasboth a geoscientific and an information tech-nological (IT) focus.

2. Spatial scale and uncertainty analysis The geoscientific objectives of the project con-

cern the investigation of the consequencesusing data originated from different scales onthe example of the derivation of the ground-water vulnerability according to Hölting et al.(1995). This method considers the intrinsicsusceptibility of the groundwater systemdepending on the properties of the aquifercoverage (field capacity of the soil, petrogra-phic structure) and the associated sources ofwater and stresses for the system (recharge,travel distance). From this investigation ruleswere compiled and implemented permittingconsistent spatial information to be derivedand displayed from geodata recorded on diffe-rent scales and in different formats. In order toderive these rules, the effects of spatial infor-mation and its uncertainties on the derivedgroundwater vulnerability were analysed atdifferent scales: at the micro scale (~1:5,000),the meso scale (~1:25,000) and the macroscale (<1:50,000). Three study areas, located in the south-western part of the Federal State of NorthRhine-Westphalia (NRW), representing the dif-ferent scale levels, have been selected as studyareas: the rivers Rur and Erft, Inde andSaubach (see Azzam et al., 2003). These areaswere selected in particular because they showthe greatest possible variability of natural areaand anthropogenic factors, which is alsoreflected in the data available, their resolutionand quality. The method, data and results of the assess-ment of groundwater vulnerability has already

Azzam R. (1), Kappler W. (2), Kiehle C. (1), Kunkel R. (3), Meiners H.G. (2), Wendland F. (3)

(1) RWTH Aachen University, Chair for Engineering Geology and Hydrogeology, Lochnerstraße 4-20,

52064 Aachen, Germany, {azzam|kiehle}@lih.rwth-aachen.de

(2) ahu AG Wasser Boden Geomatik, Kirberichshofer Weg 6, 52066 Aachen, Germany, {w.kappler|g.meiners}@ahu.de

(3) Research Centre Jülich, Programme Group Systems Analysis and Technology Evaluation, 52425 Jülich, Germany,

{r.kunkel|f.wendland}@fz-juelich.de

2

been described in detail by Bogena et al.(2004) and will not be discussed in detail in thiscontext. However, the results show that theassessment groundwater vulnerability leads todifferent results depending on the scale of thedata used for the derivation. The reason forthis is mainly due to the fact that the know-ledge of the deeper subsurface, which is animportant factor for vulnerability assessment,depends on the scale of the data. Whereas onthe micro scale borehole data provide detailedknowledge on the three dimensional structureof the subsurface, macro scale data, oftengeneral geological maps, are containing onlytwo dimensional information. Therefore,assumptions required for the »unknown« cha-racteristics of the subsurface missing in thedata may lead to different results concerninggroundwater vulnerability assessment. On the other hand, the usability of the infor-mation system requires information on thequality of the presented results. Therefore,methods for quality assessment of the inputdata and calculation results have been develo-ped. These methods include the quantificationof numerical uncertainties using Gaussian errorpropagation. For the groundwater rechargelevel - one input parameter for the groundwa-

ter vulnerability assessment - this has beendiscussed already in detail in Bogena et al.(2005). For other input parameters uncertaintyassessment has been performed on the basisof best case / worst case calculations. In Figure1 the results of the groundwater vulnerabilityand uncertainty assessment is shown for theexample of the macro scale area. It becomesclear that the vulnerability as well as its uncer-tainty varies strongly within the investigationarea depending on the individual situation at acertain site.

It can be concluded that a spatially differentia-ted assessment of groundwater vulnerabilityand the estimation of uncertainties is possibleon the different scale levels. Results are credi-table from geoscientific point of view.However, the results are different on differentscale levels due to different data. Because theconcept of groundwater vulnerability accor-ding to Hölting et al. (1995) contains a signifi-cant amount of »expert judgement« and inter-pretation of the data, validation of the finalresult (GWV) is not possible. Therefore, thequality of groundwater vulnerability assess-ment can only be evaluated within and notacross scale levels.

Figure 1: Groundwater vulnerability and uncertainty ofvulnerability classes due to data uncertainties

3

3. Implementation of the Spatial DataInfrastructure The Spatial Data Infrastructure (SDI) shouldreach pilot operation within the project’s ove-rall duration processing distributed geodata ina standardised way. If the effort is acceptable,user demands should be met in an appropria-te way. User workshops have been organizedto investigate the demands of providers andusers. An incremental, iterative model of soft-ware development has been applied to meetchanging requirements. The SDI consists of web-based componentsand services, which can be divided into threetiers (see figure 2): 1. The data tier, representing the distributed

data inventories and providing access tometadata and catalogue service.

2. The business logic tier, implementing thegeoprocessing capabilities and the compo-nents needed for the rule based derivation.

3. The presentation tier, providing a web-based graphical user interface to interactwith the system.

3.1 Data Tier The data tier contains both descriptive datainformation (metadata) required by the systemto select the appropriate input data forgroundwater vulnerability assessment and theinput data itself. All elements of the data tierwere implemented as individual web-servicescompliant to current OGC specifications (fore-most Web Coverage Service, Web Feature

Service and Catalogue Service Web (see Evans2003, Vretanos 2002, and Nogueras-Iso et al.2005, respectively)). Metadata about available and used data arerequired for the tasks of the other tiers. Fromthe case studies it became evident that infor-mation concerning data quality is very impor-tant for uncertainty analysis and the assess-ment of the results presented through thesystem by the users. The different availablecommon metadata standards (e.g. FGDC, ISO,Dublincore) were evaluated with respect totheir capability in specifying data quality infor-mation. From this evaluation the ISO 19115standard has been selected to be used, themandatory elements to be specified were defi-ned. Data quality was characterized by generalnumeric uncertainties (e.g. 10%), referencesto digital uncertainty maps provided by web-services and / or other qualitative information.For all input data used by the system appropri-ate metadata sets, containing information e.g.about data formats, coordinate systems, dataquality and online access (addresses, accessrestriction etc.) have been created. Technically, the data structure was defined byan XLM Schema document according to theISO 19115 specification and reproduced in anORACLE 10g database by registering the XMLSchema. All metadata have been verifiedaccording to this schema and stored in thedatabase as the XMLTYPE records. Business logic and presentation tiers communi-cate with the metadatabase via an OGC CS-W

4

Figure 2: Spatial Data Infrastructure, conceptual view

2.0-compliant catalogue service. For this pur-pose, the existing deegree2 (http://www.dee-gree.org) catalogue service was modified andextended to support both the used ISO 19115metadata and XMLTYPE data.

3.2 Business-Logic Tier The business logic tier is the mediator betweenthe data tier and the presentation tier. It provi-des the main service for generation of infor-mation and rule based derivation of geoinfor-mation. The two main components implemen-ted are the data processing component andthe rule based derivation component, whichare introduced in the following two sections.

3.2.1 Data Processing Component According to figure 2 the business logic tier isthe centre of any SDI, thus it is placed betweenthe data tier and the presentation tier. It is capa-ble of integrating several data sources of hetero-geneous origin. The business logic tier consistsof several components which encapsulate alltasks needed for accessing data, generatinginformation out the data and forwarding resultsto the presentation tier (i.e. the user). Inside thebusiness logic, all kinds of tasks performed con-ventionally in geo information systems, can beimplemented. The tasks needed for geoproces-sing, i.e. the processing of spatial data accor-ding to a strictly defined geospatial algorithm,consist mainly of three elements: 1. Connection of web services which serve as

data providers 2. Application of the geospatial algorithm (in

this case: the calculation of groundwatervulnerability)

3. Preparation of information for output (e.g.on a handheld device or through a web-based information system)

According to the principles of service-orientedarchitecture (Chappell and Jewell, 2003), fromwhich the concept of spatial data infrastructu-res is derived, the business logic componenthas also been implemented as a web service. To generate uniform information, three stepshave to be undertaken: 1. Providing factors as OGC compliant services

(either WFS or WCS)

2. Accessing the services in order to acquirethe provided data, to transform all data togrid data and to use map algebra (Tomlin,1990) for calculating the geoinformation»Groundwater Vulnerability«

3. Providing the result as W3C compliantSOAP web service for platform-indepen-dent consumption

Figure 3 gives an overview of the service inter-action. Clients (here: a web application, gene-rally: any kind of http client) access an integra-tive SOAP web service »Groundwater Vulner-ability« which encapsulates services forauthentication and the assessment of ground-water vulnerability. The base data are not cou-pled with the system. An OGC-compliant cata-logue service acts as a data broker for distribu-ted data services. The data is computed by amap algebra service. The results of this processare afterwards incorporated into a map-likedisplay by the »base service groundwater vul-nerability mapping«. The groundwater vulner-ability map classifier acts as a statistic calcula-tion module, computing map parameters, suchas mean distribution of values, minimum/maximum values of computed results, etc.

All the complexity remains hidden to the servi-ce consumer who just accesses the integrativesoap web service which is well defined by aWSDL-interface . The web application consuming this web servi-ce has not to be installed on the same machi-ne as the one providing the service. Figure 4shows four screens of the information systemsin front of a map of the study area. The maintasks performed by users can be described as: 1. select the area of interest based on a

topographic map provided by a publicdata provider

2. inform by means of generating geoinfor-mation from distributed data inventories

3. compare by generating several maps basedon different data sources and afterwards letthe system compare the maps by e.g. mapalgebra tasks like subtraction, addition, etc.

4. decide on a specific topic by using theinformation generated by the system

5

6

3.2.2 Rule Based Derivation Component – TheBusiness Rules Approach (Kardasis and Loucopoulos, 2004) define busi-ness rules as »Projections of external cons-traints on an organisation’s way of working,and on its supporting information systems»functionality«. Those »external constraints«projected on spatial information are for exam-ple the limits of a groundwater model, the vali-

dity of geological assumptions related to scalelevels or restrictions regarding generalizationof data sets. This concept is enhanced by Wan-Kadir andLoucopoulos (2004) who extend the businessrule approach by the concept of »evolvablesoftware architecture«. This definition aims toseparate the code from the business logic. Thestrict separation allows software to grow with

Figure 3: Communication schema inside the developed SDI

7

new requirements. It focuses on a just-in-timeintegration of business logic. The advantages,especially for the enterprise technologies, arefast integration of altered market situations,integration of new laws, etc.

The relevant aspects of business rules for theproject are: - Strict separation of programming code from

business logic - The definition of restrictions on base data

(e.g. scale levels) - User-defined rule alteration (during runtime) - Integration of already existing metadata This architecture (figure 5) enables a strictseparation of domain experts, software engi-neers and users. The software engineers choo-se the implementation technology which isfully transparent to the user. Strictly defined metadata schemes are not sui-table for inserting business rules. Their purpo-se is to describe the underlying spatial datasets. Business rules have to be defined in aseparate rule repository. To ensure interaction

in distributive software environments the defi-nition of business rules in form of eXtensibleMarkup Language (XML) (Harold and Means,2004) files is a suitable way which has beenfollowed in this project. An XML schema definition has been developedto specify the valid data types for any givenrule set. The XML schema defines for exampleminimum or maximum scale for integrating afile into the process of information retrieval ordefining another data set, from which the pro-posed information could be derived.

The XML schema allows an easy integration ofconstraints in the process of data selection.XML files validated against the schema containsimple rules which are transformed from textu-al representation of expert’s knowledge. A business rule declaration could be specifiedcontaining at least the following entries forany given spatial data set: - minimum and maximum values for scale

restrictions - algorithms for generalization

Figure 4: The main tasksof the developed spatialdata infrastructure

8

- URL to alternative data sets - URL to data set, from which the current

data set is derived - rules for deriving factors from base data

(e.g. how to convert usable field capacity tofactor soil)

- textual use restrictions (containing non-numeric information about data quality incontext of different scales)

- any other rule description Any XML file containing rules has to be valida-ted against XML schema in order to ensurecorrectness of all entries (e.g. preventing theuser from entering a negative scale level). Thegeneration of XML files is easy. Users simply fillin web forms to define, for example, scalerestrictions. Three interfaces provide standardized commu-nication between the rule component and the

user (human or machine): - GetCapabilities: well-known of any Open

Geospatial Consortium compliant WebService (OWS), this interface provides theservice metadata

- GetRules: retrieves all rules defined for aspecific data set

- DefineRule: Definition (and alteration) ofnew rules for a specific data set

Those interfaces allow the implementation ofthe rule component into the SDI and also theintegration in any other distributed environ-ment due to its generic nature and high-level-abstraction as well.

3.3 Presentation Tier The main objective of the presentation tier wasthe development of an exemplary web appli-cation which provides easy to handle functio-

Figure 5:System of rulebased derivation ofgeoinformation

9

nalities to analyse the different possibilities ofcalculating groundwater vulnerability on diffe-rent data sets. The application should workwithout proprietary software components. Itshould also communicate via standardisedinterfaces (W3C, OGC) with other componentsin order to be exchangeable. To meet technical and user requirements,which were discussed in workshop sessions, anexemplary web application has been develo-ped using Apache turbine framework andvelocity. It contains modules - to communicate with the Business logic - to analyse user requests - to transform GWV results into graphical out-

put (maps, diagrams) and - to calculate statistics.

The Business logic and presentation tier arecommunicating via http using WSDL interfaces(W3C compliant).

The most important functionalities of the pre-sentation tier are: - An easy to handle predefinition of the area

under investigation (figure 6) using an exi-sting Web Map Service with backgroundinformation (Topography)

- Provision of original data and GWV results - Provision of metadata and data uncertainties - Clear presentation of GWV results in the

form of maps, diagrams or tables - Calculation of GWV variations by varying

input parameters (e.g. scale) and - Calculation of differences between GWV

variations (figure 7).

In workshop users emphasized the benefits of calculations on distributed data and the adequate presentation of metadata anduncertainties.

4. Conclusions Referring to the objectives of the BMBF-/DFG-programme »Geotechnologien – InformationSystems in Earth Management«, the project»Geo Web Service Groundwater Vulnerability«delivers different results of research. First of all, the Geo Web Service GWV showshow distributed heterogeneous data can beprovided as OGC compliant data services(WFS, WCS). The data services are directly usedby the Geo Web Service GWV, but can also beused by other spatial data infrastructures. The service oriented provision of data leads to

Figure 6:Predefinition ofthe area underinvestigation

10

syntactical interoperability. Establishing a newOGC specification for Web Processing Services(WPS) produces new possibilities for geo webservices. It fills the gap between different datasets, because now intersecting and other ope-rations on data services can be provided in astandardised way. Using strategy patterns inthe Geo Web Service GWV, algorithmic modi-fications can easily be carried out. The Geo Web Service GWV enables integrationof domain expertise just-in-time. A rule basedreasoner controls the data sets used for calcu-lation. Rules are represented by XML-files andprocessed by SAX-events.

The implementation is based on componentsand services, provided by distributed servers.Standardised interfaces (OGC, W3C) can be

used to communicate with other services andapplications or clients. Additional functionalitytherefore can easily be used by clients whenimplemented in the business logic. The used free and open source software hasproved its applicability. It is used to reduceexpense and to alleviate transferability to otherfields of investigation. The Geo Web Service GWV is a very practicalexample of an online-calculating, easy tohandle web application. Users who take part inthe pilot operation confirm the advantages ofthe system. The automated online-calculationoffers new possibilities for data providers andclients. The system reduces the need for secon-dary data storing, especially when data arechanging fast. The possibilities of combiningdata from different providers facilitates data

Figure 7:Comparison of the computationresults for differentinput parameter(e.g. scales)

11

handling. The need of ordering or convertingdata is minimized. The transferability to otherareas of application is given. Many geoscienti-fic or engineering questions can be answeredby similar systems. Therefore it can be regar-ded as a contribution to widen the range ofexisting geo web service applications and toincrease the acceptance using web technolo-gies for geoscientific matters.

5. References Azzam, R.; Bauer, Ch.; Bogena, H.; Kappler,W.; Kiehle, Ch.; Kunkel, R.; Leppig, B.;Meiners, H.-G.; Müller, F.; Wendland, F.;Wimmer, G. (2003): Geoservice GroundwaterVulnerability - Development of an InformationInfrastructure for the Rule-based Derivation ofGeoinformation from Distributive, Heteroge-neous Geodata Inventories on Different Scaleswith an Example regarding the GroundwaterVulnerability Assessment. Geotechnologien;Information Systems in Earth Management,Kick-Off-Meeting, University of Hannover, 19.February 2003, Projects. GeotechnologienScience Report, 2, Koordinierungsbüro GEO-TECHNOLOGIEN, Potsdam, Germany, 31-35.

Bogena, H.; Kunkel, R.; Leppig, B.; Müller, F.;Wendland, F. (2004): Assessment of Ground-water Vulnerability at Different Scales. Geo-technologien; Information Systems in EarthManagement, Status Seminar RWTH AachenUniversity, 23-24 March 2004, Programme &Abstracts. Geotechnologien Science Report, 4,Koordinierungsbüro GEOTECHNOLOGIEN,Potsdam, Germany, 30-34.

Bogena, H.; Kunkel, R.; Montzka, C.; Wend-land, F. (2005): Uncertainties in the simulationof groundwater recharge at different scales.Advances in Geosciences 5, 1-6.

Chappell, D.A.; Jewell, T. (2003): Java WebServices. Köln (O’Reilly).

Evans, J.D. (Hrsg.) (2003): Web CoverageService (WCS), Version 1.0.0. OGC03-065r6.

Online:https://portal.opengeospatial.org/files/?artifact_id=3837 Harold, E.R., Means, W.S. (2004): XML in anutshell. Cologne: O’Reilly, 3rd edition.

Hölting, B.; Haertlé, T.; Hohberger, K.-H.;Nachtigall, K.-H.; Villinger, E.; Weinzierl, W.;Wrobel, J.P. (1995): Konzept zur Ermittlung derSchutzfunktion der Grundwasserüberdeckung.Geologisches Jahrbuch (63). Hannover(Schweizerbart): 7-20.

Kardasis, P., Loucopoulos, P. (2004): Expressingand organising business rules. Information andSoftware Technology 46, 701-718.

Nogueras-Iso, J.; Zarazaga-Soria, J.; Muro-Medrano, P. (2005): Geographic InformationMetadata for Spatial Data Infrastructures.Berlin (Springer).

Tomlin, C.D. (1990): Geographic informationsystems and cartographic modelling. NewJersey (Prentice Hall).

Vretanos, P. (ed.) (2002): Web Feature Service1.0. OGC, 02-058. Online: https://portal.open-geospatial.org/files/?artifact_id=7176.

Wan-Kadir, W.M.N., Loucopoulos, P. (2004):Relating evolving business rules to softwaredesign. Journal of Systems Architecture 50,367-382.

12

Overcoming Semantic Heterogeneity in SpatialData Infrastructures

1 Introduction Spatial data infrastructures (SDIs) play a majorrole for searching, accessing and integratingheterogeneous geographic data sets and geo-graphic information (GI) services. The standardsof the Open Geospatial Consortium (OGC) pro-vide a syntactical basis for data interchange bet-ween different user communities. But this is onlythe first step, as semantic heterogeneity (Bishr1998) still presents an obstacle on the waytowards full interoperability (Sheth 1999; Sond-heim et al. 1999; Egenhofer 2002). In SDIs, exi-sting standards fail to address semantic problemsthat occur due to heterogeneous data contentand heterogeneous user communities (e.g. diffe-rent languages, terminologies, and perspectives).Semantic heterogeneity occurs at differentlevels. At each of these levels, it can inhibit tasksthat are essential to the success of SDIs. - At the metadata level, semantic heteroge-

neity impedes the discovery of geographicinformation;

- at the schema level, semantic heterogeneityimpedes the retrieval of geographic informa-tion; and

- at the data content level, semantic hetero-geneity impedes the interpretation and inte-gration of geographic information.

It is the goal of the work presented in thispaper to enhance important tasks in SDIs byovercoming these semantically heterogeneityproblems. We present an ontology-basedmethodology for enhancing GI dis-covery,retrieval, interpretation and integration in SDIs,which has been developed in the meanInGsproject1. To illustrate its benefits and practical

use, we introduce two examples: - an example from the hydrology domain for

illustrating discovery, retrieval and transfor-mation, and

- an example from the geology domain forillustrating interpretation and integration.

The remainder of the paper is structured as fol-lows. Chapter 2 elaborates on the problemscaused by semantic heterogeneity on the meta-data, schema and data content levels. In chap-ter 3, we explain the notion of ontologies andintroduce the ontology architecture and langua-ge employed in both presented ap-proaches.Section 4 describes the proposed methodologyfor overcoming semantic heterogeneity at themetadata and schema levels. The approach fordealing with semantic heterogeneities particu-larly at the data level is described in section 5.We conclude the paper with a discussion ofrelated work (section 6) and a conclusion andoutlook to future work (section 7).

2 Problems Caused by Semantic Hetero-geneityThe Metadata Level In current SDIs, catalogues (OGC 2004) provi-de query functionalities based on keywordsand/or spatial filters. The metadata fields thatcan be included in the query depend on themeta-data schema used, e.g. ISO 19115(ISO/TC-211 2003), and on the query functio-nality of the service that is used for accessingthe metadata. Even though natural languageprocessing techniques can increase the seman-tic relevance of search results with respect tothe search request (e.g. Richardson & Smeaton

Lutz M. (1), Christ I. (2), Witte J. (3), Klien E. (1), Hübner S. (3)

(1) Institute for Geoinformatics (IfGI), Münster, {m.lutz|klien}@uni-muenster.de

(2) Delphi InformationsMusterManagement (DELPHI IMM), Potsdam, ingrid.christ@delphi-imm.de

(3) Center for Computing Technologies (TZI), Bremen, {witte|huebner}@tzi.de

13

1995), keyword-based techniques are inherentlyrestricted by the ambiguities of natural langua-ge. If different terminology is used by providersand requesters keyword-based search can havelow recall, i.e. not all relevant information sour-ces are discovered. If terms are homonymous orbecause of the limited ability to express complexqueries in keyword-based search, precision canalso be low, i.e. some of the discovered servicesare not relevant (Bernstein & Klein 2002).

The Schema LevelOnce an appropriate data source has beendiscovered, it can be accessed through a stan-dardized interface like a Web Feature Service(WFS) (OGC 2002). Here, semantic heterogenei-ty can cause another difficulty. While the servicecan be queried for the schema of a data source,a requester might still run into trouble when for-mulating a query filter if the property names arenot intuitively interpretable. Also, when theretrieved data are to be consumed by anotherservice (e.g. in a composite service chain) theymight have to be mapped from the providingservice’s (source) schema into the consumingservice’s (target) schema.

The Data Content LevelProblems can also occur when interpreting thecontent of data, in particular if the semanticsof values depend on some reference system(e.g. units of measure or a classificationsystem). For example, it is difficult to interpreta value correctly, if the unit of measure is notgiven. Problems can also occur when classifica-tion systems (e.g. for rock, soil, or vegetationtypes) are used. They can differ between infor-mation communities (e.g. between geologyand soil science), but also within one informa-tion community when the vocabulary used bythe information community changes over time.The resulting heterogeneities present seriousproblems when several datasets using differentclassification schemes are to be represented ina common map, interpreted by a user or com-bined for analysis. Like-wise, the data integra-

tion task within a composite service chainrequires the detection and elimination of sem-antic heterogeneities, e.g. transformations ofvalues between different units of measure.

3 Building Blocks for OvercomingSemantic HeterogeneityOntologies can be employed for making thesemantics of the information content of geo-spatial web services explicit. In this section, wedescribe the ontology architecture (section3.1), ontology language (section 3.2) and rea-soning procedures for matchmaking (section3.3) that are employed in the proposedmethodology. The notion of registration map-pings (section 3.4) is required to establish alink between a data schema and its semanticdescription, which is crucial for the tasks ofdata retrieval and schema transformation.The rule-based method for semantic media-tion (section 3.5) is required for detecting andeliminating semantic heterogeneities for thetask of data transformation.

3.1 Ontology ArchitectureThe backbone of our methodology is an infra-structure of geospatial domain and applicationontologies (Fig.1). Domain ontologies repre-sent the basic concepts and relations that areknown to all members of an information com-munity. Together they form the shared vocabu-lary of that domain.

Based on these common terms applicationontologies are derived that further constrainspecific concepts and relations and thus des-cribe a particular application, e.g. a geogra-phic dataset or the categories in a classifica-tion scheme. A user searching for data or acategory with certain properties can also usethe concepts and relations from the sharedvocabulary to specify a query. As both appli-cation ontologies and queries are based onthe same terms, they become comparable –and thus the commitment of providers andrequesters to a common shared vocabularyensures semantic interoperability.

14

3.2 Description LogicsThe ontologies shown in this paper are expres-sed using a Description Logic (DL) (Baader &Nutt 2003) notation used in the RACER system(Haarslev & Möller 2004). DL is a family ofknowledge representation languages that aresubsets of first-order logic (for a mapping fromDL to FOL, see e.g. Sattler et al. 2003). Theyprovide the basis for the Ontology WebLanguage (OWL), the proposed standard lang-uage for the Semantic Web (Antoniou & VanHarmelen 2003).

The basic syntactic building blocks of a DL areatomic concepts (unary predicates), atomicroles (binary predicates), and individuals (con-stants). The expressive power of DL languagesis restricted to a small set of constructors forbuilding complex concepts and roles. Implicitknowledge about concepts and individuals canbe inferred automatically with the help of infe-rence procedures (Baader & Nutt 2003).

A DL knowledge base consists of a TBox con-taining intensional knowledge (declarations

that describe general properties of concepts)and an ABox containing extensional knowled-ge that is specific to the individuals of thedomain. In our work, we only use TBox lang-uage features, namely- concept definition: (define-concept C D),- concept inclusion: (implies C D), and- role definition: (define-primitive-role

R :parent P :domain C :range D).The domain of a role is a concept describingthe set of all things from which this role canoriginate. This notion of the term should notbe confused with the notion »domain of inte-rest« (as in domain ontology). The range of arole is a concept describing the set of all thingsthe role can lead to. Concepts can be definedusing the following constructors:

Figure 1: The hybrid ontology approach, modified from (Wache et al. 2001)

D (and E F) (intersection)(or E F) (union)(all R C) (value restriction)(some R C) (existential quantification)(at-least | at-most | exactly n R) (number restrictions)

15

3.3 Subsumption ReasoningThere are two different types of user queries.The user can either choose an existing conceptfrom a domain or application ontology (simplequery), or she can define a concept based onthe concepts and relations in the shared voca-bulary (defined concept query). Both casesresult in a query concept that can be used inthe matchmaking.

To determine whether a concept describing adata source or category in an application onto-logy is a match for a given query concept, weuse one of the available DL inference procedu-res, computing subsumption relationships bet-ween concepts. Subsumption reasoning deter-mines whether a concept is more specific orgeneral than another one. A data source orcategory is a match for a given query, if thecorresponding concept is subsumed by, i.e.more specific than, the query concept.

For a more detailed introduction to DL langua-ges and different subsumption algorithms see(Baader & Nutt 2003).

3.4 Registration Mappings for GI RetrievalIn order to establish a mapping between adata source’s schema and its description in anapplication ontology, we have introduced regi-stration mappings (Bowers & Ludäscher 2004).An example registration mapping2 for a featu-re type representing a water level measure-ment is shown in Fig. 2. The complete feature

type is mapped to an application concept(chmi_Measurement) and its properties aremapped to a contextual path in the ontology.For example, the mapping from tok tochmi_Measurement.quantityResult.observedWaterBody.name states that the property repre-sents the name of the water body, in which themeasurement was taken.

The main idea of registration mappings is tohave separate descriptions of the applicationconcept C (called semantic type in Bowers &Ludäscher 2004) and of the structural details ofthe feature type it de-scribes (called structuraltype). This has the advantage that the semanticsof the feature type can be specified more accu-rately in application concepts if the specificationdoes not try to mirror the feature type’s structu-re. This is especially true for feature types thathave a »flat« structure that does not well reflectthe conceptual model of the domain. The pro-perty tok in figure 5, for example, represents thename of the river, where the water level measu-rement was taken.

3.5 Semantic Mediation for GI ServiceCompositionThe goal of the matchmaking approach descri-bed in section 3.3 is to identify data sources thatexactly match the semantics required by therequester. When matchmaking is done in thecontext of composing a complex service chain,these requirements are also determined by theapplication semantics and schema of an existing

Figure 2: An example registration mapping for the GML documentshown on top

16

service (which is to be combined with the servi-ce to be discovered). In this case, mediation bet-ween the schema of the discovered service andthat of the target service can become necessary.

Our approach to mediation is based on theidea of semantic mediation introduced inWache (2003). When compared to approachesfrom the area of schema integration (for anoverview, see Conrad 2002), semantic media-tion especially focuses on semantic heteroge-neities. The specification of the integrationmapping on the semantic level is identified asthe most difficult task when developing amediator for integrating heterogeneous infor-mation sources. As information on the seman-tics of an information source is often only avai-lable inherently inside the information systemsmanaging the information, an explicit seman-tic description of information sources is intro-duced in addition to syntactic descriptions.This semantic description consists of two parts:a description of meaning and a description ofcontext (Fig. 3).

The meaning is defined unanimously across alldata sources in a domain e.g. by a conceptWaterLevel. It can be used to identify semanti-cally equivalent information or semantic hetero-geneities within one domain (Wache 2003). Thecontext describes the different representationsof semantically equivalent information in diffe-rent data sources, e.g. the unit or the scale ofWaterLevel. It can be used to identify and solve

semantic heterogeneity problems between spe-cific data sources within one domain. The specification of the integration mapping isbased on a rule-based approach. The mappingconsists of transformation rules, which are divi-ded into rules for solving structural and sem-antic heterogeneities. Accordingly, two kindsof transformation rules can be distinguished: - Query decomposition rules are based on the

»global-as-view« principle (Levy 1999; Halevy2001) and enable the splitting of a queryagainst a global (target) schema into severalsubqueries against the respective informationsources. They are used to solve conflicts cau-sed by structural heterogeneities.

- Context transformation rules specify how apiece of information can be transformedfrom one context into another. They areused to solve conflicts caused by semanticheterogeneities.

4 Enhancing GI Discovery, Retrieval andIntegrationThis sections illustrates how semantic hetero-geneity can be overcome to enhance thediscovery, retrieval and integration of geogra-phic information. We first introduce a scenarioin section 4.1, which we use for illustration. Insection 4.2, an approach for ontology-basedGI discovery and retrieval is introduced. Sec-tion 4.3 describes an approach for semantictranslation, and section 4.4 introduces the SDIarchitecture for implementing the presentedapproaches.

4.1 Hydrology Scenario In this section, we use the following scenariofor illustration. A service developer wants toimplement a service chain (ISO 2005) that pro-vides fast and up-to-date access to water levelmeasurements in a certain river, interpolatesthese measurements along the river courseand visualizes the interpolation results. Such aservice chain could for example enable thedetection of hazard areas during flood events.The central component of this service chain isthe service that interpolates the water level

Figure 3: Semantics as a composition of meaning andcontext

17

measurements. In order to execute this servicein an open and distributed environment thefollowing steps are necessary: (1) appropriateinput data (provided by WFS (OGC 2002)) haveto be discovered, (2) the input data have to beretrieved using given constraints, and (3) theretrieved data have to be transformed to fit there-quirements of the interpolation service.

In current SDI architectures the user will facethe problems described in section 2. On themetadata level, keyword-based search in acatalogue will result in (1) low recall, e.g. if theuser searches for »water level measurement«whereas the service is described with »tidescale«, and (2) low precision, e.g. if the usersearches for the water level in rivers whereasthe services provides water level in groundwa-ter, and both use »water level measurement«in their descriptions. On the schema level, itmight be difficult for a user to correctly inter-pret the meaning of the discovered featuretype’s property names, e.g. if a water levelmeasurement is called »height«. This makesthe formulation of the WFS GetFeature requesta difficult task. Finally, on the data contentlevel, the results returned by the WFS may beincorrectly interpreted by the consuming inter-polation service due to missing information onthe provided data. If, e.g., the interpolationservice expects water level measurements inmeters and the WFS provides water level mea-sure-ments in centimeters, this will lead towrong interpolation results.

4.2 Ontology-Based Discovery and Retrieval ofGeographic DataTo address the problems on the metadata andschema levels, we have proposed an appro-ach that enhances the discovery of geogra-phic data in SDIs by providing ontology-baseddescriptions (section 3.1) and integrates thediscovery and retrieval processes to make theoverall task of GI retrieval more user-friendly(Lutz & Klien 2006).

Rather than having to formulate separate que-ries for discovering data and retrieving it, therequester only has to formulate one query forthe data she is interested in. This query isbased on terms from existing domain ontolo-gies and is automatically translated into one orseveral DL concepts. These concepts are usedas query concepts for discovering semanticallyappropriate feature types. The matchmakingbetween the query concept and the applica-tion concepts describing feature types is basedon subsumption reasoning and follows theapproach described in section 3.3.

After the requester has selected one of thediscovered feature types, a GetFeature query,which uses the property names of the featuretype's application schema, is constructed fromthe user query. This step requires registrationmappings (section 3.4) between the featuretype's properties and the roles from thedomain ontology. As a final step, the derivedWFS query is then executed and its results arereturned to the user. In the following, we illu-strate the translation of a requester's query (1)into a DL query concept (2) and subsequentlyinto a GetFeature request (3) based on the sce-nario described above.(1) An example query statement for finding

water level measurements for the Elbe Riverprovided in centimeters for a given date(2004-04-22) and location is presented inFig. 4. This query is based on concepts andrelations from three domain ontologies (forthe domains of MEASUREMENTS, HYDRO-LOGY and GEOGRAPHIC FEATURES). Notethat the constraints are either type restric-tions or comparisons with a value specifiedby the requester (value constraints). Valueconstraints can only be defined for roleswhose range is an XSD datatype or a GMLgeometry type. In addition to commonstring and number comparators (such as ≥or startsWith), spatial comparators such aswithinBoundingBox, intersects or within-distance-of can be used.

18

(2) This query can be translated following theguidelines in Lutz & Klien (2006) into thefollowing DL query concept. Type cons-traints are expressed through universallyquantified value restrictions in DL. Valueconstraints only become relevant when thedata are retrieved from the WFS. In order tobe able to express these constraints as a fil-ter expression in the GetFeature query, it isimportant that the feature type containsthe property to be constrained. Therefore,in the discovery phase, value constraints areexpressed as existential quantification onthe specified roles.

This query concept is then used for discoveringappropriate data sources based on DL sub-sumption reasoning (section 3.3).(3) In the next step we want to retrieve the

requested information from the discoveredWFS. This requires formulating a GetFea-ture request including a filter expression. Inorder to do this, the structure of the WFS’sfeature type has to be known. Also, theproperty names of the selected feature typethat are equivalent to the domain ontologyterms used in the query statement have tobe derived. All the required information canbe accessed from the feature type’s regi-stration mapping. By using the registrationmapping shown in Fig. 2 (p. 4), the exam-ple query statement shown in Fig. 4 can betranslated into the WFS GetFeature request(OGC 2002b) and filter expression (OGC2001) shown in Fig. 5.

4.3 Semantic Translation of Geographic DataWhen considering a service chain composed ofseveral separate services (as in the presentedexample), an additional step might be requiredafter discovering and retrieving appropriatedata. If the structure and semantics of the dataprovided by one service (e.g. the WFS providingthe water level measurements) do not matchexactly those required by the consuming service(e.g. the interpolation service) a translationbecomes necessary.

Figure 5: A WFS query thatrequests the property stav from afeature type called StavVody. The fil-ter expression constrains the queryto features whose tok propertyequals »Elbe«, whose datum proper-ty equals »2004-04-22« and whoseposition property is within the speci-fied bounding box

Figure 4: Example for a semantic query statement. Thekeywords of the proposed syntax are shown in capitals,the comparators in italics

19

In general, translations connect one or moredata sources to a destination with the help ofappropriate conversion rules. The challenge of asemantic translation is not to process the trans-lation, but to discover and to specify the trans-lation, in our case of the GML documents retur-ned by a WFS.

For the discovery, the semantics and the struc-ture of the source(s) (i.e. the WFS) as well as ofthe target (i.e. the interpolation service) have tobe examined. Meaning and context can beextracted for each feature type property fromthe registration mappings and application onto-logies of the source and target services. Basedon the meaning, semantic correspondences canbe identified between the different attributesfor each feature type based on subsumptionreasoning (step 2 in section 4.2). After matchingconcepts have been discovered, appropriatemappings between the corresponding proper-ties of the feature types are generated automa-tically for a defined domain. These mappingsare combined for different feature types asquery transformation rules. The rules (section3.5) contain the information, how queries can

be applied to one schema of a feature type, spe-cified by a corresponding description in the rulehead, and can be transformed into equivalentqueries against other feature types, specified inthe rule body. Thus, for example, a query withthe schema of the feature type required by theinterpolation service can be transformed on pro-perty level into a request for a feature typeStavVody as is shown in Fig. 6. The rule decom-poses a query against a feature type calledMeasurement from the interpolation serviceinto a feature type called StavVody (as instancesof the meaning Measurement). Additionally, itdefines correspondences between different pro-perties that are semantically equivalent, e.g.value from the feature type Measurement andstav from the feature type StavVody.

After this, we have to take a look at the contextof the semantic description. In our scenario, theinterpolation service requires water levels inmeters (see context description in Fig. 6) where-as the discovered WFS provides water levels incentimeters. Therefore the measurement valueshave to be converted. For this conversion, alibrary of predefined context transformation

Figure 6: A query decomposition rule that decomposes a query against a feature type cal-led Measurement from the WLIS into a feature type called StavVody.

20

rules is used. These rules specify how to convertbetween different contexts. As an example, Fig.7 shows a rule for transforming measurementvalues from centimeters to meters.

To build query plans containing detailed sequen-ces of query transformation rules and contexttransformation rules we are using logical infe-rence based on a representation of the featuretypes and the rules in Horn Logic.

4.4 SDI ArchitectureIn order to use the methods for enhancing disco-very, retrieval and transformation in SDIs theyhave to be encapsulated in software compo-nents. In this section, we introduce an architectu-re that includes such components in addition toexisting SDI components and describe the flow ofinformation between them. Fig. 8 depicts a sim-plified view of the architecture including the tasksfulfilled by each of the components.

The central component of the architecture is aclient that manages the overall workflow (WSClient). In our current architecture, this service istailored to its specific application, i.e. to executethe interpolation service chain. In a future versionof the architecture, this client could be substitu-ted by a generic workflow client that executes acustomised description of the service chain to beexecuted, e.g. using the Business Process Exe-cution Language (BPEL) (Andrews et al. 2003).

The entry point for the ontology-based discove-ry (step 1 in Fig. 8) and retrieval (step 4) is thequery client. This client provides a user interfacefor query formulation. It is built on the querylanguage described in section 4.2, thus enablingusers to pose formal and precise queries basedon existing domain ontologies (Fig. 9).

Figure 7: Example for a simple context transformationrule for the conversion between centimeter and meter(domain-independent)

Figure 8: Simplified view on the service architecture

<?name1,

(?meaning)::[meas:unitOfMeasure->> meas:Centimeter],

?datatype1,

?value1> @?source1

-->

<?name2,

(?meaning)::[meas:unitOfMeasure ->> meas:Meter],

?datatype2,

?value2> @?source2

:-

?value1 is ?value2 / 100.

21

After a user query has been submitted, it istranslated into a DL query concept (section 3.3).When submitting a query, the user might choo-se whether the transformation rules offered bythe Semantic Translation Specification Service(STSS, see below) should be taken into account(step 2). If, for example, a transformation rulebetween the units meter and centimeter exists,it makes sense to search for feature types thatoffer water level measurements in both units,thus extending the potential result set.

In our scenario, the STSS offers unit transfor-mations for length measurements (e.g. fromcentimeters to meters) and the query conceptis relaxed accordingly, i.e. the range restrictionfor the unitOfMeasure role is relaxed fromCentimeter to the disjunction of unit conceptsthat are transformable into centimetres: (allunitOfMeasure (or Centimeter Meter Millime-ter Yard Inch … )). Note that now it is no lon-ger guaranteed that the discovered featuretypes exactly match the requirements of theinterpolation service. However, it is guaranteedthat all feature types can be transformed in thecorrect schema by the STSS.

The query concept is sent to the SemanticCatalog Service to discover WFSs that provide

semantically appropriate feature types (section3.3). In order to access the discovered featuretype through its WFS interface, the query clientthen automatically constructs a GetFeaturerequest from the user’s query using the pro-perty names of the feature type’s applicationschema (section 4.2).

The query concept is sent to the Ontology-Based Reasoner (OBR), which stores the regi-stered ontologies and provides the reasoningfunctionality (step 4). Based on subsumptionreasoning the OBR discovers semanticallyappropriate feature types and returns theirmetadata to the query client. (section 3.3). Inorder to access the discovered feature typethrough its WFS interface, the query clientthen automatically constructs a GetFeaturerequest from the user’s query using the pro-perty names of the feature type’s applicationschema (section 4.2 (3)).

For the discovery and specification of thetransformation between a source and targetschema the Semantic Translation SpecificationService (STSS) has been developed. The inputparameters of the STSS include the results ofthe ontology-based discovery and retrieval, i.e.the application ontologies, registration map-

Figure 9: User interface of the query client.

22

pings, and GetFeature requests of the sourceservices (the discovered WFSs) and the corres-ponding information for the target service (theinterpolation service). The STSS determines thesemantic correlations, selects the needed con-text transformation rules, adds changes to thestructure and specifies a transformation, whichis noted in XSLT (step 5). The actual transfor-mation is performed by the TransformationService (TS), which simply executes the XSLTrules and returns a GML document with thetransformed water level measurements inmeters (step 6).

Now, the interpolation service correctly inter-prets all the data provided by the WFSs and theresults of the interpolation (step 7) can bedisplayed in a WMS (step 8).

5 Enhancing GI Interpretation andIntegrationIn this section, we address the problem of sem-antic heterogeneity on the data content levelfrom a different angle. In section 5.1, we intro-duce another scenario (from the geologydomain), which differs from the one introdu-ced in the previous section in addressing theproblem of different classification systems wit-hin one user community. In section 5.2, wepresent how ontology-based methods can beapplied to support the tasks of interpretatingand integrating geographic information.

5.1 Geology ScenarioSemantic heterogeneity problems at the datacontent level occur not only between differentinformation communities (e.g. from thedomains of soil science and geology) but alsowithin the same information community at dif-ferent times. The latter applies for example tothe Geology Survey of Saxony-Anhalt wheredifferent authors of geological maps have useddifferent classifications at different times inhistory. This leads to the problem of synony-mous and homonymous stratigraphic termswithin the geological database. There arenumerous questions concerning geological

data integration, like »Where are good conditions for groundwater formations?« or»Where is the geological subsoil suitable for adump site?«. Obviously, this makes the crea-tion of a common map presentation or dataanalysis a difficult task (if done in a conventio-nal way). We use the geology scenario to illu-strate how these problems are tackled usingontology-based data integration methods.

5.2 Ontology-based Data Integration In our approach, the meaning of the stratigra-phic terms used in the different classificationsystems is made explicit by means of DL onto-logies. As in the approach for GI discovery andretrieval, the use of ontologies should be trans-parent for the user. Therefore, we have imple-mented a function for translating user queriesinto DL query concepts, which can subse-quently be used by an inference engine to dothe actual matchmaking.

The ontologies were modelled according tothe hybrid approach as depicted in Fig. 1. Ashared vocabulary for a subdomain of geologywas created and, based on this, more specificapplication ontologies were defined for eachstratigraphic concept from the Rogenstein-Zone of the Lower Buntsandstein3 (like »su4«)according to different classification schemeslike that of Jung or Fulda & Huelsemann.

The graphical user interface (Fig. 10) allows twokinds of queries: A concept-based query provi-ding pre-defined stratigraphic concepts fromexisting classification schemes and a more com-plex user-defined query based on petrographiccharacteristics. The latter allows, for example,the search for rocks offering a good protectionagainst ground water pollution. Translated inpetrographic characteristics this means it mustbe a solid rock or a soil based on clay or silt. Thisquery would be translated into the following DLquery concept:

23

The query is sent to the Ontology-based reaso-ner (OBR), which performs the matchmakingbetween the query concept and the conceptsrepresenting stratigraphic terms in one of theavailable classification schemes. The matchingstratigraphic concepts in different classificationschemes are shown in Fig. 11.

These concepts are then used to generate afilter for the GetFeature request to the WFSthat provides a standardized access to thegeological database. The retrieved featuresare displayed by a WMS. As a result of theuser-defined query, Fig. 12 shows the suc-cessful data integration in two neighbouring

map sheets in Saxony-Anhalt that use diffe-rent classification schemes.

The ontology-based approach presented in thissection has a tremendous benefit compared tothe conventional ones: Users can either useterms from a classification scheme they are alre-ady familiar with and still find features using adifferent (unknown) classification scheme.Furthermore, they can create their own searchterms based on the petrographic characteri-stics available in the shared vocabulary andfind the appropriate features irrespective ofthe classification scheme used by it. This is acrucial feature for using geological information

Figure 10: User Interface for the geology scenario

(define-concept query (and

Gestein

(or

(and

(all hat-konsistenz Fest)

(exactly 1 hat-hauptbestandteil))

(and

(all hat-konsistenz Locker)

(exactly 1 hat-hauptbestandteil)

(or

(at-least 1 hat-hauptbestandteil Ton)

(at-least 1 hat-hauptbestandteil Schluff)))))

)

24

Figure 11: Results (extract) of the user-defined query for rocks with a good protection againstgroundwater pollution

25

as a basis for engineering decisions, for exam-ple the building of a dump site.

6 Discussion and Related WorkThe approach presented in this paper is rela-ted to previous work in the fields of geogra-phic information science, information disco-very and retrieval, data integration and artifi-cial intelligence.

A first step towards overcoming semanticheterogeneity in the geospatial domain hasbeen the proposal of Integrated GeographicInformation Systems (IGIS), i.e. systems thatintegrate diverse GIS technologies or reflect aparticular point of view of a community(Hinton 1996). This idea has been advanced inFonseca et al. (2002a) and Fonseca et al.(2002b) by introducing ontologies as meansfor supporting representations of incompleteinformation, multiple representations of geo-graphical space, and different levels of detail. InSDIs, where geographic information is usuallyhighly distributed and heterogeneous, solvingheterogeneity problems becomes a prerequisi-te. One focus of the research presented here isto transfer the ontology approach for dealingwith semantic heterogeneity to the SDI domainand to demonstrate how it can be integratedinto existing standards-based architectures.

Work in the field of information discovery andretrieval is manifold. There is widespread agree-ment among researchers in this field that decla-rative content descriptions and query capabili-ties are necessary (Mena et al. 1998; Czerwinskiet al. 1999; Guarino et al. 1999; Heflin &Hendler 2000). The vision of most research inthis domain is that users should be able toexpress what they want, and the system shouldfind the relevant sources and obtain the answer(Levy et al. 1996). As this might involve combi-ning data from multiple sources, informationdiscovery and retrieval is closely related to dataintegration, whose goal it is to provide a uni-form interface (through a global schema) to amultitude of data sources (each with a localschema). In data integration terminology (Levy2000), our approach can be considered as a»Local As View« approach. This means thatthe contents of a data source are described asa query over the mediated schema, which inour case is substituted by the ontology (seeMädche et al. (2001); Guha et al. (2003) forother examples, where ontologies are used insearch and retrieval mechanisms). Usually, aquery through the mediated schema in thisapproach requires complex transformationrules. With his semantic mediator, Wache(2003) suggests a way to generate these rulesfrom fully annotated data sources semi-auto-matically with the help of assistants. The assi-

Figure 12: Graphical Result for rock types offering a goodprotection against ground water pollution

26

stants attempt to find inter-correspondencesbetween data elements of query and sources. Inour SDI-scenario, it is often not possible to askthe user for confirmation, especially not on low-level relations between sources the user is notfamiliar with. We circumvent this problem withthe specialisation to a specific domain withexplicitly pre-modelled information and rela-tions, e.g., transformation rules between units(see chapter 3.5).

In the BUSTER (Bremen University SemanticTranslator for Enhanced Retrieval) project(Vögele et al. 2003), DL descriptions have beenused to describe and query classifications (Visser& Stuckenschmidt 2002) and data content(Hübner et al. 2004; Vögele & Spittel 2004).However, these approaches use simple ontolo-gies, and queries only have limited expressivity.They show how well established cataloguesystems for electronic devices, namely ETIM andecl@ss (Visser et al. 2002a), or landuse classifi-cation, ATKIS and CORINE Landcover (Visser etal. 2002b), can be used as the grounding sha-red vocabularies for semantic translation.However, as these classification schemes areoften imprecise, miss details and contain hardlyunderstandable verbal circumscriptions andeven inconsistencies, a lot of adjustments wereneeded in order to transform them into ontolo-gical descriptions.

Providing an ontology-based query interfacethat enables uniform access to heterogeneousdata sources and supports the user in formula-ting a precise query has also been proposed bythe SEWASIE4 project (Dongilli et al. 2004),which employs the same ontology and match-making approach for information retrieval.Moreover, the SEWASIE query interface enablesan iterative refinement process of the query andutilizes natural language as query representa-tion. While this certainly represents a user-friendly approach, it also additionally requiresthat the ontology engineer provides verbaliza-tions for each ontology term. In contrast, wepropose an intuitive but still formal query lang-uage. And whereas the SEWASIE query interfa-ce is developed for the needs of the Semantic

Web in general, we are focused on geospatialinformation infrastructures.

A strategy based on semantics for supportingthe discovery and integration of datasets andservices is used in the Science Environmentfor Ecological Knowledge (SEEK) project(Pennington et al. 2004). We have benefitedfrom the work conducted in SEEK by adap-ting the method of registration mapping(Bowers & Ludäscher 2004) for our purposes.The key difference of our approach lies in thecombination of both tasks, the informationdiscovery and retrieval, and to hide the com-plexity from the user.

Integrating information from different usercommunities based on ontologies with the goalof displaying a single map (with a single wellunderstood legend) has also been the focus ofother research projects. In the GEON project,geological data from different US states havebeen combined according to a simple sharedvocabulary comprising geological age, composi-tion, fabrics, texture and genesis (Lin &Ludäscher 2003). Rather than using a hybridontology approach as proposed in this paper,the authors propose to use explicit mappingsbetween different ontologies. Also, the ontolo-gies used are simple is-a hierarchies and do notcontain roles. In the HarmonISA project , landu-se data from the border region of Austria,Slovenia and Italy have been combined in a sin-gle landcover map. In this project, a compre-hensive shared vocabulary for defining land useclasses has been developed based on the diffe-rent national landuse classification systems. Incontrast to the approach presented in thispaper, where all areas matching certain require-ments were searched, the HarmonISA project5

aimed at producing a landuse map that com-pletely covers the area. Therefore, the authorsused a complex similarity measurement bet-ween landuse type definitions rather than sub-sumption reasoning.

27

7 Conclusions and Future WorkProblems caused by semantic heterogeneitiescan occur on different levels in SDIs. In thispaper, we have illustrated how these problemscan be overcome by using DL ontologies andreasoning. We have also shown how the propo-sed methodology can be encapsulated in servi-ces and clients and how these can be combinedwith existing SDI components. The two usecases illustrate how these intelligent serviceseffectively support the semantic query, retrieval,translation and integration of geographic data.Moreover, we have shown that the approachalso supports dynamic service chaining in orderto answer complex queries.

In our future work we will address the followingissues:- Extensions of the tested scenarios. The tested

scenario comprises requests and applicationschemas with a relatively simple structure.Also, the effects of the scale of a data sourcehave not been taken into account. Futuretests of the approach will include more com-plex request possibilities (like support for spa-tial comparators and nested queries) and datasources at different scales. Also, the effective-ness of the approach will be tested in a moregeneric setting with complex applicationschemas and examples from other domains.

- Semantics of geoprocessing services. In sce-narios where a service chain is required toanswer a complex question, the semanticsnot only of the data but also of the servicesfor processing the data are of vital importan-ce. In our future work, we will therefore inve-stigate approaches for the semantic descrip-tion and discovery of geoprocessing servicesand examine how these can be combinedwith the presented approach for the discove-ry and retrieval of geographic data. For firststeps in this direction, see Lutz (2005a,b).

- Template Service Chains. In recent years,many researchers have addressed the auto-mated generation of complex service chainsbased on user queries (e.g. Burstein et al.2005). However, these approaches still face

many problems of complexity. A simpler solu-tion for supporting the creation of complexservice chains by the user could be basedon providing generic templates for servicechains that solve a particular type of task.Such a template should be a fixed combi-nation of several generic service types, eachof which performs a subtask of the overallfunctionality. In an iterative process, reque-sters could subsequently instantiate thesetemplates with services discovered for eachof these subtasks.

- User-friendly generation of applicationontologies. While our approach hides muchof the complexity of the ontology-based GI retrieval from the requester, the dataprovider still has to create and register rat-her complex application ontologies. We are aware that this is one of the crucialbottlenecks for our approach to be accep-ted and used in future SDIs. Future workwill therefore address how the process of creating formal descriptions of the geo-data could be automated. First ideas onhow this can be achieved using spatial ana-lyses of geographic datasets are presentedin Klien & Lutz (2005).

ReferencesAndrews, T., Curbera, F., Dholakia, H., Goland,Y., Klein, J., Leymann, F., Liu, K., Roller, D.,Smith, D., Thatte, S., Trickovic, I. & Weerawara-na, S. (2003): Business Process ExecutionLanguage for Web Services, Version 1.1, BEASystems, IBM, Microsoft, SAP,Siebel Systems.

Antoniou, G. & Van Harmelen, F. (2003): WebOntology Language: OWL, in: Staab, S. & R.Studer (ed.): Handbook on Ontologies,Springer: 67-92.

Baader, F. & Nutt, W. (2003): Basic DescriptionLogics, in: Baader, F., D. Calvanese, D.McGuinness, D. Nardi & P. Patel-Schneider (ed.):The Description Logic Handbook. Theory,Implementation and Applications, Cambridge,Cambridge University Press: 43-95.

28

Bernstein, A. & Klein, M. (2002): TowardsHigh-Precision Service Retrieval, in: Horrocks,I. & J. Hendler (ed.): The Semantic Web - FirstInternational Semantic Web Conference(ISWC 2002): 84-101.

Bishr, Y. (1998): Overcoming the Semanticand Other Barriers to GIS Interoperability, In-ternational Journal of Geographical Informa-tion Science 12 (4): 299-314.

Bowers, S. & Ludäscher, B. (2004): An Onto-logy-Driven Framework for Data Transfor-mation in Scientific Workflows, InternationalWorkshop on Data Integration in the LifeSciences (DILS'04).

Burstein, M., Bussler, C., Pistore, M. & Ro-man, D. [ed.] (2005): Proceedings of theWorkshop on WWW Service Compositionwith Semantic Web Services 2005 (wscomps-05), University of Technology of Compiegne,France.

Conrad, S. (2002): Schemaintegration –Integrationskonflikte, Lösungsansätze, aktu-elle Herausforderungen, Informatik – For-schung & Entwicklung 17 (3): 101-111.

Czerwinski, S., Zhao, B. Y. & Hodes, T. (1999):An architecture for a secure service discoveryservice, Fifth ACM/IEEE International Con-ference on Mobile Computing and Networ-king: 24-35.

Dongilli, P., Franconi, E. & Tessaris, S. (2004):Semantics driven support for query formula-tion, in: Haarslev, V. & R. Möller (ed.): Inter-national Workshop on Description Logics(CEUR Workshop Proceedings).

Egenhofer, M. (2002): Toward the SemanticGeospatial Web, The 10th ACM InternationalSymposium on Advances in GeographicInformation Systems (ACM-GIS).

Fonseca, F., Egenhofer, M., Agouris, P. &Camara, G. (2002a): Using Ontologies forIntegrated Geographic Information Systems,Transactions in GIS 6 (3).

Fonseca, F., Egenhofer, M., Davis, C. &Câmara, G. (2002b): Semantic Granularity inOntology-Driven Geographic InformationSystems, Annals of Mathematics and ArtificialIntelligence 36 (1-2): 121-151.

Guarino, N., Masolo, C. & Vetere, G. (1999):OntoSeek: Content-Based Access to the Web,IEEE Intelli-gent Systems 14 (3): 70-80.

Guha, R., McCool, R. & Miller, E. (2003):Semantic Search, 12th International Con-ference on World Wide Web: 700-709.

Haarslev, V. & Möller, R. (2004): RACER User’sGuide and Reference Manual. Version 1.7.19,URL: http://www.cs.concordia.ca/~haarslev/racer/racer-manual-1-7-19.pdf. Last accessed:July 29, 2005

Halevy, A. Y. (2001): Answering Queries UsingViews: A Survey, Very Large Data Bases 10 (4):270-294.

Heflin, J. & Hendler, J. (2000): Searching theWeb with SHOE, Papers from the AAAIWorkshop (In Artificial Intelligence for WebSearch): 35-40.

Hinton, J. (1996): GIS and Remote SensingIntegration for Environmental Applications,International Journal of Geographical Infor-mation Science 10 (877-890.

Hübner, S., Spittel, R., Visser, U. & Vögele, T.(2004): Ontology-Based Search for Inter-active Digital Maps, IEEE Intelligent Systems19 (3): 80-86.

ISO (2005): Geographic Information - Ser-vices. ISO 19119, International Organizationfor Standardization.

29

ISO/TC-211 (2003): Text for FDIS 19115Geogaphic information - Metadata. FinalDraft Version, International Organization forStandardization.

Klien, E. & Lutz, M. (2005): The Role ofSpatial Relations in Automating the SemanticAnnotation of Geodata, Conference on Spa-tial Information Theory (COSIT 2005).

Levy, A. Y. (1999): Combining ArtificialIntelligence and Databases for Data Integra-tion, in: Wooldridge, M. & M. M. Veloso (ed.):Artificial Intelligence Today: Recent Trendsand Developments (LNCS 1600), Berlin,Springer: 249-268.

Levy, A. Y. (2000): Logic-Based Techniques inData Integration, in: Minker, J. (ed.): LogicBased Artificial Intelligence, Dordrecht, NL,Kluwer: 575-595.

Levy, A. Y., Rajaraman, A. & Ordille, J. (1996):Querying heterogeneous information sourcesusing source descriptions, 22nd VLDB Con-ference: 251-262.

Lin, K. & Ludäscher, B. (2003): A System forSemantic Integration of Geologic Maps viaOntologies, Semantic Web Technologies forSearching and Retrieving Scientific Data(SCISW).

Lutz, M. (2005a): Ontology-based Descrip-tions for Semantic Discovery and Composi-tion of Geoprocessing Services, GeoInfor-matica (in press).

Lutz, M. (2005b): Ontology-Based ServiceDiscovery in Spatial Data Infrastructures, in:Jones, C. & R. Purves (ed.): ACM Workshopon Geographic Information Retrieval (GIR'05).

Lutz, M. & Klien, E. (2006): Ontology-basedRetrieval of Geographic Information, Inter-national Journal of Geographical InformationScience(forthcoming).

Mädche, A., Staab, S., Stojanovic, N., Studer,R. & Sure, Y. (2001): SEAL - A Framework forDeveloping SEmantic portALs, 18th BritishNational Conference on Databases (LectureNotes in Computer Science): 1-22.

Mena, F., Kashyap, V., Illarramendi, A. &Sheth, A. (1998): Domain Specific Ontologiesfor Semantic Information Brokering on theGlobal Information Infrastructure, First Inter-national Conference on Formal Ontologies inInformation Systems.

OGC (2002): Web Feature Service Implemen-tation Specification, Version 1.0.0, Open GISConsortium.

OGC (2004): Catalogue Services Specifi-cation, Version 2.0 (OGC ImplementationSpecification 04-021r2), Open GeospatialConsortium.

Pennington, D., Michener, W. K., Berkley, C.,Higgins, D., Jones, M. B., Schildhauer, M.,Bowers, S., Ludäscher, B. & Rajasekar, A.(2004): Building SEEK: The Science Environ-ment for Ecological Knowledge (SEEK): ADistributed, Ontology-Driven Environment forEcological Modeling and Analysis (Abstract),in: Egenhofer, M., C. Freksa & H. Miller (ed.):The Third Conference of Geographic Infor-mation Science (GIScience 2004).

Richardson, R. & Smeaton, A. F. (1995): UsingWordNet in a Knowledge-based Approach toInformation Retrieval (Technical Report CA-0395), Dublin City University.

Sattler, U., Calvanese, D. & Molitor, R. (2003):Relationships with other Formalisms, in:Baader, F., D. Calvanese, D. McGuinness, D.Nardi & P. Patel-Schneider (ed.): The Des-cription Logic Handbook. Theory, Implemen-tation and Applications, Cambridge, Cam-bridge University Press: 142-183.

30

Sheth, A. P. (1999): Changing Focus onInteroperability in Information Systems: FromSystem, Syntax, Structure to Semantics, in:Goodchild, M. F., M. Egenhofer, R. Fegeas &C. A. Kottman (ed.): Interoperating Geo-graphic Information Systems, Dordrecht, NL,Kluwer: 5-30.

Sondheim, M., Gardels, K. & Buehler, K. [ed.](1999): GIS Interoperability, New York, JohnWiley & Sons (Geographic Information Sys-tems 1, Pronciples and Technical Issues).

Visser, U. & Stuckenschmidt, H. (2002): Inter-operability in GIS - Enabling Technologies, in:Ruiz, M., M. Gould & J. Ramon (ed.): 5thAGILE Conference on Geographic Informa-tion Science: 291-297.

Visser, U., Stuckenschmidt, H., Schlieder, C.,Wache, H. & Timm, I. (2002a): TerminologyIntegration for the Management of distribu-ted Information Resources, Künstliche Intelli-genz 16 (1): 31-34.

Visser, U., Vögele, T. & Schlieder, C. (2002b):Spatio-Terminological Information Retrievalusing the BUSTER System, in: Pillmann, W. &K. Tochtermann (ed.): Environmental Com-munication in the Information Society, 16thConference on Informatics for EnvironmentalProtection (EnviroInfo): 93-100.

Vögele, T., Hübner, S. & Schuster, G. (2003):BUSTER - An Information Broker for the Se-mantic Web, KI - Künstliche Intelligenz 03 (3):31-34.

Vögele, T. & Spittel, R. (2004): EnhancingSpatial Data Infrastructures with SemanticWeb Technologies, 7th Conference on Geo-graphic Information Science (AGILE 2004).

Wache, H. (2003): Semantische Mediation fürheterogene Informationsquellen, Berlin, Aka-demische Verlagsgesellschaft.

Wache, H., Vögele, T., Visser, U., Stucken-schmidt, H., Schuster, G., Neumann, H. &Hübner, S. (2001): Ontology-Based Inte-gration of Information – A Survey of ExistingApproaches, IJCAI-01 Work-shop: Ontologiesand Information Sharing: 108-117.

1 see http://www.meanings.de/

2 taken from the hydrology example describedin section 4.1.

3 Oolitic Limestone

4 Semantic Webs and Agents in IntegratedEconomies, see http://www.sewasie.org/

5 see http://harmonisa.uni-klu.ac.at

31

32

Advancement of Mobile Geoservices:Potential and Experiences

1. IntroductionMobile information technology opens newperspectives and dimensions for the geoscien-ces, by providing experts in governmental andnon-governmental authorities, industry andscience with ubiquitous access to geoscientificinformation. With this new instrument thedigital acquisition, management, visualization,and analysis of geodata needed for the under-standing of geoscientific processes and naturaldisasters can be supported directly in the field.

The number of applications is increasing wheregeoinformation systems (GIS) have to coopera-te with distributed mobile applications andwith suitable geodatabase managementsystems (Balovnev, Bode, Breunig, Cremers,Müller, Pogodaev et al., 2004). The currentparadigm shift from the development ofmonolithic GIS to flexible and mobile accessi-ble geoservices can be recognized in manyapplication fields. New geoservices will provideubiquitous access to geodata needed in appli-cations such as environmental monitoring anddisaster management. Client applications com-municating with geoservices have to efficientlyacquire, visualize and manage application-spe-cific 2D and 3D objects and complex spatio-temporal models (Breunig, Cremers, Shumilov& Siebeck, 2003).

In this contribution a geoscientific case studydealing with the analysis of land slides showsthe potential behind mobile geoservices.Contributions to a distributed software system(Breunig, Malaka, Reinhardt & Wiesel, 2003)consisting of geoservices used by on-siteclients for geodata acquisition, viewing, aug-mented reality, and geodata management arepresented. The clients communicate over net-work with geodatabase services. Experiencesare reported and finally, conclusions and ashort outlook are given which address furtherresearch in the field of mobile geoservices.

2. Objectives of the projectThe concrete problem we are referring to inthis project is the analysis of land slides at anarea near Balingen in south-west Germany(Ruch, 2002). Since several years there are acti-ve creeping movements of the terrain, whichmay endanger the traffic and people using anearby road. The geodetic measurementsshow a gradual sinking of the soil and rocks. Aforecast for a slowing down or speeding up ofthe movements cannot be given. However,mobile data acquisition of the ongoing move-ments and remote data access to a central sta-tion help to watch the situation. The move-ment measurements are done by extensome-

Breunig M. (1), Bär W. (1), Thomsen A. (1), Häußler J. (2), Kipfer A. (2), Kandawasvika A. (3), Mäs S. (3),

Reinhardt W. (3), Wang F. (3), Brand S. (4), Staub G. (4), Wiesel J. (4)

(1) Research Centre for Geoinformatics and Remote Sensing, University of Osnabrück, Kolpingstr. 7,

49069 Osnabrück, Germany; E-Mail: Martin.Breunig@uni-osnabrueck.de

(2) European Media Laboratory GmbH, Villa Bosch , Schloss-Wolfsbrunnenweg 33, 69118 Heidelberg, Germany

E-Mail: Jochen.Haeussler@eml.villa-bosch.de

(3) GIS Lab, University of the Bundeswehr Munich, Werner-Heisenberg-Weg 39, 85577 Neubiberg, Germany

E-Mail: Wolfgang.Reinhardt@UniBW-Muenchen.de

(4) Institute of Photogrammetry and Remote Sensing (IPF), University of Karlsruhe, Englerstr. 7, 76128 Karlsruhe,

Germany, E-Mail: Joachim.Wiesel@ipf.uni-karlsruhe.de

33

ters located in some of the biggest clefts (seefigure 1). If conspicuous extensions of a moni-tored cleft are registered, an alarm is triggeredand the local road is closed immediately. In thiscase a geologist then has to go to the area anddecide if this was a false alarm or if the nextlandslide can be expected shortly. The mainobjective of the project is to show the potenti-al of mobile geoservices prototypically for geoscientific applications like the Balingen landslide example.

The available primary data of the Balingen exa-mination area are fixed points with directionvectors, measurement plots of the extensome-ters, a digital elevation model, contour lines,path network, structural edges and slopes inscale 1:250. From these primary data the follo-wing interpreted data are constructed: strati-graphic boundaries and 3D strata bodies.

Typical requirements of the Balingen case studyto geoservices are:- Storage of 2.5D geodata (digital elevation

model), measurements data, and 3Dmodels.

- Retrieval of stored geodata and computeddeduced 2D profile sections.

- Online geodata acquisition and analysis ofthe terrain.

- Geodata editing of rocks and clefts in theterrain.

- Viewing of primary and interpreted data inthe terrain. Overlapping of the 3D modelwith the physical reality by AR methods.

The Balingen case study is a well-suited exam-ple to demonstrate the use of modern geoser-vices supporting environmental monitoringand prediction in the geosciences.

3. Methods & results

3.1 Mobile acquisition of geodataFor the mobile geodata acquisition in the givenproject environment of the Balingen case study,four main objectives have been investigated:

Figure 1: Clefts in the Balingen case study area with extensometer measurement units

34

(a) Refinement of concepts for mobile acquisi-tion of geodata.

(b) Development of a prototype system.(c) Definition of a detailed concept for the

quality assurance.(d) Proof of concepts – Application of the

system to the »Balingen test area«.

3.1.1 Refinement of concepts for mobile dataacquisitionThis point included in particular the followingimportant research issues:

- The development of refined workflows formobile acquisition of geodata, which makefully use of ubiquitous access to varioussources of information. This includes theselection of the servers and the download ofthe data usable for the current application,the feature acquisition, update etc. The spe-cific requirements of these workflows duringmobile online data acquisition have beenanalysed, evaluated and finally the applica-tion has been adjusted accordingly.

- Multi-sensor treatment: In the user scenarioin the Balingen test area, different kinds ofsensors like GPS receivers, total stations,extensometers and even laser scanning devi-ces have to be considered. In future OGCstandards like SensorWeb or SensorML willallow for an interoperable access to thesesensors. Until the sensors manufacturers sup-port these protocols alternative options haveto be considered. The use of common stan-dardised protocols like e.g. NMEA for GPSreceivers allows for an access to many sensorsof that specific kind independent of a certainvendor. This enables the application to accessand control a maximum number of sensors.

- Technical issues like the connectivity viawireless techniques have been investigated.In rural and especially forested areas cellularradio and WLAN have to be combined inorder to fully cover an area of interest and totransfer the data to the server. Some of theexperiences made are summarised in section

3.1.2 Development of a prototype systemA prototype for mobile data acquisition of geo-data has been developed. The most importantguidelines for this development have been:

- Development of an open architecture basedon standards, which means that no proprie-ty vendor dependent modules and interfaceshave been included. Proprietary interfacesconstrict in particular the connection to thedisparate data services and the applied sen-sors. Hence for the access of the heteroge-neous distributed servers standards like theOGC web map and feature services (WMSand WFS) and the geographic markup lang-uage (GML) have been employed. As men-tioned in the previous section the control ofthe sensors and the transfer of the measure-ment results should also be based on stan-dardised interfaces (if possible).

- A generic approach of data acquisition hasbeen developed which allows using thesystem in various applications. Therefore theclient application has to be able to adjustitself to the requirements imposed by thedata model. In particular, the measuring pro-cess and the templates for input of furtherattributes must be flexible and adaptable.The standardised service interfaces mentio-ned above are self-contained and self-descri-bing. The client application is using thesefeatures for the purposes of data access andacquisition. This means the client applicationdownloads capabilities (supported opera-tions and existing feature classes) and sche-ma information of the server at runtime.Such XML-schema contains all necessarydetails about the modeled feature types,their geometry and associated attributes aswell as the interrelations between featuresof one or different types. With the informa-tion contained in the XML-schema, it is pos-sible for the client application to adjust theacquisition process with regard to the requi-red attributes, geometry types and relations-hips of a particular feature type and to guidethe user through the whole data collectionprocedure. The templates for the input of

35

attribute values are generated automaticallyat runtime and the process of measuringgeometry elements is adjusted to the requi-rements of the feature type currently beingmeasured. This will assure that the collectionof the data is conforming to the schemaprovided by the particular server. (Mäs, S.,Reinhardt, W. & Wang, F. 2005a)

- The architecture of the client softwareallows for an easy extension and adaptationto the requirements of a particular applica-tion. Section 3.2 (Graphical geodata editor)includes further explanations regarding this.

3.1.3 Quality assurance conceptThe mobile interoperable access to heteroge-neous geodatabases and their update from thefield has far reaching consequences for thedata acquisition process. As mentioned before,this approach provides the possibility to checkthe newly acquired data in terms of quality andreliability directly in the field, which makes qua-lity management investigations necessary. In ourwork specific focus was given on finding a wayto define integrity constraints and transfer themto the client in a standardized way, as additionalinformation to the XML schema availablethrough the WFS. These constraints allow forrelated automatic checks during data collection

in the field. Therefore it has been investigatedhow spatial and other constraints can be for-malised in SWRL (Semantic Web Rule Language,W3C 2004b), which is a combination of OWL(Web Ontology Language, W3C 2004a) andRuleML (Rule Markup Language).

The defined constraints can be applied, forexample:

- on spatial relations between objects of thesame or of different classes,

- on a single or numerous attribute values, - on a defined relation between two attribute

values of one object,- or on a combination of spatial relations and

attribute values of different objects.

The rules are not restricted to relate only twoobject classes or attributes. Even complex spa-tial and topological relations between nume-rous spatial objects, together with their attri-bute values, can be described.

A simple example of a quality constraint forgeospatial data is given in figure 2. In naturallanguage the meaning of this rule is: »a clea-ring is always within a forest«. The two atomsin the antecedent define variables for each oneof the object classes. In the consequent thesevariables are used to set the object classes in

Figure 2: Example quality constraint encoded in SWRL.

36

relation. Therefore the »Within« relation isemployed. The denotation and the definitionof such spatial relations refer to the spatialoperators defined in OGC Filter EncodingImplementation Specification (OGC 2001).More details regarding the constraint formali-sation in SWRL and the quality assurance con-cept can be found in Mäs et al. (2005b).

3.1.4 Proof of concepts – application of thesystem to the »Balingen test area«As mentioned before the proposed mobileclient system should support the decision-making process of the geologist in case of analarm. The concepts and prototype implemen-tations have been proven in the Balingen testarea. Therefore a data model has been definedin cooperation with the users and a WFS serverhas been set up. This data model and thealarm scenario are described in more detail inKandawasvika et al. (2004). Figure 3 shows thesystem configuration for the field tests.

Central component of the configuration hasbeen the »Geotech in-field server«, which is

normally installed in a car at a point where aconnection to the central geodata warehousevia GSM / GPRS / UMTS is possible. This con-nection might not be necessary in every case.Sometimes it is better to have the databaseand the service directly running on the in-fieldserver, depending on the data volume and theavailable bandwidth / transfer rate. With thein-field server and a locally installed WLAN it ispossible to support several mobile users at thesame time. Mobile units are preferably tabletPCs, because of their capacity and performan-ce, but the client application should also sup-port other devices. For the data collection GPSand total station have been used as measure-ment devices.

In practical tests we found out that the wholearea of around 200* 150 m2 can be covered byusing only 2 WLAN access points (high-endAPs and antennas). Please notice that our exa-mination area is a very steep (50m height dif-ference), undulated terrain, which is coveredby tall trees. The user is able to move in thewhole area e.g. with a tablet PC always being

Figure 3: System configuration for the Balingen test area

37

connected to the geodata warehouse (via thein-field-server).

The performed field test verified the practicaladvantage of the developed concepts for thegeologist field tasks. The support to the geolo-gists included:

- Online request and visualisation of availableexisting data

- Positioning in the map- Possibilities to analyse data and do inspec-

tion measurements- Validation of the alarm- Acquisition of new features like e.g. ditches

or gaps- Quality assurance of these data

The quality assurance process not only does ithelp to acquire data conform to the server datamodel, but it also supports the decision-makingprocess and helps to identify dangerous situa-tions that are not always obvious to see. Forexample a constraint prohibiting a publiclyaccessible way to be in a certain distance to aditch would lead to an automatic warning to

the geologist while measuring such a newly for-med ditch. It is then up to his decision if andhow to react: he might close the way for publicaccess or at least mark the danger with somewarning signs. Anyway, for traceability he has todocument his decision in the system.

3.2 Graphical geodata editorA central component of the mobile acquisitionsystem is the graphical editor for geodata (seefigure 4). It is implemented as a lightweightJava application running on the mobile device,e.g. a ruggedized Tablet PC. The editor consti-tutes the user interface of the mobile dataacquisition system and provides the core func-tionality for acquiring and editing geodata inthe field. The central element of the editor GUIis a map which displays the geodata receivedfrom the server. The usual tools for navigatingthe map (pan, zoom), getting informationabout features and editing their attribute dataand geometries are being implemented.

A straightforward possibility for the visualiza-tion of the GML feature collections received by

Figure 4: Architecture of the mobile acquisition system

38

WFS is the transformation to SVG using XSLtransformations (XSLT). In this context, wehave investigated whether and how SVG canbe used to visualize and store geodata withina Java SVG implementation. As implementa-tion of SVG, we have chosen the open sour-ce Apache Batik framework. This frameworksupports SVG rendering and access to theSVG DOM from Java applications. We investi-gated how to manage the geodata with all itsnon-geometric attributes on the client (inmemory). It is not desirable to store the dataredundantly, e.g. as SVG DOM and as objectsof a GIS library in parallel. So, we need tohold all spatial as well as non-spatial attribu-tes of the GML features in the SVG represen-tation. While the GML geometries can betransformed to SVG geometries, the non-spa-tial attributes of the features can be stored in»svg:metadata« elements. This is a standardi-zed mechanism to embed arbitrary metadatawithin SVG documents. It is also possible toinsert elements of other namespaces into aSVG document. The Java SVG binding provi-des possibilities to address particular nodes inthe SVG DOM directly. This allows manipula-ting SVG subtrees representing single geogra-phic features. We realized a simple way tokeep all the geodata in one SVG DOM inclu-ding a change history (during one editing ses-sion) for each feature attribute.

When geodata is represented as SVG it is neces-sary to transform from GML to SVG and alsoback from SVG to GML in order to write editeddata back to the server. In order to develop ageneric application we had to investigate thegeneral possibilities and restrictions for bidirec-tional XSL transformations between GML andSVG. We used Styled Layer Descriptor (SLD)documents for the definition of the visual attri-butes of the different feature types. In Merdes,Häußler & Zipf (2005) it is shown, that it is pos-sible to build a generic application without thenecessity for application developers to write anyapplication or domain specific code by using theself-describing mechanisms of the WFS servicesand SLD as styling definition.

To integrate additional functionality for morespecific application scenarios we have develo-ped an architecture and runtime engine forplugins. These way parts of the additionalfunctionality can be integrated into the coreeditor keeping the actual editor componentthin. Thus the system can be adapted to theapplication scenario and also fosters the des-ired reusability of the software application. TheSVG representation of the geodata inside thecore editor is transparent to the plugins.

One group of plugins is »position sources«. Asposition sources we denote plugins that providegeo-positions with semantics well known to theuser. Examples of position sources are GPS andtotal station. A position source plugin encapsu-lates e.g. a single GPS device and provides itsmeasurements to the editor environment,together with additional information like time-stamp, precision etc.. With such a plugin thecurrent position can be displayed on the map.Several position sources can be connectedsimultaneously. The plugin infrastructure makesit possible for all plugins to connect to all regi-stered position sources at any time making theeditor a very flexible platform for additional andmore advanced functionality. Other deviceswhich do not function as position sources canconnect e.g. other measuring devices whichdeliver measurements for non-geometric attri-butes of new or existing features (temperature,soil parameters, precipitation measurements,etc.). Triggering of a single measurement or aseries of measurements in certain spatial or tem-poral intervals and insertion of the respectivelocated measuring point(s) into the database arepossible that way.

A third group of plugins are those that do notlink hardware devices to the editor, but provi-de other kinds of functionality. Examples ofsuch plugins include:- A feature acquisition plugin which generates

and adds new features using the positionsources as input for the new geometries.

- A plugin for quality assurance that controlsthe correctness of the edited features per-forming topological tests.

39

- A 2D Profile Section plugin: The user definesa planar profile section in the map of theeditor and gets a 2D profile generated bythe 3Dto2D service described in section 4.4.

The described infrastructure provides a flexibleand extensible solution for a mobile open stan-dards-based geodata editor.

For the described Balingen case study there areseveral ways in supporting the geologist withsuch a system. As there is online access to thegeodata, the geologist does not have to go tothe office (which in our case is hundreds ofkilometres away) to consult the latest data. Inthe case of a false alarm, this makes it possibleto bring the endangered road into serviceagain very quickly. Furthermore, observationsof the geologist can be added to the geodat-abase server directly in the field, thereforebeing immediately accessible by other specia-lists. The quality control plugin could widelymake post processing of the data superfluous,which is again important due to the big distan-ce between the monitored area and the office.Decision making is assisted by the availability

of calculation-intensive services like the men-tioned 3Dto2D service, which is important for(infield) interpretation of the acquired data.

3.3 Augmented reality clientA mobile AR prototype system (see figures 5 &6) has been designed and developed (Wiesel,Staub, Brand & Coelho, 2004), to support geo-scientists in the field. The system is based onan IEEE1394 camera and a monoscopic HeadMount Display (HMD), hardware for naviga-tion, an inertial measurement unit (IMU) andthe necessary computing equipment mountedon a backpack.

The system has been designed to allow ahuman to move around in the test area andanalyse the geological structures and landsli-des by - Inspecting the scene.- Overlaying the terrain with time stamped 3D

geodatabase content (e.g. profiles or displa-cement vectors, geological data).

- Gathering new geodata (e.g. new clefts or rifts).- Entering and editing geodata in real time

Figure 5: Proposal for an Augmented Reality System Architecture and Hardware Mockup

40

into the geodatabase.- Entering attribute data into the database.- Writing reports about current situations in

the field.

Navigation and orientation of the sensorsystem (based either on a camera or a headmounted display) is crucial for the usability ofsuch mobile AR clients. We have combined aGPS-receiver and a low cost IMU to achieve apositioning precision in the cm range.

By using Real Time Kinematic (RTK) GPS posi-tioning we can achieve a positioning precisiondown to ±1 cm. Yet, in a typical geoscientificapplication, we have to deal with GPS dro-pouts while moving around in the field. Toovercome this problem we are using an IMUmounted on top of the system, which can pro-vide velocity, position and attitude of the HMDand camera for a short time period. Ongoingstudies (Staub, Coelho & Leebmann, 2004)calibrate and filter sensor readings to close thegaps of satellite signal outages.

Furthermore, we use a so called wrist-wornkeyboard to interact with the system. Manyimportant features are implemented so far and

can be triggered by predefined keyboardshortcuts. For example, it is possible to changethe transparency, line width of the virtualobjects or the lighting conditions of the virtualscene, as well as loading or removing objects.It is also possible to zoom in the scene, panand rotate the virtual objects. The feedbacksignal sent by the ARS after receiving such acommand is either visual or acoustic. Thisdepends on the action performed by the user.

The human computer interface had to be desi-gned straightforward without occluding thearea of interest in the »real world«. Therefore,a transparent interface with minimal contentsand alternative controls on demand is propo-sed. It consists of permanent output of theuser’s position (Gauß-Krüger coordinates andellipsoidal height) and orientation, which isshown in the upper-most position of thedisplay. An overview window is placed at thelower right corner of the display, which can beremoved if it is occluding some importantobjects. These three components combine allthe necessary positioning and orientationinformation to give the user knowledge abouthis (or her) location in the field. In the centre ofthe field of view a crosshair is displayed. This is

Figure 6: Testing the proposed ARS outdoors

41

used for capturing additional information fromvirtual and real objects. This is a useful featureof the ARS, because it offers the possibility toreceive information about the objects in real-time. Non-visible information is gathered fromthe artificial objects.

To achieve a realistic impression of the super-imposed scene, it is important to provide asmooth transition of virtual and real objects.Depth information is needed to fit virtualobjects into the environment, which mayocclude parts of the virtual scene. In urbanenvironments, the developed ARS uses additio-nal building models to retrieve depth informa-tion and to compute occlusion (Coelho, 2004).In the context described in this article, the userhas to operate in a forest. No information onthe location and size of trees, which are the

main source of occluding objects, is available.To operate in such an environment, a head-mounted stereo camera system is used toobtain necessary depth information on the flywith a dense two-frame stereo algorithm.

In the Balingen test area, newly discoveredclefts or rifts have to be surveyed by a geo-scientist. Therefore, (Leebmann, 2005) propo-ses a methodology to gather such informationfrom a distance. It is necessary to survey theobject of interest from a minimum of two dif-ferent points of view. This way it is possible tocalculate the Gauß-Krüger coordinates. Figure7 shows the approach tested by surveying anedge and the augmented view on it after cal-culating its position in the field.

Figure 7: Survey an edge in terrain

42

3.4 3D geodatabase servicesIn order to provide geoservices accessible byarbitrary mobile clients, a 3D geodatabasesystem should provide an open service archi-tecture giving access to the whole functionali-ty of the underlying geodatabase, while ensu-ring the communication with the mobileclients based on interoperable open protocols.

In the prototype of our 3D geodatabasesystem the service framework is provided by arich set of single services implemented asremote method calls in Java. The service fra-mework supports the combination of the sin-gle services to so called service chains – whichare then capable of providing complex proces-sing capabilities inside the database system.The data transfer between services and clientsis primarily based on the extensible markuplanguage with an associated XML schema. Theoutput services currently cover a specialisedXML format, VRML and X3D, and are extensi-ble by the user through XSL transformation toarbitrary XML and text based formats.

To support applications like the Balingen landslide scenario, the geodatabase services mustprovide access to and update capabilities ofentire 3D-models and related geometric andthematic data from mobile devices in the field.The following sections give two examples ofour 3D geodatabase system which are meantto address these capabilities. The first exampledescribes the support of constraint mobiledevices (PDAs), which are not yet capable ofworking with complex 3D models, through aspecial application service. The second exam-ple explains our ongoing research with sup-porting update capabilities on mobile devicesthrough the usage of mobile databases inte-grated with the server geodatabase system.

3.4.1 Supporting constrained mobile devicesA comprehensive subsurface model may con-sist of hundreds of geological bodies, eachrepresented by complex objects, e.g. triangula-ted surfaces or volumes, composed of up tomore than a hundred thousand elements (e.g.

triangles or tetrahedrons). Considering cons-traint clients, e.g. PDAs combined with a GPS,both the transmission and the graphical repre-sentation of such a complex model are not yetrealistic, because of insufficient available band-width and performance of the graphicaldisplay. On the other hand, the geoscientist inthe field often needs only a selected part of theinformation, specified by e.g. a 3D region, astratigraphic interval, a set of thematic attribu-tes or some other geometric and thematic cri-teria. Even such reduced information may betoo large for use in the field, motivating theuse of techniques of data reduction and pro-gressive transmission (Shumilov, Thomsen,Cremers & Koos, 2002).

Therefore – due to today’s hardware restric-tions on PDAs - graphical representation of a3D model could be reduced to a sequence of2D sections and projections. By sliding throughsuccessive sections, even a 2D display can pro-vide insight into the form and structure of acomplex 3D body. However, this means thatservices have to be provided that compute 2Dprofile sections for arbitrary planes of a 3Dsubsurface model. Such a service allows thefield geologist to compare the actual observedsituation with information provided by thesubsurface model, and to take decisions onsampling accordingly.

We are exemplary presenting such a service,the so called 3Dto2D-Service. It provides thederivation of 2D geological profiles from a 3Dsubsurface model for a specified arbitraryplane in the 3D space. Additionally, furtherobjects spatially located in a specified distanceto the plane, which are of interest for interpre-tation, can be projected onto the computed2D profile. The service is composed of the fol-lowing single services provided by our serviceframework (see figure 8):

- RetrieveService – supports queries on com-plex geoscientific 3D models.

- PlaneCut – cuts a planar profile through the3D model for a spatially specified arbitrary3D plane.

43

- PlaneProjection – projects interesting 3Dobjects onto the plane profile, which arespatially located in a specified certain distan-ce of interest to the 3D plane.

- AffineTransform – transforms the resulting3D objects into a 2D xy plane.

Figure 8 shows the principle steps. The usermay specify a planar profile section betweenendpoints A and B, with further data such asspatially neighboured boreholes b1 and b2.Figure 8 (a) shows the location in map planeview. The block view of the 3D model is givenin figure 8 (b) and figure 8 (c) shows the viewof profile section with part of model removed.Finally figure 8 (d) shows the resulting 2D pro-file section with the projected borehole profilesas additional information.

Each of the single services of the 3Dto2D-Service implies geometric operations requiringa considerable amount of time (Breunig, Bär &Thomsen, 2004). Therefore, in order to reducethe length of transactions, the single servicesare operating each in transactional mode.Single failures of one service can be compen-sated by restarting this single service, and donot require starting the whole service chainfrom the beginning.

3.4.2 Supporting update operations with deta-ched mobile databasesA 3D geodatabase system for geological appli-cations should enable the geologists in thefield, as well as in the laboratory, to refer to ashared common 3D model during the process

of data caption, processing, interpretation andassessment. The cycle of steps involved in upda-ting a geological model can be rather long andthe result may never be free of subjective appre-ciation. Therefore it is advisable to use strategiesof version management to control the evolutionof the 3D model rather than supporting directediting by transaction management.

In the following we will give an overview ofhow we address update capabilities in the fieldusing mobile databases (Bär & Breunig, 2005).The approach is based on a version manage-ment extension of our 3D geodatabase server.The mobile database is regarded as a specialclient to this version management system andtherefore updates on local 3D objects duringoffline mode are integrated back to the 3Dgeodatabase system as new revisions of thepreviously replicated 3D object. This approachmakes it possible to review changes in 3Dobjects or to complete 3D models before theyare merged to the original 3D model of thedatabase system.

Therefore our 3D geodatabase system hasbeen extended with version managementcapabilities. The generic version managementextension is motivated by the object-orientedversion model of Schönhoff (2002) and provi-des the management of history graphs of ver-sions and a hierarchy of workspaces as versionrepositories. For an overview of version modelswe refer to Katz (1990).

Figure 9 gives an overview of the general struc-ture of the version management system as

Figure 8: 3Dto2D service

44

seen by the user. The existing 3D databasesystem is called the release database in whichthe releasable database objects reside. Tomodify a database object from the releasedatabase it has to be put under version controlfirst. This operation creates a new designobject for the database object and the initialversion in the main workspace. Starting fromthis initial version, new modifications result ina new version of the design object. Inside aworkspace there exist only revisions of versions(linear history). Modifications to versions whichare meant to provide a rather alternative repre-sentation of an object are called alternativeversions. Such alternative versions have to becreated in a new child workspace with an ownrevision history. Propagating a later revision ofan alternative version back to the parent work-space is called merging. To execute a merge,no conflicts with the latest revision in theparent workspace are allowed. Otherwise aconflict resolution must be done beforehandby the user. This way mature versions can bepropagated upwards the workspace hierarchyand finally replace the original database objectin the release database.

Besides the concepts of design objects andtheir versions, the concept of configurationshas to be supported in the version manage-ment system. Configurations allow groupingspecific versions of several design objects. Withconfigurations the notion of 3D models fromgeology as a consistent set of several designobjects can be realized. Therefore they must beextensible to allow forcing constraints on theadded versions such as »no geometric inter-section between the objects represented bythe added versions is allowed«. Furthermore,configurations provide a way for batch propa-gation of a consolidated set of versions bet-ween the workspaces.

The version management presented so fardoes not force a specific representation of ver-sions in the system. As the complex 3D objectsused in geo applications can internally consistof up to several hundred thousands of simple-xes, the storage of a complete object for eachversion is impracticable. Therefore, we repre-sent a version as the set of changes to its revi-sion or alternative predecessor in the versionhistory (delta storage). Beside the reduction of

Figure 9: Structure of the version management system

45

used storage, this approach enables efficientalgorithms for conflict detection and also forproviding change histories. Having used simpli-cial complexes as the underlying data model ofour 3D objects, the changes are representedinside the version management system asadditions / deletions of single simplexes, com-plete components and the associated thematicchanges. Conversion operations between theversion representation and the database objectrepresentation ensure that all the operationsfrom the geodatabase system can also beapplied to the versions of 3D objects. Thismakes it possible for example to create profi-les sections with the described 3Dto2D-Service from different versions of a 3D objectand therefore to compare differences also onconstraint devices in the 2D space.

The version management extension is integra-ted with the service framework of the geodat-abase system. The communication betweenmobile databases and the version manage-ment is based on the XML representation of3D objects or change sets between versions.Although the version management system wasdesigned with the support of detached mobiledatabases (offline usage mode) in mind, theintegration with the service framework alsoenables every mobile or static client to use theversion management capabilities provided.

3.4.3 The 4D-Extension: Managing spatialobjects varying with timeLandslides obviously involve changes of loca-tion and form of spatially extended objectsdepending on time. The modelling of displace-ments and deformations can be done bynumerical models, or by a scientist designing asufficient number of discrete states of themodel at different time instants, based ofobservations and measurements. The task ofthe geodatabase is to manage the resultingtime-dependent spatial objects (4D-objects),and provide services that allow to retrieve thestate of a spatial object at any given time of itslifespan by appropriate searching and interpo-lation methods. The 4D-extension of the

geodatabase is based on earlier experienceswith the timescene tree (Polthier & Rumpf,1995), with the GeoToolKit (Balovnev et al,2004), and on concepts presented byWorboys (1995). Rolfs (2005) presents adetailed discussion of the spatiotemporalextension and its implementation as well asmore extensive references.

A time-dependent spatial object is consideredas a function defined on a time interval, withvalues in a set of spatial 3D-objects. Thisimplies that in addition to the spatial modeldiscussed in previous chapters, a model and adiscretisation of time is required. It consists oftime instants t and time intervals (ti,ti+1) thatare concatenated to form time sequences. Anumber of temporal operations support setoperations and predicates, especially to deter-mine intersections. Searching is supported by atemporal index based on Bentley’s segmenttree, cf. (de Berg et al., 2000). The temporalbehaviour of objects is defined in an interfacethat is inherited by all temporal and spatio-temporal classes.

The central questions concern the discretisa-tion of time and the necessary interpolationbetween discrete states of the object, changesof topology, i.e. of meshing and of connectivi-ty. As the static 3D-objects of a geologicalmodel may already comprise meshes of con-siderable size (up to several 100000 elements),a simple repetition of slightly changed copiesat each time step may result in intolerably bigand redundant 4D-objects. Therefore, at-tempts are made to reduce redundancy inparts of 4D-objects that are either static orshow only very small changes over time orchanges that dependent linearly on time, byallowing for different density of discretisationin different parts.

Whereas the geometry (location, extent andform) of a 4D-object may vary continuously orby steps, its topology (meshing, connectivity)can only change at discrete steps. Moreover, itseems reasonable to assume continuous defor-mations and displacements taking place more

46

frequently than disruptions or re-meshingbecause of extreme deformations or changesof size. The time-dependent geometry therefo-re assumes that a number of contiguous dis-cretisation intervals where continuous displa-cements and deformations occur withoutchange of topology, can be grouped togetherto larger time intervals at the boundaries ofwhich meshing and connectivity change.

A 4D-object is composed of spatiotemporal(ST-) elements of two kinds: one is defined as apair (t, s d(t)) , with a time instant t and a d-sim-plex s d(t), the other one is defined as a tuple (ti, ti+1, s d(ti), s d(ti+1), f) – it consists of an opentime interval ]ti, ti+1[, a pair of spatial d-simple-xes s d(t) defined at the interval boundaries, andan interpolation function f that, for any t in theopen interval ]ti,ti+1[, yields a snapshotf(t)=(t,s d(t)), The present geodatabase supportslinear interpolation of vertex co-ordinates, butthe approach can be generalised to more ela-borate interpolation methods. Considered as a4D-geometry object, such an ST-elementresembles a deformed prism (figure 10).

The ST-elements are grouped into a number ofspatiotemporal (ST-) components, each with acommon discretisation of time and a constantand connected mesh. Between spatially neigh-bouring ST-components, time discretisationmay vary, and at the contact of subsequent ST-components in time, meshing and connectivitymay change (figure 11). Different discretisation in space or time maycause inconsistencies at the contact of ST-com-ponents. These might be avoided by carefullydesigning the 3D-objects, or by the user impo-sing appropriate constraints.

In a simple case, an ST-object may consist of asingle ST-component, with a common time dis-cretisation, and no discontinuities.

Besides methods for the loading and checkingof ST-objects, for intersection with 4D searchboxes, numerical functions etc., there are twomain operations supported by an ST-object Od

to be mentioned: 1. The calculation, for any given time t within

its interval of definition, of its 3D-snapshotSd(t), yielding a ST-object, which is a d-sim-plicial complex in R3 with an additional timestamp, and can be subject to any spatialoperation defined in the 3D-geodatabase.

2. The intersection with a 4D query box, resul-ting in a new ST-object defined by the inter-section with the query box.

In principle, a combined spatiotemporal indexcan be defined by extending the well known R-tree to four dimensions. In the present model,however, separate indexes for time – a seg-ment tree, and an R-tree for space are used,thus keeping in line with the static 3D- model(Rolfs, 2005).

Figure 10: A spatiotemporal (ST-) element

47

4. ConclusionsThis contribution reported about typical geo-scientific requirements to new geoservices. In acase study dealing with land slides at Balingen,south-west Germany, it has been shown thatthe mobile acquisition, visualization andmanagement of spatial data can simplify geo-scientific work by digitally supporting geo-scientists directly in the field. Contributions ofthe project partners to a prototype of a distri-buted software system of geoclients and servi-ces usable by mobile geoscientific applicationswere discussed. A mobile graphical editor forgeodata acquisition, a mobile AR client andgeodatabase services were presented in detail.We are optimistic that in the future, the mer-ging of the 3D database content with the livescene in real time executed by AR methods willhelp the geoscientific expert in the field effi-ciently to examine geological subsurface struc-tures and to compare them with visible faultlines at the surface. The 3Dto2D geodatabaseservice, for example, meets the introducedgeoscientific requirements by remotely compu-ting 2D profile sections from a 3D subsurfacemodel and by visualizing the database queryresults on the mobile client. For the future wesee research demands in integrating single

geoservices into geodata infrastructures and indeveloping new mobile data acquisition andvisualization tools coupled by efficient geodat-abase services.

AcknowledgementsThe funding of the research project »Advance-ment of Geoservices« (Weiterentwicklung vonGeodiensten) by the German Ministry ofEducation and Research (BMBF) by grant no.03F0373B et al. within the framework of thegeotechnology initiative (http://www.geotech-nologien.de) is gratefully acknowledged. The responsibility for the contents of this publi-cation is by the authors. We also thank Dr.Ruch from the LGRB Baden-Württemberg forproviding an interesting application and datafor the Balingen test example. An earlier version of this contribution has been publishedas GEOTECH-126.

Figure 11: A spatiotemporal (ST-) object composed of 3 ST-components with different time discretisation

48

ReferencesAgarwal, S., Arun, G., Chatterjee, R., Speck-hard, B. & Vasudevan, R. (2003) Long transac-tions in an RDBMS. 7th Annual Conferenceand Exhibition: A Geospatial Odyssey, Geo-spatial Information and Technology Associa-tion (GITA).

Balovnev, O., Bode, T., Breunig, M., Cremers,A.B., Müller, W., Pogodaev, G., Shumilov, S.,Siebeck, J., Siehl, A. & Thomsen, A. (2004)The story of the GeoToolKit – An Object-Oriented Geodatabase Kernel System. Geo-Informatica, 8 (1), 5-47.

Batty, P.M. (2002) Version management revisi-ted. Proc. of GITA Annual Conference, Florida.Bär, W. & Breunig, M. (2005) Usage of MobileDatabases for Mobile Geoscientific Applications.Accepted for publication in AGILE 2005 Pro-ceedings, 8th AGILE Conference on GeographicInformation Science, Estoril, Portugal, 9 p.

Bernard, G., Ben-Othman, J., Bouganim, L.,Canals, G., Chabridon, S., Defude, B., Ferrie, J.Gaucarski, S., Guerraoui, R., Molli, P., Pucheral,Ph., Roncancio, C., Serrano-Alvarado, P. & Val-duriez, P. (2004) Mobile Databases: a Selectionof Open Issues and Research Directions. ACMSIGMOD Record, Vol. 33, No. 2, June, 6 p.

Breunig, M. (2001) On the Way to Compo-nent-Based 3D/4D Geoinformation Systems.Lecture Notes in Earth Sciences, No. 94,Springer, 199 p.

Breunig, M., Cremers, A.B., Shumilov, S. &Siebeck, J. (2003) Spatio-Temporal DatabaseSupport for Long-Period Scientific Data. DataScience Journal, International Council forScience, Vol. 2, 175-191.

Breunig, M., Malaka, R., Reinhardt, W. &Wiesel, J. (2003) Advancement of Geoservices.Geotechnologien Science Report No. 2,Information Systems in Earth Management,Potsdam, 37-50.

Breunig, M., Türker, C., Böhlen, H., Dieker, S.,Güting, R.H., Jensen, C.S., Relly, L., Rigaux, P.,Schek H.J. & Scholl M. (2003) Architecture andImplementation of Spatio-Temporal DBMS.Spatio-Temporal Databases – The CHORO-CHRONOS Approach, Lecture Notes in Com-puter Science Vol. 2520, Springer, 219-264.

Breunig, M., Bär, W. & Thomsen, A. (2004)Usage of Spatial Data Stores for Geo-Services.Proceeding 7th AGILE Conference on Geo-graphic Information Science, Heraklion,Greece, 687-696.

Brinkhoff, T. (1999) Requirements of TrafficTelematics to Spatial Databases. Proceedingsof the 6th Intern. Symposium on Large SpatialDatabases, Hong Kong, China. In: LNCS, Vol.1651, 365-369.

Coelho, A. H. (2004) Erweiterte Realität zurVisualisierung simulierter Hochwasserereignis-se. Phd thesis, Karlsruhe, Univ., Fak. Für Bau-ingenieur-, Geo- und Umweltwissenschaften.

Dahne, P. & Karigiannis, J. N. (2002) Archeo-guide: system architecture of a mobile outdooraugmented reality system. Proceedings of theInternational Symposium on Mixed and Aug-mented Reality, ISMAR 2002, 263-264.

de Berg, M. et al. (2000): Computational geo-metry – algorithms and applications, Springer.Feiner, S., MacIntyre, B., Höllerer, T. & Webster,T. (1997) A touring machine: Prototyping 3Dmobile augmented reality systems for explo-ring the urban environment. Proceedings ofthe 1st Int. Symposium on Wearable Com-puters, ISWC97, October 13-14, 1997, Cam-bridge, 208-217.

Gollmick, Ch. (2003) Client-Oriented Repli-cation in Mobile Database Environments. Je-naer Schriften zur Mathematik und Informatik,Math/Inf/08/03, University of Jena, Germany.

49

Güting, R. H., Bohlen, M. H., Erwig, M.,Jensen, C. S., Lorentzos, N. A., Schneider, M. &Vazirgiannis, M. (2000) A foundation for repre-senting and querying moving objects. ACMTransactions on Database Systems 25, 1, 1-42.

Höllerer, T., Feiner, S., Terauchi, T., Rashid, G. &Hallaway, D. (1999) Exploring mars: Develo-ping indoor and outdoor user interfaces to amobile augmented reality system. Computer &Graphics, Vol.23, No. 6. Elsevier Publishers,779-785.

Kandawasvika, A., Mäs, S., Plan, O., Reinhardt,W. & Wang, F. (2004) Concepts and developmentof a mobile client for online geospatial dataacquisition. GeoLeipzig 2004 - Geowissenschaf-ten sichern Zukunft. Schriftenreihe der Deut-schen Geologischen Gesellschaft, Vol. 34, p. 79.

Kandawasvika, A. & Reinhardt, W. (2005)Concept for interoperable usage of multi-sen-sors within a landslide monitoring applicationscenario, Accepted for publication in AGILE2005 Proceedings, 8th AGILE Conference onGeographic Information Science, Estoril,Portugal, 10 p.

Katz, R.H. (1990) Towards a Unified Frameworkfor Version Modeling in Engineering Databases.ACM Computing Surveys 22, No. 4, 375-408.

Leebmann, J. (2005) Dreidimensionale Skizzen inder physikalischen Welt. Phd thesis, Karlsruhe,Univ., Fak. Für Bauingenieur-, Geo- und Umwelt-wissenschaften, submitted on July 5th 2005.

Livingston, M.A., Brown, D. & Gabbard, J.L.(2002) An augmented reality system for milita-ry operations in urban terrain. Proceedings ofInterservice/ Industry Training, Simulation andEducation Conference, IEEE, Zurich, 31-38.

Mäs, S., Reinhardt, W. & Wang, F. (2005a)Concepts for quality assurance during mobileonline data acquisition. Accepted for publica-tion in AGILE 2005 Proceedings, 8th AGILEConference on Geographic Information Sci-

ence, Estoril, Portugal, 10 p.Mäs, Stephan; Wang, Fei; Reinhardt, Wolfgang(2005b): »Using Ontologies for IntegrityConstraint Definition«, In: Proceedings of the4th International Symposium On Spatial DataQuality, pp. 304-313, August 25- 26, 2005,Peking, China

Merdes, M., Häußler, J. & Zipf, A. (2005)GML2GML: Generic and Interoperable Round-Trip Geodata Editing – Concepts and Example.Accepted for publication in AGILE 2005 Pro-ceedings, 8th AGILE Conference on GeographicInformation Science, Estoril, Portugal, 10 p.

Mostafavi, M.-A., Edwards, G. & Jeansoulin, R.(2004): Ontology-based method for qualityassessment of spatial data bases. In: ISSDQ '04Proceedings, 49-66.

Mutschler, B. & Specht, G. (2003) Implementa-tion concepts and application development ofcommercial mobile database systems (in ger-man). Proceedings Workshop Mobility and In-formation Systems, ETH-Zürich, Report No.422, 67-76.

Newell, R.G. & Easterfield, M. (1990) VersionManagement – the problem of the long trans-action. Smallworld Systems Ltd., TechnicalPaper 4.

OGC (2001): Filter Encoding ImplementationSpecification, Version: 1.0.0, OpenGIS® Imple-mentation Specification, OpenGIS project do-cument: OGC 02-059, 19 September 2001

Polthier, K. & Rumpf, M. (1995): A Concept ForTime-Dependent Processes. In: Goebel, M.,Mueller, H., Urban, B. (eds.): Visualization inScientific Computing, Springer, 137-153.

Pundt, H. (2002): Field Data Acquisition withMobile GIS: Dependencies Between DataQuality and Semantics, GeoInformatica 6:4,2002, Kluwer Academic Publishers, 363-380.

50

Roddick, J.F., Egenhofer, M.J., Hoel, E. &Papadias, D. (2004) Spatial, Temporal andSpatio-Temporal Databases – Hot Issues andDirections for PhD Research. ACM SIGMODRecord, Vol. 33, No. 2, June, 6 p.

Rolfs, C. (2005): Konzeption und Implemen-tierung eines Datenmodells zur Verwaltungvon zeitabhängigen 3D-Modellen in geowis-senschaftlichen Anwendungen. DiplomarbeitGeoinformatik, Fachhochschule Oldenburg FBBauwesen u. Geoinformation, 90p.

Ruch, C. (2002) Georisiken: Aktive Massenbe-wegungen am Albtrauf. LGRB-NachrichtenNo.8/2002, Landesamt für Geologie, Rohstoffeund Bergbau (LGRB), Baden-Württemberg, 2 p.

Schönhoff, M. (2002) Version Management inFederated Database Systems. DISDBIS 81,Akademische Verlagsgesellschaft (Aka), Berlin.Sellis, T. (1999) Research Issues in Spatio-tem-poral Database Systems. In: Güting, Papadias& Lochovsky (Eds.): Advances in SpatialDatabases. Lecture Notes in Computer Science1651, Springer.

Staub, G., Coelho, A. & Leebmann, J. (2004) Ananalysis approach for inertial measurements of alow cost IMU. Proceedings of the 10thInternational Conference on Virtual Systemsand Multimedia, Ogaki, Japan, November,2004, 924-933.Shumilov, S., Thomsen, A.,Cremers, A.B. & Koos, B. (2002) Managementand Visualization of large, complex and time-dependent 3D Objects in Distributed GIS. Proc.10th ACM GIS, McLean (VA).

W3C (2004a): OWL web ontology languageReference. Editors: Dean, Mike; Schreiber,Guus, W3C Recommendation, 10. February2004, Available at:http://www.w3.org/TR/owl-ref/W3C (2004b): SWRL: A Semantic Web RuleLanguage Combining OWL and RuleML, W3CMember Submission 21 May 2004, Availableat: http://www.w3.org/Submission/ + 2004/SUBM-SWRL-20040521/

Wiesel, J., Staub, G., Brand, S. & Coelho, A.H.(2004) Advancement of Geoservices – Aug-mented Reality GIS Client. GeotechnologienScience Report No. 4, Aachen, 94-97.Wolfson, O. (2002) Moving Objects Informa-tion Management: The Database Challenge.Proc. of the 5th International Workshop onNext Generation Information Technologies andSystems, LNCS 2382, Springer-Verlag, 75-89.

Worboys, F. M. (1994): A unified model forspatial and temporal information. ComputerJournal vol. 37 no. 1, 26-34.

51

52

Development of a data structure and tools for the integration of heterogeneous geospatial data sets

AbstractThe integration of heterogeneous geospatialdata sets offers extended possibilities of deri-ving new information which could not beaccessed by using only single sources. Differentacquisition methods, data schemata and upda-ting periods of the topographic content leadsto discrepancies in geometry, accuracy andtopicality which hampers the combined usageof these data sets. The integration of differentdata sets – in our case topographic data, geo-scientific data and imagery – allows for a con-sistent representation, the propagation ofupdates from one data set to the other and theautomatic derivation of new information. Inorder to achieve these goals, basic methods forthe integration and harmonisation of datafrom different sources and of different typesare needed. To provide an integrated access tothe heterogeneous data sets a federated spati-al database is developed. We demonstrate twogeneric integration cases, namely the integra-tion of two heterogeneous vector data sets,and the integration of raster and vector data.

1. IntroductionGeospatial data integration is often applied tosolve complex geoscientific questions. To ensu-re successful data integration, i.e. ensure thatthe integrated data sets fit to each other and

can be analysed in a meaningful way, an intel-ligent strategy is required due to the fact thatthese data sets are mostly acquired using diffe-rent methods, quality standards and at differentpoints in time. Differences between printedanalogue maps were not as apparent as arethose of digital data of today, when differentdata sets are overlaid in modern GIS-applica-tions. Integrating different data sets allows for aconsistent representation and for the propaga-tion of updates from one data set to the other.To enable the integration of vector data sets, astrategy based on semantic and geometricmatching, object based linking, geometricalignment, change detection, and updatingwill be used. With this described strategy theactual topographic content from an up-to-datedata set can be used as a reference to enhan-ce the content of certain geoscientific datasets. In addition, the integration of two datasets with the aim to derive an updated data setwith an intermediate geometry based on givenweights is possible. The integration of rasterand vector data sets is the second integrationtask dealt with in this paper. As an example,field boundaries and wind erosion obstaclesare extracted from aerial imagery exploitingprior GIS knowledge. One application area aregeoscientific questions, for example the deri-vation of potential wind erosion risk fields,which can be generated with field boundaries

Butenuth M. (1), Gösseln G. v. (2), Heipke C. (1), Lipeck U. (3), Sester M. (2), Tiedge M. (3)

(1) Institute of Photogrammetry and GeoInformation, University of Hannover, Nienburger Str. 1, 30167 Hannover,

Germany, {butenuth, heipke}@ipi.uni-hannover.de

(2) Institute of Cartography and Geoinformatics, University of Hannover, Appelstr. 9a, 30167 Hannover, Germany

{goesseln, sester}@ikg.uni-hannover.de

(3) Institute of Practical Informatics, University of Hannover, Welfengarten 1, 30167 Hannover

{ul, mti}@dbs.uni-hannover.de

53

and additional input information about theprevailing wind direction and soil parameters.Another area is the agricultural sector, whereinformation about field geometry is importantfor tasks such as precision farming or themonitoring and control of subsidies.The paper is structured as follows: The followingsection gives an overview of the state of the artconcerning the topic of data integration.Afterwards, the used data sets are presentedand an architecture for database supportedintegration is described. Methods for the inte-gration of vector/vector and raster/vector dataintegration are highlighted in the following sec-tion. Results demonstrate the potential of theproposed solution, finally a set of conclusions isgiven and further work is discussed.

2. State of the art of geospatial data integrationThe integration of vector data sets presentedin this paper is based on the idea of comparingtwo data sets, while one is used as a referenceand a second one – the candidate – is alignedto the first one, which is a general matchingproblem, see e.g. Walter and Fritsch (1999).For the integration of multiple data sets, it hasbeen shown how corresponding objects canbe found when several data sets have to beintegrated (Beeri et al., 2005). Due to the com-plexity of the integration problem it is very dif-ficult to solve this task with one closed system,therefore the development of a strategy basedon component ware technology was proposed(Yuan and Tao, 1999) and a software prototy-pe for the vector data integration has beendeveloped as a set of components to ensurethe applicability in different integration tasks.While this approach uses a reference data setto enhance and update the topographic con-tent of a candidate data set, data integrationcan also be used for data registration, whenone data set is spatially referenced and theother has to be aligned to it (Sester et al.,1998). In order to geometrically adapt datasets of different origin, rubber sheeting me-chanisms are being applied (Doythser, 2000).

Strategies applied to cadastral data based ontriangulation to enhance the rubber-sheetingprocess have been presented by Hettwer andBenning (2000).The recognition of objects with the help ofimage analysis methods starts often with anintegration of raster and vector data, i.e. usingprior knowledge to support object extraction.An integrated modelling of the objects of inte-rest and the surrounding scene exploiting thecontext relations between different objectsleads to an overall and holistic description(Baltsavias, 2004). In this paper, the extractionof field boundaries and wind erosion obstaclesfrom imagery is chosen to demonstrate themethodology integrating raster and vectordata. In the past, several investigations regar-ding the automatic extraction of man-madeobjects have been carried out (e.g. Mayer,2001). Similarly, the extraction of trees hasbeen accomplished, cf. Hill and Leckie (1999)for an overview of approaches suitable forwoodland. In contrary, the extraction of fieldboundaries is not in an advanced phase: a firstapproach to update and refine topologicallycorrect field boundaries by fusing raster-ima-ges and vector-map data is represented inLöcherbach (1998). The author focuses on thereconstruction of the geometry and features ofthe land-use units, however, the acquisition ofnew boundaries is not discussed. In Torre andRadeva (2000) a so called region competitionapproach is described, which extracts fieldboundaries from aerial images with a combi-nation of region growing techniques and sna-kes. To initialise the process, seed regions haveto be defined manually, which is a time andcost-intensive procedure.In order to connect heterogeneous databases,first so-called multi-database architectures hadbeen discussed for loose coupling. Subse-quently, so-called federated databases havebeen chosen to support closer coupling (Con-rad, 1997). Federated databases allow integra-ting heterogeneous databases via a globalschema and provide a unified database inter-face for global applications. Local applicationsremain unchanged, as they still access the

54

databases via local schemata. For databaseschema integration a broad spectrum ofmethods has been investigated (Batini et al.,1986), but identifying objects is typicallyrestricted to one-to-one-relationships. In con-text of geospatial integration more sophistica-ted methods are needed, to incorporate com-plex correspondences between objects (many-to-many-relationships), which usually are notconsidered in federated databases. Whereasthere are a lot of overview articles of spatialdatabases (e.g. Rigeaux, 2002), federated spa-tial databases are hardly investigated with theexception of (Devogele, 1998; Laurini, 1998).

3. Architecture for integrationDifferent geospatial data sets which representthe same real world region, but cover differentthematic aspects, are acquired with respect todifferent needs. In this section we present anarchitecture that provides an integrated accessto heterogeneous data sets. It is designed tostore and export results of the vector/vectorand the raster/vector integration steps. Thistask is accomplished according to the para-

digm of federated databases. For this purposethe known architecture of a federated databa-se is expanded to handle geospatial data. Inorder to select certain objects satisfying givensemantic criteria it is possible to define map-pings to harmonise the attributes of the diffe-rent data sets. Furthermore, the database pro-vides mechanisms to pre-process geospatialobjects for the integration of raster and vectordata. Fig. 1 gives a simplified overview of therealised system architecture with respect to theinteraction between the federated databaseand the integration process, namely objectmatching and extraction.

In the next section, the involved vector and rasterdata sets are described to demonstrate howmuch the geospatial data models differ structu-rally and semantically. Then the architecture andmodelling concepts of the database integrationare explained; they provide an organisational fra-mework for the approaches of geospatial dataintegration given in section 4.

Figure 1: System overview

55

3.1 Data setsThe vector data sets used in this project includethe German topographic data set (ATKISDLMBasis), the geological map (GK) and the soilscience map (BK), all at a scale of 1:25000.Simple superimposition of different data setsalready reveals some differences. These differen-ces can be explained by looking at the creationthe maps. For ATKIS the topography is the mainthematic focus, for the geoscientific maps it iseither geology or soil science. Thus, these mapshave been produced using the result of geologi-cal drilling, and according to this punctual infor-mation, area objects have been derived usinginterpolation methods based on geoscientificmodels. They are, however, related to theunderlying topography. The connection bet-ween the data sets has been achieved by usingthe topographic information together with thegeoscientific results at the point of time, whenthe geological or soil science information wascollected. The selection and integration ofobjects from one data set to another one wasperformed manually and in most of the casesthe objects have been generalised by the geo-scientist. While the geological content of thesedata sets keeps its topicality for decades, thetopographic information in these maps doesnot: In general, topographic updates are not

integrated unless new geological informationhas to be inserted in these data sets. The geo-scientific maps have been digitised to use thebenefits of digital data sets, but due to the digi-talisation even more discrepancies occurred.Another problem which amplifies the deviationsof the geometry is the case of different datamodels. Geological and soil science maps aresingle-layered data sets which consist only ofpolygons with attribute tables for the represen-tation of thematic and topographic content,while ATKIS has a multi-layered data structurewith objects of all geometric types, namelypoints, lines and polygons, equally with attribu-te tables. In addition to the described vectordata, raster data sets are used to enable objectrecognition while exploiting the prior ATKISknowledge. The raster data sets are aerial ima-ges or high resolution satellite images, whichinclude an infrared channel.

3.2 Architecture and concepts of integrationAs the previous section has shown, the variousgeospatial data sets differ significantly due tothe various objectives of their acquisition. Inorder to integrate the corresponding databa-ses we have chosen the architectural paradigmof federation (Conrad, 1997), as it gives a close

Figure 2: Architecture of a federated database.

56

coupling at the same time and keeps the data-bases autonomous. Hereby, the matching andextraction processes are given an integratedview to the different databases via a globaldatabase schema (global applications). Never-theless, particular applications (like import andexport processes) may still access the databa-ses locally as shown in Fig. 2.

The federation service requires an »integrationdatabase« (cf. section 3.2.4) on its own tomaintain imports and descriptions of the invol-ved data sets (component databases), and toincorporate qualified links between object asthe result of the matching process as well asfurther findings such as geometric adjustedand new extracted objects.

3.2.1 Schema adaptationTo make the structurally different data setsaccessible to the federation service a genericbut flexible export schema was designed basedon experiences with geospatial data sets con-taining topographic objects with respect toobject-relational databases (Kleiner, 2000).The schema contains all objects, object classes,attribute types and attribute values, each ofthem in one entity type (or table in the relatio-nal DBMS). Fig. 3 shows the schema for topo-graphic data (ATKIS), the geoscientific datasets get isomorphic export views; in moredetail they have application-specific attributetypes and object classes according to their ownrepresentation model.

A geoobject of entity type ATKIS_Objects, e.g. aroad, has several entries of type ATKIS_ Attri-butes, namely (attribute, value)-pairs like e.g.(width, 10 meters). The corresponding type ofthe attributes or the classification of the geoob-jects can be found in the collections ATKIS_AttributeTypes and ATKIS_ObjectClasses.

3.2.2 Object linkingGiven the structural adaptation of the differentdata sets, the federated database can be ena-bled to incorporate correspondences throughso called links. Linking objects, however,should not only involves simple one-to-one-relationships, as real-world objects are repres-ented differently with respect to differentmaps. The federation service has to cope withmore complex correspondences namely one-to-many- and even many-to-many-relations-hips as shown in Fig. 4, which represents diffe-rent partitions of a real world object in twomaps. This task is accomplished with a flexibleschema, that integrates these general corre-spondences as attributed one-to-one links bet-ween aggregated objects. Fig. 4 shows aninstance of three and two objects, respectively,e.g. a section of a water body segmented intwo different ways, whose aggregations(denoted by dashed lines) are linked.

Figure 3: Export schema for the topographic map (ATKIS).

57

3.2.3 Attribute harmonisation and semanticselectionIn order to provide the applications with amodel independent and uniform method toaccess certain objects with respect to thematicattributes, a mechanism for the semantic des-cription of geoobjects was developed, to cha-racterise comparable object sets for the mat-ching process and to characterise object selec-tion for the extraction process. To fulfil theserequirements, the architecture of federateddatabases had to be expanded to unify thehandling of semantic descriptions. Fig. 5shows two simplified semantic selections of

topographic objects, namely of open landsca-pe and a partitioning network.

Semantic object selections are defined in thefollowing three stages: Coarse semantic classi-fication is achieved through the references toobject classes given by the export views. Fig. 5depicts some object classes of the topographicmap, e.g. farmland and roads. Next, a moreprecise characterisation is provided throughthe specification of object attributes, i.e. thecoarse selection via object classes is restrictedby attribute conditions. For instance, roadobjects appear as both one-dimensional and

Figure 4: Realisation of a many-to-many-relationship as a link between object aggregations.

Figure 5: Semantic selections for regions and networks.

58

two-dimensional objects due to acquisitionrules. In order to build a partitioning networkonly the one-dimensional road objects are nee-ded. Finally, fine object classes are merged toclass sets, which provide semantic selectionsfor the global applications, independent of theoriginal data set’s semantic specifications. Nextto the structural unification through exportviews attribute harmonisation is achieved byconnecting two conforming semantic selec-tions of two different data sets (e.g. waterbodies both in the topographic and the geolo-gical map). It is necessary to provide this sem-antic description for any representation modelonly once, independent of the quantity ofinstances of this particular model (componentdatabases).

3.2.4 Integrated schemaFig. 6 summarises the schema architecture ofthe integration database. The componentdatabases are both original involved geospati-al data sets based on the previously describedexport views, and the term »Objects« stands

for all objects of the integration database, i.e.adjusted and extracted geometric objects. Thedifferent parts of Fig. 6 show that the federa-tion service is supported with respect to thefollowing tasks for- the model description, characterisation of

object classes and attribute types of a cer-tain data set model

- the registration, registering the componentdatabases

- the semantic selection as described in theprevious section

- the application control, which stores metadata about extraction and matching pro-cesses, in particular about the used seman-tic selections, and links between the invol-ved component databases

- the linking objects from different data sets(object linking, cf. Section 3.2.2)

4. Methods of data integrationIn this section the methodologies of the vec-tor/vector and the raster/vector data integra-tion are described. First, the integration of

Figure 6: Overview of the integrated schema.

59

heterogeneous vector data sets which havebeen acquired for different purposes and withunequal updating strategies is presented basedon a component based strategy. Subsequently,the integration of raster and vector data ishighlighted with the example of the extractionof field boundaries and wind erosion obstaclesfrom imagery exploiting prior GIS knowledge.

4.1 Integration of vector dataAt the beginning of the integration process thesemantic content of all data sets was compa-red. According to this step, certain selectiongroups were built up for each data set (e.g.water area). This selection is mandatory toavoid comparing »apples and oranges« andhas to be the first step to ensure a successfulintegration. An area-based matching process isused for the creation of links between objectcandidates. These links are stored in the fede-rated database using a XML-schema, followedby an alignment process which reduces geo-metric discrepancies to a minimum to ensuresatisfying results in the subsequent intersectionprocess, but will still be capable of decidingbetween geometric discrepancies based onmap creation or topographic changes whichoccurred during the different times of acquisi-tion. A rule-based evaluation of the intersec-tion results is used for change detection.

4.1.1 Revelation of links between correspon-ding objectsVarious data sets have different forms of repre-sentations for certain topographic objects (e.g.rivers), the decision which kind of representa-tion to take often depends on specific attribu-tes, e.g. in (ATKIS DLMBasis, cf. Section 3) thewidth of the river is used for this decision, thin-ner than 12 meters – polyline, wider than 12meters polygon. Due to the fact that there aredifferent thresholds for each data set, thesedifferences have to be resolved using harmoni-sation strategies. To ensure a suitable result inthe revelation of links, line objects have to betransformed into polygons by applying a bufferalgorithm using the width attribute.Another problem is the representation of grou-ped objects in different maps. For a group ofwater objects, e.g. a group of ponds, therepresentation in the different data sets couldeither be a group of objects with the same ora different number of objects, or even a singlegeneralised object (see Fig. 7). Finally, alsoobjects can be present in one data set and notrepresented in the other. All these considera-tions lead to the following relation cardinalitiesthat have to be integrated: 1:0, 1:1, 1:n, andn:m. After the corresponding relations havebeen identified, each selection set will beaggregated, so they can be handled as 1:1relations, so called relation-sets (Goesseln andSester, 2004).

Figure 7: Different representations - ATKIS (solid line), GK (dotted line).

60

These relation-sets will be visualised to theoperator - using a GUI based application - en-abling a manual correction of the derived links.With this software each relation-set can beinspected and edited, to check whether theautomated process has failed to build up thesuitable correspondences between the selec-ted data sets. Because of to the fact that theobjects from all three data sets are representa-tions of the same real world objects, they showapparent resemblance in shape and position.Nevertheless the alignment of the geometriesis required after the evaluation of the matchingresults. As it will be described later, there aredifferent geometric alignment method requi-red for covering all alignment tasks. Thereforethe technique offering the most suitable resultcan be selected for every single relation-set.

4.1.2 Geometric Alignment of corresponding objectsObjects which have been considered as a mat-ching pair could be investigated for changedetection using intersection. At this stage thementioned differences will produce more pro-blems which are visible as discrepancies in posi-tion, scale and shape. These discrepancies willlead to unsatisfying results in the evaluation ofthe resulting elements almost and this wouldevoke an immoderate estimation of the areainvestigated as change of topographic content.Therefore a geometric adaptation will be app-lied, leading to a better geometric correspon-dence of the objects. For these adaptation pro-cesses thresholds are required which allow thereduction of discrepancies which are based onmap creation, but will not cover the changeswhich happened to real world objects betweenthe different times of data acquisition.

Iterative closest point (ICP)The iterative closest point algorithm (ICP) deve-loped by (Besl and McKay, 1992) has beenimplemented to achieve the best fitting bet-ween the objects from ATKIS and the geo-scientific elements using a rigid 7 parametertransformation. The selection of a suitablealgorithm used for ICP is depending on the

alignment to be performed, in this case theproblem is reduced to a 2D problem requiringfour parameters (position, scale and orienta-tion) an solved using a Helmert-transforma-tion. These calculations are repeated iterativelyand will be evaluated after each calculation;the iteration stops when no more variation inthe four parameters occur. At the end of theprocess the best fit between the objects usingthe given transformation is achieved. Evalu-ating the transformation parameters allows forclassifying and characterising the quality of thematching: in the ideal case, the scale parame-ter should be close to 1 and rotation and trans-lation should be close to 0. Assuming, that theregistration of the data sets is good, these fourparameters exactly meet the reasons for theintegration of analogue produced data sets,that have been created by manual copying ofprinted maps. Therefore a greater scale factorcan be an indicator for differences betweentwo objects that are not based on map crea-tion, but on a change on the real world object,that occurred between the different times ofdata acquisition (Goesseln and Sester, 2004).At the end of the process the best fit betweenthe objects using the given transformation isachieved. The result of this transformation isstored as a set of shifting vectors, which arerequired in a subsequent step in which theneighbourhood of the transformed objectswill be aligned. This step will be describedlater on (cf. Section 4.1.3). The application ofthe iterative adaptation using the ICP appro-ach based on Helmert-transformation sho-wed very good results and revealed the possi-bility of reducing the amount of objectswhich have to be evaluated manually.However there are some situations where thisapproach does not generate sufficient results(e.g. objects which cover several map-sheetsor at least touch the map boundaries).

Dual interval alignment (DIA)The DIA approach has been implemented,enabling the alignment of local discrepanciesof corresponding geometries by calculating thetransformation of single vertices, based on theideas of Kohonen (1997), however this appro-

61

ach handles each vector separately. Corres-ponding objects which have been assigned asrepresentations of the same real world objectthrough the matching process are investigatedbased on their vertices. For every point in oneobject the nearest neighbour in the correspon-ding partner object is determined using the cri-terion of proximity. The conformation approachevaluates the distance between these coordina-tes, based on an interval which is predeterminedby the human operator. This threshold definesthe largest distance – representing a change ingeometry – which will be suitable for the candi-date data set. Distances exceeding this thres-hold implicate a topographic difference whichhas to be investigated during field-work.

As it can be seen in Fig. 8, for each point (PC)from object C and the corresponding point (PR)of the linked object R, the point transformationis calculated based on the euclidean distance (d)between these points. The new coordinates aredetermined taking interval ranges a and b intoaccount. Points within the first distance interval(0<d<a) are aligned to a single point, a distancefalling into the second interval (a<d<b) will leadto an approximation of the selected points.Points with a distance beyond b will not beadapted (see Eq. 1).It seems to be a paradox that the completealignment must not be the perfect result. Theintegration of data sets which cover the samearea, which are based on the same method ofrepresentation and are acquired at nearly thesame point of time, can be performed by usingan alignment strategy with elimination of all dif-

ferences. While integrating data sets whichhave been acquired at different points in time itis obvious that a certain amount of change totopography, built-up area and/or vegetation hasoccurred. Therefore an alignment threshold isrequired which allows the operator to decidebetween errors due to map creation or real-world changes. The introduction of a secondthreshold follows the idea of »fuzzy« logic andensures that there are no points of hard discon-tinuities at the geometry of aligned objects.The integration of a weight p (see Eq. 1) to thealignment process does not only take differentaccuracies of geometries into account, butopens this approach to a much wider range ofconflation tasks. E.g. in a project where onedata set is handled as a reference data set andmust not be changed (weight set to 1). In othercases, when two data sets have to be alignedand no single data set can be selected as refe-rence the alignment is performed using thecommon idea of conflation by aligning two datasets to a new improved data set.

(1)

Calculating the shift distance based on the nea-rest neighbour enables very good alignment,but can result in topologic errors. This requiresthe integration of an additional criteria which is

Figure 8: Application of DIA for the partial alignment of object geometries (schematic).

62

vertex orientation. Therefore the orientation ofthe polygon segments will be calculated for allcorresponding objects. If a point and its corre-sponding partner are selected using the distan-ce criterion, the direction to each correspondingsuccessor will be calculated. If the differencebetween these directions exceeds a given thres-hold, the points must not be aligned due to theassumption that they do not represent the same»side« of the real world.Comparing the adaptation approaches ICP andDIA, each is suitable for a different kind ofobjects in this project. ICP matches the idea,that the majority of the geometric discrepanciesis caused in the way the data sets have beencreated by integrating topographic elementsthrough manual copying. The resulting parame-ters can be used for the investigation and theevaluation of the influences which were respon-sible for the geometric discrepancies: An objectwhich can be aligned by just using translationswith a small change in scale can be judged asminor error based on manual copying. A largerscale factor can reveal topographic changes onthe real world objects, which have to be investi-gated by a human operator.The transformation, the implemented ICP algo-rithm is based on, is very fast and reliable, butdepending on the chosen transformation algo-rithm it does not give satisfying results for larger,irregular shaped objects like rivers, or objectsthat have been changed during different peri-ods of time and therefore only match partially.But as good as the alignment results of DIA are,it is much more time consuming and susceptibleto errors. The combination of both approachesdelivered very good results, offering the possibi-lity to assess the geometric discrepancies by eva-luating the resulting ICP-parameters, and alig-ning large object groups or partially matchingobjects using DIA. The automatic guided deci-sion between these methods has not been com-pleted yet. So far both methods will be appliedfor every relation-set and the most suitableresult will be chosen by comparing the resultswith certain geometric operators (e.g. anglehistogram, symmetric difference).

4.1.3 Neighbourhood adaptation using rub-ber-sheetingThe individual alignment of selected objectswould result in gaps, overlaps or inconsistenciesconcerning the rest of the data set, so that theneighbourhood of the aligned objects must betransformed equivalent. To ensure an overallalignment, the results which originate from theindividual alignment processes are stored as acollection of displacement vectors. All vectorswill build up a vector field which is the basis ofthe neighbourhood adaptation ensuring ahomogeneous data set. Using a distance weigh-tened interpolation the rubber-sheeting methodcalculates a new transformation target for everypoint in the data set based on the vectors deri-ved from the alignment.This strategy has to be carefully adapted forevery adaptation process regarding to the useddata set. Different data sets require differentconstraints the rubber-sheeting algorithm mustbe able to take into account. These constraintscan be e.g. points or areas which must not bechanged, like fixed points or areas which havebeen updated manually in advance, or objectsof higher category.

4.2 Integration of raster and vector dataThe integration of raster and vector data is high-lighted by means of the extraction of fieldboundaries and wind erosion obstacles fromimagery exploiting prior GIS knowledge. First,the integrated modelling and the derived strate-gy are described, followed by the presentationof fully automatic methods to extract the fieldboundaries and wind erosion obstacles.

4.2.1 Model and strategyThe semantic model comprises the integrationof raster data (imagery) and vector data (GISdata) as starting point for the object extraction,as described in detail in Butenuth (2004). Themodel is differentiated in an object layer, a geo-metric and material part, as well as an imagelayer (cf. Fig. 9). It is based on the assumption,that the used images include an infrared (IR)channel and are generated in summer, when

63

the vegetation is in an advanced period ofgrowth. The use of vector data as prior know-ledge plays an important role, which is repres-ented in the semantic model with an additionalGIS-layer (ATKIS DLMBasis, cf. Section 3): Fieldboundaries and wind erosion obstacles areexclusively located in the open landscape, thus,further investigations are focussed to this area.Additionally, the objects road, river and railwayare introduced in the semantic model as fieldboundaries with a direct relation from the GIS-layer to the real world (i.e. a road is a field boun-dary). Of course, the underlying assumption isbased on correct GIS-objects. Modelling of theGIS-objects in the geometry and material layertogether with the image layer is not of interest,because they do not have to be extracted fromthe imagery; thus, the corresponding parts arerepresented with dashed lines in Fig. 9.Nevertheless, additionally extracted objectswhich are not yet included in the GIS databasecan be introduced at any time.The field is divided in the semantic model intofield boundary and field area in order to allowfor different modelling in the layers. The fieldboundary is a 2D elongated vegetation bounda-ry, which is formed as a straight line or edge inthe image. The field area is a 2D vegetationregion, which is a homogeneous region with a

high NDVI (Normalised Difference VegetationIndex) value in the colour infrared (CIR) image.The wind erosion obstacle is divided in hedgeand tree row due to different available informa-tion from the GIS-layer, which is partially storedin the database. The wind erosion obstacles arenot only described by their direct appearance ingeometry and material, but also through thefact, that due to their height (3D object) there isa 2D elongated shadow region next to theobject and in a known direction. In particular,the relationships between the objects to beextracted are of interest leading to connectionswithin the layers: One object can be part ofanother one or be parallel and nearby, andtogether they form a context network in the realworld. For instance, wind erosion obstacles arenot located in the middle of a field because ofdisadvantageous cultivation conditions, butsolely on the field boundaries.

The strategy derived from the modelled charac-teristics of the field boundaries and wind ero-sion obstacles aims at realising an automaticprocessing flow. Imagery and GIS-data are theinput data to initialise the process: First, fieldboundaries and wind erosion obstacles areextracted separately. At the end, a combinedevaluation of the preliminary results is advanta-

Figure 9: Semantic Model.

geous due to the modelled geometrical andthematic similarities of the objects of interestgetting a refined and integrated solution.The strategy extracting the field boundariesstarts with the derivation of the open landscapefrom the GIS data. In addition, within the openlandscape, regions of interest are selectedusing the roads, rivers and railways as border-lines (cf. Section 4.2.2.). Consequently, theborderlines of the regions of interest are fieldboundaries, which are already fixed. In eachregion of interest a segmentation is carried outin a coarse scale ignoring small disturbingstructures and thus exploiting the relativehomogeneity of each field. The aim is to obtaina topologically correct result, even though thegeometrical correctness may not be very high.Afterwards, network snakes are used to impro-ve the preliminary field boundaries.The strategy extracting the wind erosion obsta-cles starts again within the open landscape dueto the modelled characteristics. Search bufferscan be defined around the GIS objects roads,rivers and railways, because tree rows or hedgesare often located alongside these objects, andhave to be verified using the imagery. In contra-ry, there is no prior information about the loca-

tion of all other wind erosion obstacles, whichcan lie anywhere within the open landscape. Inaddition to the modelled material characteri-stics, the geometrical part such as the straight-ness or minimum length has to be considered.Finally, the combined evaluation of the prelimi-nary results identifies discrepancies between thefield boundaries and wind erosion obstacles. Forexample, extracted wind erosion obstacleswithout a corresponding extracted field boun-dary have to be checked, whether nearby a fieldboundary is missed, or whether the extractionof the wind erosion obstacle is wrong.Consequently, the combined evaluation andrefined extraction process leads to a consistentand integrative final result.

4.2.2 Preparation of GIS dataAs described in the previous section the regionsof interest are primarily derived from roads,rivers and railways according to the GIS data, asfar as these objects are located in the openlandscape. The fact that these network genera-ting objects imply a segmentation of the openlandscape is used as starting point, as allnecessary borderlines are already present in

Figure 10: Generation of regions of interest (a) and adjustment of tree rows and hedges (b).

64

65

this data set – however, this segmentation istoo extensive (e.g. because of administrativereasons). In order to detect only the borderli-nes concerning regions of interest, a topologi-cal data model is generated (Egenhofer et al.,1989), consisting of a embedded graph struc-ture, which contains the open landscaperegions and the network generating objects.This graph based data model represents boun-daries of area objects and one-dimensionalobjects as edges and therefore allows decidingif certain segmentations are a result of theseparating network. The removal of all edges,which are not caused by the separating net-work, implies the merging of the adjacentregions and finally results in the generation ofthe regions of interest (Fig. 10 a).

Furthermore, the initial topological data model(i.e. before edge removal) is used to prepare thetree row and hedge objects used in the winderosion obstacle extraction process by extendingthem to the next respective edge (Fig. 10 b, darklines). This process of alignment is similar totopological error correction of inaccurately pro-duced maps (Ubeda et al., 1997), whereas theseobjects are not inaccurately acquired but treerows or hedges often end with a short distanceto roads, rivers or railways.

4.2.3 Extraction of field boundariesThe extraction of field boundaries starts with asegmentation within each region of interestexploiting the modelled similar characteristicsof each field. The border area of each region ismasked out due to disturbing heterogeneities,which are typical for fields and deteriorate thesubsequent steps. A multi-channel region gro-wing is carried out using the RGB- and IR-channels of the images with a resolution offew meters. The four channels give rise to a 4-dimensional feature vector: Neighbouringpixels are aggregated into the same fieldregion, if the difference of their feature vectorsdoes not exceed a predefined threshold. Inconcert with the modelled constraints, theresulting field regions must have a minimumsize. The case of identical vegetation of neigh-

bouring fields may lead to missing boundaries.In order to overcome this problem, the stan-dard deviation of the grey values in the imagewithin a quadratic mask is computed, i.e. highvalues typically belong to field boundaries.Extracted lines from the standard deviationimage within sufficiently large field regions areevaluated concerning length and straightness.Positively evaluated lines are used to split theinitially generated field regions.The result of the segmentation leads to topo-logically correct but geometrical inaccurateresults. Network snakes are used to improvethe geometrical correctness of the preliminaryfield boundaries while maintaining the topolo-gical constraints. Snakes were originally intro-duced in Kass et al. (1988) as a mid-level imageanalysis algorithm, which combines geometricand/or topologic constraints with the extrac-tion of low-level features from images. A tradi-tional snake is a parametric curve (Kass et al.,1988; Butenuth and Heipke, 2005)

(2)

where s is the arc length, t the time, and x andy are the image coordinates of the 2D-curve.The image energy is defined as

(3)

where I represents the image, | I(v(s,t))| is thenorm of the gradient magnitude of the imageat the coordinates x(s) and y(s) and |v| is thetotal length of v. In practice, the image energyEI(v) is computed by integrating the values | I(v(s,t))| in precomputed gradient magnitudeimages along the line segments that connectthe polygon vertices. The internal energy isdefined as

(4)

where the function α(s) controls the first-orderterm of the internal energy: the elasticity. Largevalues of α(s) let the contour become verystraight between two points. The function β(s)controls the second-order term: the rigidity.Large values of β(s) let the contour become

66

smooth, small values allow the generation ofcorners. α(s) and β(s) need to be predefinedbased on experimental data and experience.The total energy of the snake, to be minimised,is defined as Esnake = Ev(s,t) + EI(v). A minimum ofthe total energy can be derived by embeddingthe curve in a virtual viscous medium solvingthe equation

(5)

where γ is the viscosity of the medium and κ isthe weight between internal and image ener-gy. After substituting of

(6)

in equation 5, a solution for the contour attime t depending on time t-1can be computed:

, (7)

(I: identity matrix)Vs,t stands for either X or Y, the vectors of thex and y coordinates of the contour. A is a pen-tadiagonal matrix, which depends only on thefunctions α(s) and β(s). A main problem of snakes is the necessity tohave an initialisation close to the true solution.Methods to increase the capture range of theimage forces are not useful in our case, becau-se there are lots of disturbing structures withinthe fields, which can cause an unwanted imageenergy and therefore a wrong result. Thus, onlythe local image information is of interest. Asdescribed above, the result of the segmentationis used to initialise the processing.In addition to the good initialisation the deri-vation of the topology of the initial contours ismost important. The global framework of theaccomplished segmentation gives rise to a net-work of the preliminary field boundaries:Enhancing traditional snakes, network snakesare linked to each other in the nodal pointsand thus interact during processing (cf. Fig. 13b). Similarly, the connection of the end pointsof the contours to the borders of the region ofinterest must be taken into account: In con-trast to the nodal points, a movement of the

end points is only allowed along the borders ofthe regions of interest. These topological cons-traints are considered, when filling the matrixA (see equation 7) with the functions α(s) andβ(s), which in our case are taken to be con-stant.

4.2.4 Extraction of wind erosion obstaclesThe extraction of wind erosion obstacles isconcentrated to the open landscape, as poin-ted out in the semantic model. No other GISdata (road, river, railway) is used, i.e. no priorgeometric information reduces the searcharea, in order to acquire all tree rows and hed-ges within the open landscape. A texture seg-mentation is accomplished in the CIR imageswith a resolution of few meters yielding thetexture classes tree/hedge, settlement area andagricultural area, for details concerning theapproach cf. Gimel´farb (1996). The trainingimages are generated manually by a humanoperator. The texture class of interest tree/hedge is, as expected, fragmented and notcomplete. Therefore, the elongated and smallregions of the class are vectorised. Startingpoint are the left and right boundaries of theseregions: centrelines are then computed, whichare evaluated concerning length and straight-ness. Currently, the third dimension, as descri-bed in the semantic model, is not used toextract the wind erosion obstacles due to amissing digital surface model.

5. ResultsIn this section some results of test areas in nor-thern Germany are presented. First, results areshown to reveal the possibility to perform thealignment and change detection for the upda-ting of vector data sets with a high degree ofautomation. Second, results of the extractionof field boundaries and wind erosion obstaclesare highlighted to demonstrate the capabilityof the described methods.

5.1 Results of the vector data integrationIn Fig. 11 the results of the different alignment

67

methods can be seen. The ICP algorithm usingan iterative four-parameter transformation isvery suitable for the alignment of objectswhich already have a similar geometry. Thealignment parameters which are the results ofthe ICP algorithm can give a first hint whetherthe geometric discrepancies are due to mapcreation and acquisition methods (a., d.) or tochanges which occurred to the real worldobject (c.). The resulting scale factor which wascalculated for the alignment of object c. wasrated as too large and therefore no alignmentwas performed. Of course changes of thetopography can not be discovered by simpleevaluation of these parameters. For object b.the algorithm achieved a best fit with fourparameters below certain thresholds, but theremaining differences between the geometriesstill have to be corrected.

The DIA implementation showed very goodresults to compensate local discrepancieswhich can not be corrected using the four-parameter-ICP, as it aims for the best align-ment of the whole object. There is no singlefour-parameter transformation which is capa-ble of adjusting large extended natural objectslike rivers object to their corresponding part-ners in ATKIS, so that parts e, f and h would beproperly aligned. The results does exhibit somesmall gaps between the geometries (see e.g.area (g)): this is due to the fact that the DIA

algorithm in the current version is only wor-king only with existing vertices without inser-ting additional ones.

In order to identify possible changes betweenthe objects in the different representations,after the alignment process has been comple-ted an intersection of corresponding objects isused for the change detection. The intersec-tion is performed on all types of topographicelements which are represented in the datasets. The results of the intersection process willbe evaluated and according to their semanticattribution sorted into three different classes.

- Type I: Segment has corresponding semanticattribute in both data-sets, no adaptationrequired,

- Type II: Segment has different semantic attri-butes, and a suitable information can bederived from the reference data set and thecandidate data set will be updated.

- Type III: Segment has different semanticattributes, but a suitable information cannot be derived from the reference data set.Manual interaction is required.

Type II will also be assigned to objects whichare represented in the reference, but not thecandidate data-set, this is the result of diffe-rent updating periods between the referenceand the candidate data set, which result in

Figure 11: Result of the approach, GK 25 (thin, dottedline) aligned on the reference German digital topographicmap (ATKIS, dark lines).

68

outdated objects. While Type I and II requireonly geometric corrections or attribute adapta-tion and can be handled automatically, Type IIIneeds more of the operators attention.Depending on the size and the shape of a TypeIII segment and by using a user-defined thres-hold, these segments can be filtered, removedand the remaining gap can be corrected auto-matically, this will avoid the integration of sliverpolygons and segments which are only theresults of geometric discrepancies.

5.2 Results of the raster and vector data integration

5.2.1 Results of the extraction of field boundariesResults of the proposed strategy to extractfield boundaries are presented in this section.The result of the first step, the segmentation, isshown in Fig. 12: The boundaries of the regionsof interest are depicted in black, the preliminaryfield boundaries are depicted in white. Com-pared to reference data, the completeness ofthe segmentation within a test area of 25 km2

(Lower Saxony, North of Germany) is 73 %, thecorrectness is 82 % and the rms error computed

by considering the horizontal derivation bet-ween extracted and reference result is 5.8 m or3 pixels. The quality of the results is promising,but as expected the geometrical correctness isnot very high.One region of interest is selected to demonstra-te the methodology of the network snakes (cf.Fig. 13 a-d): The initialisation of the snake –equivalent to the result of the segmentation – isshown in the first figure. The topology is poin-ted out in Fig. 13 b): The individual snakes for-ming the network are linked to each other inthe nodal point (black), and the end points(black with white hole) are linked to the boun-dary of the region of interest. The movement ofthe snake superimposed to the standard devia-tion image is shown in Fig. 13 c), the final resultsuperimposed to the real image in Fig. 13 d).The example demonstrates, that network sna-kes are a useful tool to improve the geometricalcorrectness of topologically correct but geome-trically inaccurate results.

5.2.2 Results of the extraction of wind erosionobstaclesThe result of the texture segmentation is pres-ented in Fig. 14: The class tree/hedge is depic-

Figure 12: Result of the segmentation.

69

ted in white, the class agricultural area in lightgrey and the class settlement area in dark grey.The parts of the image, which do not belongto the open landscape exploiting the prior GISknowledge, are marked in black. The fragmen-ted class tree/hedge is vectorised yielding thewind erosion obstacles, as depicted in Fig. 15in white. The texture segmentation works well,but an additional digital surface model is nee-ded to improve and stabilise the results.

6. ConclusionsThe geospatial federated database has provi-ded the expected access to the involved datasets and to the results of the matching andextraction processes. It provides a basis not

only for querying linked objects but also forupdate propagation. Utilised appropriate datastructures like topological data models offerfurther approaches to assure topological consi-stence during geometric alignment and toaccomplish a structural graph based matching.The geometric comparison and the derivationof object links, together with the ICP and DIAalignment followed by rubber-sheeting andthe evaluation process show good results. Sofar this strategy was used with one data set asreference which remains unchanged, while asecond data set is adjusted, but it can also beadapted to other vector based conflation tasksrequiring an intermediate geometry. Depen-ding on the selected thresholds large discre-pancies of the shape boundaries can consider

Figure 13a

Figure 13 d

Figure 13b

Figure 13c

Figure 13: Results of the use of net-work snakes: a) initialisation, b) buil-ding the topology, c) initialisation(white) and movement of the snake(black), d) extracted field boundaries.

70

as outliers and can be treated accordingly inthe subsequent overlay and analysis step. Whilematching can be performed automatically, thereare still some steps during geometric alignmentand change detection which require the deci-sion of a human operator, but the high degreeof automation reduces the manual process con-siderable. Future work will concentrate on deve-loping a strategy to also automate these pro-cesses. Especially the selection of the appropria-te alignment method and the correspondingthresholds will be enhanced.The method integrating raster and vector databy means of the extraction of field boundariesand wind erosion obstacles from imageryexploiting prior GIS knowledge has also cho-sen promising results. Concerning the extrac-tion of field boundaries the basic step of thestrategy, the segmentation, could be enhancedby using an additional texture channel to pre-vent wrong field boundaries. They occur, whenthere are large heterogeneities within a field.The control of the network snakes could beimproved by selecting variable values when fil-ling the matrix A to increase the geometricalcorrectness, the use of network snakes providea topologically consistent solution. Regardingthe extraction of wind erosion obstacles, initialresults show the potential, but also the limita-tions of the current approach. The use of adigital surface model will probably be very hel-pful to achieve better results. Finally, the com-

bined evaluation of the field boundaries andwind erosion obstacles will identify discrepan-cies between the different extracted objects,resulting in a more consistent and integrativefinal result.

ReferencesBaltsavias, E.P. (2004): Object Extraction andRevision by Image Analysis Using ExistingGeodata and Knowledge: Current Status andSteps towards Operational Systems. ISPRSJournal of Photogrammetry and RemoteSensing 58 (3-4), 129-151.

Batini, C., Lenzerini, M., Navathe, S. B. (1986):A Comparative Analysis of Methodologies forDatabase Schema Integration. ACM Comput.Surv. 18(4), 323-364.

Beeri, C., Doytsher, Y., Kanza, Y., Safra, E., Sagiv,Y. (2005): Finding Corresponding Objects whenIntegrating Several Geo-Spatial Datasets. Proc.13th ACM International Symposium onAdvances in Geographic Information Systems,Bremen, Germany, 4-5 November 2005, 87-96.

Besl, P., McKay, N. (1992): A Method forRegistration of 3-D Shapes, IEEE Transactionson Pattern Analysis and Machine Intelligence(Special issue on interpretation of 3-D scenes -part II) 14 (2), 239-256.

Figure 14: Result of the texture segmentation. Figure 15: Result of the extracted wind erosion obstacles.

71

Butenuth, M. (2004): Modelling the Extractionof Field Boundaries and Wind ErosionObstacles from Aerial Imagery. InternationalArchives of the Photogrammetry, RemoteSensing and Spatial Information SciencesXXXV (Part B4), 1065-1070.

Butenuth, M., Heipke, C. (2005): NetworkSnakes-Supported Extraction of Field Boun-daries from Imagery. In: Kropatsch, Sablatnig,Hanbury (Eds), 27th DAGM Symposium, Wien,Österreich, Springer LNCS 3663, 417-424.

Conrad, S. (1997): Föderierte Datenbanksys-teme, Springer-Verlag, Berlin.

Devogele, T., Parent, C., Spaccapietra, S.(1998): On spatial database integration, Inter-national Journal of Geographical InformationScience, 12:4, 335-352.

Doytsher, Y. (2000): A rubber sheeting algo-rithm for non-rectangular maps, Computer &Geosciences, 26 (9-10), 1001-1010.

Egenhofer, M.J., Frank, A.U., Jackson, J.P.(1989): A Topological Data Model for SpatialDatabases, Lecture Notes in Computer Science409, 271-286.

Gimel´farb, G.L. (1996): Texture modelling bymultiple pairwise pixel interactions. IEEETransactions on Pattern Analysis and MachineIntelligence 18 (11), 1110-1114.

Goesseln, G. v., Sester, M. (2004): Integrationof geoscientific data sets and the german digi-tal map using a matching approach. Interna-tional Archives of Photogrammetry and Re-mote Sensing 35 (Part 4B), 1249-1254.

Hettwer, J., Benning, W. (2000): Nachbarschafts-treue Koordinatenberechnung in der Kartenho-mogenisierung, Allg. Verm. Nachr. 107, 194-197.

Hill, D. A. and Leckie, D. G. (Eds.) (1999):International forum: Automated interpretationof high spatial resolution digital imagery forforestry. February 10-12, 1998, Natural

Resources Canada, Canadian Forest Service,Pacific Forestry Centre, Victoria, BritishColumbia.

Kass, M., Witkin, A., Terzopoulus, D. (1988):Snakes: Active Contour Models. InternationalJournal of Computer Vision 1, 321-331.

Kohonen, T. (1997); Self-Organizing Maps.Springer.

Kleiner, C., Lipeck, U., Falke, S. (2000): Objekt-Relationale Datenbanken zur Verwaltung vonATKIS-Daten. In: Bill, R., Schmidt, F.: ATKIS -Stand und Fortführung, Verlag KonradWittwer, Stuttgart, 169-177.

Laurini, R. (1998): Spatial multi-database topo-logical continuity and indexing: A step towardsseamless GIS data interoperability, Internatio-nal Journal Geographical Information Science,12:4, 373-402.

Löcherbach, T. (1998): Fusing Raster- andVector-Data with Applications to Land-UseMapping. Inaugural-Dissertation der HohenLandwirtschaftlichen Fakultät der Uni-versität Bonn.

Mantel, D., Lipeck, U. W. (2004): Datenbankge-stütztes Matching von Kartenobjekten. In:Arbeitsgruppe Automation in der Kartographie- Tagung Erfurt 2003, BKG, Frankfurt, 145-153.

Mayer, H. (1999): Automatic Object Extractionfrom Aerial Imagery - A Survey Focusing onBuildings. Computer Vision and ImageUnderstanding 74 (2), 138-149.

Rigeaux, P., Scholl, M., Voisard, A. (2002):Spatial Databases with Application to GIS,Morgan Kaufman Publishers.

Sattler, K.-U., Conrad, S., Saake, G. (2000):Adding Conflict Resolution Features to a QueryLanguage for Database Federations, 41-52.

Sester, M., Hild, H. & Fritsch, D. (1998):Definition of Ground-Control Features for

72

Image Registration using GIS-Data. In: Schenk,T. & Habib, A. (Eds.), IAPRS 32/3, ISPRSCommission III Symposium on ObjectRecognition and Scene Classification fromMultispectral and Multisensor Pixels,Columbus/Ohio, USA, 537-543.

Torre, M., Radeva, P. (2000): Agricultural FieldExtraction from Aerial Images Using a RegionCompetition Algorithm. International Archivesof Photogrammetry and Remote Sensing XXXI-II (Part B2), 889-896.

Yuan, T., Tao, C. (1999): Development of con-flation components. In: Li, B., et al. (Eds.),Geoinformatics and Socioinformatics – TheProceedings of Geoinformatics ´99 Conference,Ann Arbor, USA, 19-21 June 1999, 1-13.

Walter, V. & Fritsch, D. (1999): MatchingSpatial Data sets: a Statistical Approach, Inter-national Journal of Geographical InformationScience 13(5), 445–473.

Ubeda, T., Egenhofer, M. J. (1997): Topological Er-ror Correcting in GIS. In: Advances in Spatial Da-tabases, 5th International Symposium, SSD' 97.

73

74

ISSNEW – Developing an Information andSimulation System to Evaluate Non-pointNutrient Loading into Waterbodies

1. IntroductionSince it came into effect on 22th December2000, the European Water Framework Direc-tive (WFD 2000) is being implemented as animportant element of Community action in thearea of water resources protection andmanagement. This means a number of newrequirements to be met by the water manage-ment administrations. Not only does additionalinformation about the state of water resourcesneed to be collected and systematically prepa-red but also, extending on this data, all rele-vant waterbodies in the EU member states hadto be initially documented and assessed by theend of 2004. If a waterbody failed the para-meters of a good ecological status, then stepsare to be specified and undertaken appropria-te to meet the WFD requirements. Measuresshould be introduced based on efficiency anddifferentiated according to specific place andtime. They will be summed up in river basinmanagement plans to be set up by 2009 withthe fulfilling of WFD targets to take place nolater than 2015. Also incorporated in theseriver basin management plans will be an area-wide summary of significant pressures andimpact of human activity on the status of sur-face water and groundwater, comprisingamongst others, the estimation of diffuse sour-ce pollution including a summary of land use.All these activities are to be carried out within

a governmentally controlled procedure under-going a multi-step process of open participa-tion from the public.

In determining the appropriate measures, toolsmust be available that explicitly address spaceand time in allocating the courses of action, e.g. via scenario analysis techniques based ondistributed and process-oriented models. In do-ing so, relevant information should be factoredin as extensively as possible. Something uncon-ditionally stipulated by WFD are river-relatedregions projects, being as a rule not congruentwith the traditional administrative structure ofGermany. From the perspective of IT, this entailssuch topics as multi-user access, client/server,and data storage, among many others.

The presented ISSNEW project concerned thepreparation of software components meetingessential WFD parameters in the form of amarket-ready product family to supply many ofthe current WFD obligations with efficientsolutions. This has been exemplified by nitro-gen emission analyses, which track the subsur-face transport processes of excess N from theplant root zone via the unsaturated zone andthe groundwater zone to the surface waterbo-dies. Potential users are state and the federalenvironmental agencies responsible for plan-ning measures to meet the WFD targets.

Dannowski R. (1), Arndt O. (2), Schätzl P. (2), Michels I. (2), Steidl J. (1), Hecker J.-M. (1),

v. Waldow H. (1), Kersebaum K.-C. (1)

(1) Leibniz-Centre for Agricultural Landscape Research (ZALF), Eberswalder Straße 84, D-15374 Müncheberg,

Germany, rdannowski@zalf.de

(2) WASY GmbH Institute for Water Resources Planning and Systems Research, Waltersdorfer Straße 105,

D-12526 Berlin, mail@wasy.de

75

ISSNEW was planned and executed as a two-year cooperative project. The partners havebeen two institutes of a research centre (coe-vally the leading partner), experienced inhydrologic modelling and software design, anda consulting and software developing enterpri-se. Insofar, the project work itself was an expe-riment, apart from its intent to unite the philo-sophies of open source and proprietary soft-ware development.

2. Objectives of the ProjectISSNEW implied to unleash information tech-nology available in the fields of water resour-ces planning and management by means of amodular information and simulation system.Especially it was directed at introducing andproviding modern information technology inthe course of valuating and planning of mea-sures for river basin management. Thus theproject aimed at developing the following soft-ware components:

1. GIS and data bank based informationsystem for the gathering, structuring andvisualisation of geo-data and simulation re-sults on non-point nutrient input into water-bodies (groundwater and surface waters).

2. Simulation system for the evaluation of theeffects on water quality offered by measu-res against non-point nutrient flow fromagricultural lands and non-point nutrientinput into waterbodies.

3. Bi-directional intelligent interfaces bet-ween the information system and simula-tion system.

In view of the new questions arising fromimplementing the WFD, the software systemto be developed was meant to be utilisable asa decision support tool. In particular it had tobe qualified for drawing regionally differentia-ted conclusions on the risks from non-pointnitrogen sources to aquatic ecosystems as wellas for specifying and valuating measures ofpreventing those risks. Consequently, throughthe components themselves and especiallythrough their effective collaboration supported

by large geodatabases, the following goalswere to be attained:

- A platform-independent, completely com-ponent based software system for data sto-rage, analysis, and presentation to be avai-lable extending on a standardised informa-tion structure and taking into account theimportance of simulation models.

- The growing stocks of geodata already exi-sting as well as those still to be collected bywater management institutions and servi-ces in river basins to be supplied perfor-mantly by the implemented informationsystem, improved, and optimally furtherprocessed.

- Against the background of the WFD imple-mentation, an improved management ofknowledge to be made possible throughthe integration of space-time-based, scena-rio-capable modelling software for non-point nutrient input into the groundwaterand into the surface waters.

3. Results3.1 The information systemThe software ArcWFD (now termed WISYS)developed by WASY forms the basis for theISSNEW information system. ArcWFD/WISYS isused by numerous government agencies allover Europe to implement the WFD. Basicinformation about ArcWFD/WISYS is availablefrom www.arcwfd.com.

ArcWFD/WISYS is based on the geographicinformation system software ArcGIS develo-ped by ESRI. It consists of an extensive database structure for storing WFD-related dataand additional tools in ArcGIS. The databasestructure has been extended for purposeswithin the ISSNEW project. It allows efficienthandling and analysis of all basic data. AnESRI personal geodatabase is used within ISS-NEW for data storage.

76

For ISSNEW, the following ArcWFD/WISYStools are of major interest:- Theme overview

Theme activation according to user prefe-rences and user permissions

- Theme and meta data managerPreselection for theme and meta data visu-alisation

For general data analysis and display themeans of ArcGIS are used.

ISSNEW Object ModelEfficient data storage and transfer of basicdata into the different simulation systemsrequire a sophisticated normalised relationaldatabase system. The object-oriented ISSNEWdata model has been developed in UnifiedModeling Language UML, using the CASE toolMicrosoft Visio. The ArcWFD/WISYS datamodel has been developed and optimisedduring several years. It completely implementsthe horizontal guidance document »WaterBodies« (2002) as well as guidance document»Implementing the GIS Elements of the WFD«(2002). The ArcWFD object model containsaround 300 classes with approximately 3,000

different attributes, relations, value domainsand rules.

The ISSNEW object model consists in mainlytwo parts:1. Basic data2. Model data for newSocrates and UZModule

In the »basic data« section all relevant basicdata can be stored. Class structures and attri-butes correspond to the data sampling for-mats. The content of the »model data« sectionis derived from the basic data and provides thesimulation systems with the data structuresthey require for modelling. In Figure 2 a cutoutof the »model data« object model is shown.

From the ISSNEW-related basic data sectionthe hydrogeologic data is taken as an example.

Storing 2D and 3D hydrogeologic data in geo-databases has become a more and moreimportant topic in research during the lastyears. There are several first ideas for datastructures and standardisation, for example byStrassberg & Pierce (2001) or Strassberg &Maidment (2004).

Figure 1: ArcWFD/WISYS User Interface

77

As the corresponding data models were deve-loped simultaneously to ISSNEW, and as theywere not directly applicable to the ISSNEWproject, a completely new object model had tobe developed. The idea was to primarily storeoriginal hard field data like borehole informa-tion, adding derived and interpreted data likestratigraphic information and model layers in asecond step.

Applying complex normalised data models,there is a strong need for convenient dataimport from existing data collections and foreasy integration of additional sample data.Within ISSNEW, the software GeoDAtaeXchange (by WASY GmbH) is used. GeoDAtaeXchange allows the transfer of file-baseddata and data out of existing geodatabasesinto a normalised geodatabase structure.Linking source and destination attributes isdone graphically (Figure 3), and during thedata import the relations between the objects

are created automatically. For repeated importof data structures numerous template projectsfor ISSNEW have been created.

Editor with Mask GeneratorThe Editor has been developed to efficientlyenter and edit data in the GIS system. This tooloffers diverse functionality for editing graphi-cal and non-graphical objects in a geodataba-se. All objects in all currently loaded layersincluding all objects connected via relationscan be displayed and edited. Object selectioncan be done by attribute enquiry or by geome-tric selection.

For editing a specific form with interactiveinput features is provided, which is automati-cally adapted to the different object types(Mask Generator). Relations to other objectsare visualised and can be edited, too. Relatedobjects can be directly edited. In Figure 4 theEditor with Mask Generator is shown.

Figure 2: ISSNEW objectmodel: Model data (cutout)

78

Figure 3: GeoDAta eXchange import project

Figure 4: Editor with Mask Generator – user interface

79

3.2 The simulation systemThe ISSNEW software system is built up of a setof associated simulation components whichare coupled with each other and with theinformation system via ISSNEW-specific inter-faces (Figure 5). These simulation componentsare as follows:

- FEFLOW® (WASY): two-dimensional non-steady groundwater flow and nitrate trans-port through the saturated zone (FEFLOWis actually the main controlling componentfor simulation runs using FEFLOW’s interfa-ce manager)

- newSocrates (ZALF): stepwise steady-statesoil-vegetation-atmosphere transfer (SVAT)of moisture and dynamic nitrogen transfor-mation in soil (redesign of the open-sourceSOCRATES legacy code)

- UZModule (ZALF): one-dimensional (down-/upward) non-steady water and nitrate flowin the deeper unsaturated zone (»trailingwave« approach combined with enhancedupwind differencing algorithm – open sour-ce programming)

- MODEST (ZALF): grid-based steady-stateevaluation of the nitrate transport andtransformation in the (un-)saturated zone(functionally extended C++ implementa-tion of a GIS-based stand-alone tool fornon-point N sources risk analysis).

Direct linking of simulation components andthe information system »on the fly« permitsdetailed reproduction of the partially interrela-ted processes of leachate flow and nitratetransformation.

3.2.1 FEFLOWFEFLOW is a commercial simulation systembeing developed by WASY GmbH since morethan 20 years. It is used all over the world bygovernment agencies, universities and researchinstitutions, and private consulting companies.

FEFLOW has been designed for the most com-plex groundwater simulations in water resour-ces management, mining, for environmentaltasks, in hydrogeology and for geothermalenergy exploitation. Interaction betweengroundwater and surface water bodies can beconsidered. FEFLOW allows the simulation of

Figure 5: ISSNEW UML Component Diagram

80

very large models in high temporal and spatialresolution. FEFLOW is based on finite-elementtechnology. Its modelling abilities encompass 2Dand 3D simulation, saturated and unsaturatedconditions, flow and mass/heat transport.

FEFLOW’s open programming interface IFMenables interfacing to other simulation codesby linking user-written code to the commercialsimulation engine.

Within ISSNEW, FEFLOW is used to simulate 2Dgroundwater flow and nitrate transport ingroundwater. It is also the controlling compo-nent for the simulation network, i. e., FEFLOWis used for starting and stopping the simulationand for controlling the time stepping procedu-re. The interfaces between the different simu-lation components are included into theFEFLOW user interface via IFM.

3.2.2 newSocratesThis soil-crop-atmosphere oriented regionalmodel consists of (1) a plant growth modelbased on the evolon approach and startingfrom a yield estimate to dynamically calculatebiomass accumulation (Mirschel et al. 2002),(2) a coupled set of capacity models to simula-te water budget (Wegehenkel 2000) and nitra-te transformation (Kersebaum 1995) in soil.Infiltration is calculated following an empiricalapproach, moisture content and flux are cap-tured based on a cascade of non-linear reser-voirs. Within the nitrogen model, net-N mine-ralisation, denitrification and nitrate transport(as conveyed by vertical water flow) are simu-lated. The time step is fixed at one day, thusnewSocrates is determining the timing for thewhole ISSNEW simulation system. The verticaldomain for newSocrates use is confined to 2 mbelow surface within ISSNEW.

newSocrates module is organised into theblocks COM interface, database access (ODBC)and SVAT model, as well as the model-specificpart of the ISSNEW-PGDB database. Consis-tent data preparation (intersecting spatialinput information for defining topoi, filling

newSocrates-relevant tables of ISSNEW-PGDB)is unconditional prerequisite for successfulmodel run.

3.2.3 UZModuleUZModule was tailored of two components:KinFlow for vertical flux calculation in the dee-per unsaturated zone (> 2 m below surface),transportMPDATA for the associated nitratetransport. Apart from the other simulationmodels, there was no development before theproject start. Work was based on a thoroughexamination of the existing literature to identi-fy the appropriate methodology qualified forthe prevailing flow conditions. KinFlow imple-ments an analytic solution of the quasi-linearhyperbolic partial differential equation of moi-sture flow using a »trailing wave« approxima-tion (Charbeneau 1984) for tracking a wettingfront towards groundwater. This approachtakes the middle ground between conceptualmodels and models solving the more process-based Richards equation in terms of computa-tional cost and the physically correct descrip-tion of the process.

transportMPDATA was designed to use theMPDATA algorithm (Multidimensional PositiveDefinite Advection Transport Algorithm –Smolarkiewicz 1983) to evaluate the advectivetransport of dissolved nitrogen through theunsaturated zone based on upwind differen-cing. Smolarkiewicz & Margolin (1998) pres-ented an enhanced form of that algorithmwhich is inherently mass conservative, stableand simple to code and provides the basis fornitrate transport simulation in UZModule.

By the end of the project period UZModuledevelopment was not yet finished. In particular,the transport component transport MPDATA ismissing. Therefore, for ISSNEW test runs adummy code had to be used for the time being.

3.2.4 MODESTMODEST reproduces the fundamental aspectsof the underground dissolved N transport and

81

nitrate depletion in a spatially explicit, two-dimensional manner by means of coupled ana-lytic and empirical calculation methods assu-ming steady-state flow conditions. The actualC++ implementation relies on a grid-basedapproximation of the advective, nondispersivegroundwater-borne solute transport. Fromgiven groundwater table and known aquiferparameters, the model provides an evaluationof the transport and transformation dynamicsof nitrates along the underground path – fromentering the groundwater at any point of themodel region towards discharging into therespective draining waterbody. Pathlines are cal-culated according to Pollock (1988). Comparedto the preceding GIS implementation, MODESTwas extended by the possibility of a spatially dif-ferentiated parameterisation of the denitrifica-tion module and a nitrate retention potentialcalculation. The model is applicable for largeareas in virtually unconfined spatial resolution.Input data and the computation results current-

ly are still exchanged raster-based via a simpleGIS interface (ASCII-GRID).

Though not fully integrated into ISSNEW,MODEST represents a valuable addition to themuch more sophisticated simulation system. It isrecommended to be applied prior to running thesystem to reduce the modelling area of ISSNEWand, thus, to significantly shorten the expensesfor parameterisation and computation.

3.3 The intelligent interfaceThe intelligent interface is the central core ofthe ISSNEW simulation system. It connects theinformation system and the simulation systemand provides a graphical user interface for theentire system. Via the user interface participa-ting simulation components can be chosen,basic settings can be done and the type andpoint in time of data exchange betweenmodules can be defined.

Figure 6: ISSNEW parameter association

82

Scenario managementFor the planning of measures within the scopeof the WFD, scenario techniques have to beapplied. The graphical user interface of the ISS-NEW interface allows for handling differentscenarios and scenario versions, which can beloaded, edited and saved. All settings for a pre-pared scenario are saved within the ISSNEWdatabase.

The main settings of a scenario are the partici-pating modules (information system, simula-tion components), the workspace, the startingtime of the simulation and the options for datatransfer between different modules.

Data exchangeOne of the most important functions of theISSNEW interface is transferring data betweenthe different components of the simulation.Data exchange is defined graphically by linking»source« modules with »destination« modules(Figure 6)

Each parameter association has specific pro-perties like moment of execution (e. g., beforeor after a time step) and possible options forregionalisation.

RegionalisationCoupling different simulation components andthe information system, it is necessary to dealwith different geometric properties of the rela-ted data. For example, the simulation compo-nents newSocrates and UZModule run on an1D basis for so-called topos polygons, whereasFEFLOW uses a nodal and elemental spatialdiscretisation. The ISSNEW interface providesthe possibility to regionalise data during trans-fer. Two different regionalisation methods arecurrently supported: inverse distance weigh-ting for point-related data and a polygon-to-polygon intersection algorithm for areal data.Via COM interface additional regionalisationmethods can easily be included. TheRegionalisation Editor offers the possibility tolimit the regionalisation to subsets of the enti-re geometry (Figure 7).

Figure 7: RegionalisationEditor

83

Technical implementationDesigning the ISSNEW interface the focus wasset to a convenient and easy extensibility of thesystem. Therefore Microsoft COM technologyhas been used to capsule all modules of thesystem. Interface definitions have been develo-ped which have to be implemented by all par-ticipating components. There are three classesof modules:

- Simulation components participating in thesimulation run

- Data providers- Data consumers

Additional interface definitions, for example,allow for linking data sources and destinationsor error handling.

The IParticipator interface is required to partici-pate in the ISSNEW simulation system andmust be implemented by all components to beincluded. Its methods are used to control thesimulation run.

Data providers and data consumers use theIDataProvider and IDataConsumer interfacedefinitions. Each of the ISSNEW modules canimplement these interfaces, depending onwhether data have to be provided or received.

Data exchange between the components isrealised using the IDataLink interface whichcontains references to the data source, thedata destination and additional attributes. Ifsource and destination geometries cannot bereferenced via a common attribute, theIRegionalization interface is used to call a spa-tial interpolation routine.

As both source and destination data can beorganised in time series, an ITimeTable interfa-ce provides the functionality to access timeseries data.

Scenario management is done using a specificmodel repository within the geodatabase.Using this repository, different applications areenabled to store independent scenario or

model definitions along with arbitrary model-specific data. To manage the correspondingdatabase entries, a separate component calledStorage Server has been developed which pro-vides the interfaces to load and save data.

3.4 Pilot area, test scenario and test resultsUsability, performance and validity of the soft-ware system were to be tested and proven bythe example of a mesoscale pilot area underagrarian land use in the unconsolidated rockregion of eastern Germany. After consultingthe Brandenburg state environmental agency(Landesumweltamt Brandenburg), the area ofa groundwater body (ca. 100 km2, Figure 8)was chosen situated in the upland neighbour-hood of the Oderbruch depression in easternBrandenburg. This body of groundwater»Oder 2« (ODR_OD_2) had been identifiedduring the initial characterisation in the courseof implementing the WFD as to be »at risk offailing to meet the objectives of a good chemi-cal status« (IKSO 2005), based on the detectedconcentration of nitrates in groundwater. Forexample, nitrate concentration in the upperaquifer near the village Reichenow has beenrelatively constant at about 150 mg/l, threefoldthe maximum permissible value in drinkingwater. This fact very well corresponds with thepredominantly low nitrate retention potentialof that region as depicted in a preliminaryinspection by means of MODEST. Furthermore,the suitability of this groundwater body fortesting the ISSNEW simulation system is edu-ced from the geohydrologic situation: depth togroundwater strongly varying between up to70 m near the water divide and less than 2 min the Oderbruch depression; streamlines pre-dominantly parallelised towards the receivingsurface waterbody (the channelised Old Oderriver), though also several production wells arebeing operated near the town of Wriezen; par-tially unconfined groundwater conditions pre-vailing as combined with considerable infiltra-tion potential of the unsaturated zone, butalso, in other parts, confined or even tempora-rily water-bearing conditions are met; the aqui-fer system being pervaded by widely spread, in

84

turn sometimes but local aquitards. Alto-gether, this renders the pilot area to be suita-ble for proving the ISSNEW simulation systemby a pretty wide spectrum of practical cases.

Model runs performed during the ISSNEWworking period integrated FEFLOW andnewSocrates as a prototype simulation system.UZModule, however, was replaced by adummy component. Comparable to the simu-lation components it operated on the basis ofthe topos geometry as filed in the model-spe-cific part of the ISSNEW information system.This dummy component simulated a spatio-temporally variable flow of percolate (waterand dissolved nitrate) through the deeperunsaturated zone. This offered the possibilityof testing all the processes of simulation anddata transfer to be executed, with the excep-tion of the actual UZModule simulation.MODEST was excluded a priori from beingintegrated into the simulation system.

Technical test calculations were performed for asimulation period of 10 years, 100 days for sto-ring complete results, respectively. All parameterassociations were performed without errors,read and write operations and the simulationrun were finished without problems. The systemran stable, no major oscillations occurred.

The required simulation times were far belowthe expected ones. An AMD Athlon™ XP2800+ System with 2.08 GHz and 512 MB RAMwas used for the tests. For the polygon-to-poly-gon regionalisation necessary after initialisationin each time step, 300,000 data sets per secondcould be processed, for writing results for obser-vation points to the database using IDW regio-nalisation, 12,000 values per second were writ-ten. FEFLOW took around 1.9 second for thesimulation of a time step of 1 day.

Generally the coupling of simulation systemsand the time-step based transfer of data bet-

Figure 8: ISSNEW pilot area »Oder 2« (ODR_OD_2),Eastern Brandenburg

85

ween the components lead to reasonablesimulation times. A final estimation of thecost-value ratio, however, has to be postponedafter a real application. The results of the tech-nical tests point at the usefulness of the con-cept in real model simulation.

4. Summary and conclusionsGrowing stocks of geodata and new requestsagainst the background of the European WaterFramework Directive (WFD) for regionally diffe-rentiated conclusions on risks to aquatic ecosy-stems from diffuse nutrient entries, as well asfor suitable measures of preventing negativeimpacts from land use on the quality of waterresources and the waterbodies, imply to unleashavailable information technology by means of amodular information and simulation system.

The ISSNEW joint project aimed at developingsuch a software system also named ISSNEW tobe utilisable as a reliable decision support toolin the course of valuating and planning ofmeasures for river basin management. By inte-grating an information system containing ageodatabase, with components of a simulationsystem for modelling the nitrate loading ofground and surface waterbodies, ISSNEW pro-vided the prerequisite for scenario-based ana-lyses directed at the WFD targets.

ISSNEW consists of the components GIS- anddatabase-supported information system,modular simulation system, and bidirectionalintelligent interface. The information system(ArcWFD/WISYS) assists efficient acquisition,maintenance, visualisation and analysis of geo-data and simulation results for non-pointnutrient emissions into waterbodies. The simu-lation system combines modules for waterflow and dissolved nitrogen transport from theroot zone (newSocrates) via the deeper unsa-turated zone (UZModule) and the aquifer(FEFLOW) towards the waterbodies. Direct lin-king between simulation components and theinformation system permits detailed reproduc-tion of interrelated processes. A grid-basedstand-alone tool for N emissions risk analysis

(MODEST) is recommended to be initiated inadvance, in order to reduce the area of apply-ing ISSNEW and thus to shorten the expensesfor parameterisation and computation.

In regard to the simulation componentsnewSocrates, UZModule and MODEST to be(further) developed by ZALF during the project,besides a differentiated achievement of theobjectives set with the application, the princi-pal compatibility of proprietary (FEFLOW,Microsoft COM) and open-source softwaredevelopment within one joint project wasdemonstrated. Problems especially arose fromthe seam between both the domains (resp. theapplicants ZALF and WASY GmbH): implemen-ting interfaces, agreeing upon database struc-tures and access routines, as well as accom-plishing test runs by the example of the pilotarea. In the latter the missing of adequate sha-red capacity of personnel was a special handi-cap. The ZALF simulation components are avai-lable according to the open-source conventionand are to be published online.

The information system, as well as the prototy-pe simulation system built up of modellingcomponents in parts not yet fully tested andvalidated, were proven by the example of a100 km2 pilot region in eastern Brandenburg.

The ISSNEW project provided substantialexpertise for the participants in the fields ofcoupling information and simulation systems,as well as advances in the development andavailability of process-oriented simulation com-ponents for water flow and matter transportacross the compartments root zone, deeperunsaturated zone, and groundwater zone.

AcknowledgementsThe projectshave been fundet in the frame ofthe programme GEOTECHNOLOGIEN of BMBFand DFG, BMBF-Grant 03F0371 A, B.

86

ReferencesCharbeneau, R.J. (1984): Kinematic models forsoil moisture and solute transport. - WaterResour. Res., 20(6): 699-706.

IKSO (2005): Internationale FlussgebietseinheitOder. Merkmale der Flussgebietseinheit, Über-prüfung der Umweltauswirkungen mensch-licher Tätigkeiten und Wirtschaftliche Analyseder Wassernutzung. Bericht an die EuropäischeKommission gemäß Artikel 15, Abs. 2, 1. An-strich der Richtlinie 2000/60/EG des Europä-ischen Parlamentes und des Rates vom 23.Oktober 2000 zur Schaffung eines Ordnungs-rahmens für Maßnahmen der Gemeinschaft imBereich der Wasserpolitik (Bericht 2005).Koordination im Rahmen der InternationalenKommission zum Schutz der Oder (IKSO), 169S. + Anlagen.

Kersebaum, K.C. (1995): Application of a sim-ple management model to simulate water andnitrogen dynamics. Ecological Modelling, 81:145-156.

Mirschel, W., Wieland, R., Jochheim, H.,Kersebaum, K.C., Wegehenkel, M. & Wenkel,K.-O. (2002): Einheitliches Pflanzenwachs-tumsmodell für Ackerkulturen im Modellsys-tem SOCRATES. In: Gnauck, A. (ed.): Theorieund Modellierung von Ökosystemen: 225-243,Aachen (Shaker).

Pollock, D.W. (1988): Semianalytical computa-tion of path lines for finite difference models.Ground Water, 26(6): 743-750.

Smolarkiewicz, P.K. (1983): A simple positivedefinite advection scheme with small implicitdiffusion. - Mon. Weather Review, 111: 479.

Smolarkiewicz, P.K. & Margolin, L.G. (1998):MPDATA: A finite-difference solver for geophy-sical flows. - J. Comp. Physics, 140: 459-480.

Strassberg, G. & Pierce, S. (2001): Arc HydroGroundwater Data Model, Center for Researchin Water Resources, University of Texas at Austin.

Strassberg, G. & Maidment, D.R. (2004): ArcHydro Groundwater Data Model, AWRASpring Specialty Conference »GeographicInformation Systems and Water Resources III«,Nashville, Tennessee, 17-19 May 2004.

Wegehenkel, M. (2000): Test of a modellingsystem for simulating water balances and plantgrowth using various different complex appro-aches. Ecological Modelling, 129: 39-64.

WFD (2000): Richtlinie 2000/60/EG desEuropäischen Parlaments und des Rates vom23. Oktober 2000 zur Schaffung eines Ord-nungsrahmens für Maßnahmen der Gemein-schaft im Bereich der Wasserpolitik. Amtsblattder Europäischen Gemeinschaften, L 327, 22.Dezember 2000, 1-72.

87

88

Marine Geo-Information-System for SpatialAnalysis and Visualization of HeterogeneousData (MarGIS)

IntroductionProfound investigations of marine and terre-strial environments require compilation ofextensive and complex, related to the numberof parameters and methods, data sets for pro-cess oriented and spatial analysis. Startingfrom the »endmembers« of pure research, as expeditions by RV Challenger (1872-1876), FSGazelle (1874-1876), or FS Valdivia (1898-1899) and economically oriented surveys forfish or for oil and gas the number and diversi-ty of scientific and applied studies increasedconsiderably. Especially within the last twodecades developments of new sampling devi-ces, in situ sensors, and mobile underwaterplatforms as ROV´s, (Remotely OperatedVehicles), AUV´s (Autonomous UnderwaterVehicles) and Crawler´s (mobile, wheel drivenunderwater vehicles) provide new capabilitiesfor marine research. The multitude of measu-red parameters and the quantity of informa-tion compiled during multidisciplinary researchcruises requires new concepts for the manage-ment and spatial analysis of geodata. The termgeodata summarizes biological, chemical, oce-anographic or geological measure-ments,which are tied to the geographic coordinate (x,y, z, t) of the site of sampling or measurement.Geodata are the cornerstones for research

objectives, a requirement for management ofoffshore resources, and for implementation ofnational and international regulations as theWater Framework Directive or natural conser-vation issues. Furthermore, they support deci-sions about the possible use of the coastal sea-floor for economic demands as offshore windenergy plants, sand and gravel mining, oil andgas exploration, or mariculture. Whereas onland such decisions are supported by the avai-lability of thematic maps (e.g., about land use,soil type, fauna and flora, geology or hydrolo-gy) and environmental data, such compiledgeoinformation are rather sparse for the mari-ne environment. The increase of geodata deri-ved by application of new marine technologyand the demand for aggregated geoinforma-tion for scientific and applied purposes werestarting points for the MarGIS project.Objectives of MarGIS are the compilation ofheterogeneous marine data, the developmentof an appropriate data base model which isclosely linked to a Geo-Information System(GIS), and the application of advanced statisti-cal techniques to characterize and identify pro-vinces at the seafloor. Target areas are theNorth Sea and parts of the Baltic Sea andNorwegian continental margin. For these pur-poses data on sedimentology, geochemistry,

Schlüter M. (1), Schröder W. (2), Vetter L. (3), Jerosch K. (1), Peesch R. (2), Köberle A. (3),

Morchner C. (1), and Fritsche U. (1)

(1) Alfred-Wegener-Institute for Polar and Marine Research, Am Handelshafen, Box 120161, D-27515 Bremerhaven,

Germany, mschlueter@awi-bremerhaven.de, kjerosch@awi-bremerhaven.de, cmorchner@awi-bremerhaven.de,

ufritsche@awi-bremerhaven.de.

(2) University Vechta, Box 1553, D-49364 Vechta, Germany, wschroeder@iuw.uni-vechta.de,

rpeesch@iuw.uni-vechta.de

(3) University of Applied Sciences, Box 110121, D-17041 Neubrandenburg, Germany vetter @fh-nb.de,

akoeberle@fh-nb.de

89

benthic biology, fish stock, bathymetry or che-mical oceanography were compiled.

Marine Geodata: Sites, Tracks, Areas, and Time SeriesIn marine research, sampling and data acquisi-tion is conducted during cruises by researchvessels and, to a lesser extend, by satellite orairborne remote sensing. During cruises, avariety of devices are applied for sampling ofsurface sediments and sediment cores (Fig. 1). Different types of equipment are used to catchfish, plankton, or benthic biota. In situ sensor packages are deployed for long-term measure-ments of water currents, concentrations ofnutrients, suspended matter, or fluxes of parti-culate matter from the photic zone to the sea-floor. Acoustic techniques as multi-beam, sidescan sonar, or shallow seismic are used forbathymetric surveys, habitat mapping, andinvestigation of sub-seafloor geology (Wrightand Bartlett, 2000). High resolution still photo-graphs and videos are recorded by towed devi-ces and during dives by ROV´s, submersibles,or Crawler like the MOVE-System (MARUM,Univ. Bremen) for visual identification of geolo-gical or biological features as well as forinspection of infrastructure as pipelines. Forexample, by the ROV Victor6000 (IFREMER, FR)about 250 million soundings were recorded bya high resolution multibeam system duringtwo dives of 44 and 15 hours bottom time atthe Haakon Mosby Mud Volcano (Jerosch etal., sub., Klages et al., 2004). This allowedcomputation of a high resolution microbathy-metry map with a depth resolution of betterthan 0.1 m and a foot-print of 0.5 m at theseafloor. During just six dives of video surveysand video mosaicing more than 4300 georefe-renced mosaics were generated for classifica-tion of geochemical habitats. Furthermoresediment and water samples were gatheredand sensor packages recorded temperature,density, conductivity and concentrations ofchemical constituents along transects.Consequently, a significant amount of geoda-ta stored on DVD´s were obtained within justabout 2.5 weeks of ROV survey during onecruise. Considering that in most cases an area

of investigation is studied by several cruises a significant amount of geodata is compiled.Obviously, data management and spatial ana-lysis of geodata by GIS techniques combingseveral information layers, geostatistics, orimage analysis (Jerosch et al., subm.) are upco-ming requirements. For these purposes, stora-ge of geoinformation on a »cruise-entity-level«, for example in a file-structure separa-ting data derived during single cruises, is notefficient. Such a separation would cause con-siderable efforts to find and merge geodata.Especially during a cruise, a flexible and effi-cient access to data derived during previousexpeditions or compiled from literature andreports might support decisions about sam-pling sites or tack lines. A data model provi-ding access on a »data value entity« is benefi-cial, therefore. An integrated approach, sup-porting spatial analysis of marine geodata isrequired, which allows compilation of differentdata sets and data types (Fig. 1) within onedata base management system (DBMS).Basically, this encompass that –in best case-one query is able to retrieve a complex geore-ferenced data set which can mapped and ana-lysed within a Geo-Information-System (GIS).Besides measured or aggregated geodata, thedescription of the mode and techniques app-lied for sampling and chemical or physical ana-lysis is essential for the evaluation of the data.Such metadata have to be stored within thegeodatabase.

Marine Data ModelThe general demand for efficient and adequa-te data management initiated developments of Marine Data Models (MDM) at several institu-tes and companies. For example, the MarineData Model Group intends to develop an UMLbased, generic MDM scheme which is closelylinked to the geo-object concept and GIS soft-ware distributed by major IT companies. Insome cases, this very promising concept has tobe modified to cover specific scientific andadministrative needs. For example, in Germanythe BSH (Federal Maritime and HydrographicOffice) is establishing an advanced MDM for

90

data management of a complex set of hydro-graphic measurements, results of numericalmodels, or information about currents andtides. With different focus, MDM´s were deve-loped or are in development at institutes likeIFREMER (FR), National Oceanographic CentreSouthampton (UK), Oregon State University(US), or Scripps (US). For the purposes of theMarGIS project, we started from the genericMDM scheme. Our aim was to integrate geo-chemical, geological, biological, and oceano-graphic data measured along vertical profilesor transects into this scheme to support spati-al analysis and process oriented investigations.The latter include studies on marine habitats,geochemical turnover processes and sediment-water-interaction. For these specific purpose,we decided after contacts and discussion withother research groups, not to fully adopt thegeo-object approach of the generic MDM

scheme. Instead we developed a Data Modelwhich intends to bridge the gap betweendistributed, file-based data storage, geo-objectrelated MDM´s, and elaborated DBMS used inlong term archives as the world data centrePANGAEA. The MarGIS geodatabase(MarGis_GDB) model is implemented on aclient-server DBMS which allows integration ofmaps (vector and raster data), raw data, geo-referenced images (geotiff´s), and metadata(Fig. 2). Central part of the data model aretables containing information about samplingsites and tracks studied during research cruisesor obtained from the literature or reports.These tables are attributed as sites of measure-ments, locations of time series, or starting/endpoints of line features as tracks lines, dives, ornet hauls. Furthermore, data derived by multi-level sampling, in situ sensors, and obtainedalong track lines and areas are stored. In addi-

Figure 1: During research cruises a multitude of sampling devices and sensor are applied to derive geological, chemical, oceanographic or biological data. The multitude of measured parametersand the quantity of information compiled during multidisciplinary research cruises requires newconcepts for the management and spatial analysis of geodata. Photographs are kindly provided by MARUM (MOVE-System), AWI (AUV Bluefin, in situ Profiler), MPI-Bremen (in situ profiler), andIFREMER (ROV Victor6000)

91

tion to raw data, aggregated information asthematic maps on sedimentology or bathyme-try were integrated into the relational databa-se management system.

MetadataComparison and analysis of heterogeneousdata derived from different sources or obtai-ned by a multitude of research cruises has toconsider the different sampling techniques oranalytical methods. These methods mightchange over the years. Metadata, providinginformation where and how samples wereobtained or which analytical methods wereapplied, are essential for data compilation andretrospective data analysis.For example, in theMarGIS_GDB metadata about cruises anddives are stored. This includes informationabout the mode of navigation (e.g. for ROVdives, navigation by USBL (ultra-short-base-line) or inertial navigation systems), the coordi-nate system, or the geodetic datum (WGS84,ED50). Essential information about the techni-cal specification of sampling devices as sedi-

ment corers or water samplers, about in situsensors, camera systems, and chemicalmethods for sediment and water analysis arearchived as metadata as well. For aggregatedinformation as geological or geochemicalmaps, which were georeferenced, digitized,and integrated into the geodatabase, the datasource, geographic projection or the appliedgeodetic datum are stored as metadata. Inaddition, information about the data quality orthe applied classification scheme are stored.The latter is important for e.g., sedimentologi-cal data, were the Folk or the Wentworth clas-sification are applied to describe grain sizedistributions. International initiatives establis-hed different formats and schemes for descrip-tion of geodata by metadata. This includes theDublin Core Metadata Element Set (basic stan-dard for the resource description), the FGDC(Federal Geographic Data Committee) widelyapplied in the US, or the ISO 19115 (metada-ta). For the MarGIS project we decide for theISO 19115, which includes information aboutthe identification, the extent, the quality, thespatial and temporal scheme, the spatial refe-

Figure 2: Simplified scheme of the data model, integrating geological, geochemical, oceanogra-phic, and biological information into a geodatabase.

92

rence, and the distribution of digital geogra-phic data (Kresse & Fadaie, 2004). A»Metadata ISO 19115 document« describeseach parameter set compiled within MarGIS.ISO standard organizes the structure of themetadata as core metadata and comprehensi-ve metadata elements. The metadata are sto-red as XML document in the database. As con-sidered subsequently, metadata are assignedto each information layer and are distributionin conjunction with geodata by an InternetMap Server.

Geodata compiled in the MarGIS_GDBThe geodata compiled within the MarGIS pro-ject and integrated into the data base werederived by an intensive recherché on literatureand reports, retrievals of published geodatafrom marine data base systems (MDBS), and byclose co-operation with scientist from variousresearch disciplines. A considerable amount ofpublished data were retrieved from marinedata base systems established by internationalinitiatives or by Federal Hydrographic andOceanographic Agencies. Prominent examplesare the data base systems operated by ICES(International Council for the Exploration ofthe Sea) or the Marine Environmental

Database (MUDAB) initiated and operated bythe German Federal Maritime and Hydro-gra-phic Office (BSH) and the FederalEnvironmental Agency (UBA). In total, informa-tion about the following parameters werecompiled (Tab. 1): bathymetry, salinity, tempe-rature, concentrations of oxygen, ammonium,nitrate, nitrite, phosphate, silicic acid, pH,alkalinity and suspended matter, data on ben-thic biology as epibenthic and endobenthicorganisms, fish populations, fish ages andlength, and on the geology and geochemistryof sediments. The latter includes sedimentmaps and spatial distribution of distinct featu-res at the seafloor like pockmarks, seeps andreefs. These compilations were supported bythe METROL project (Methane Flux Control inOcean Margin Sediments, http://www.metrol.org, coordinated by the Max Planck Institutefor Marine Microbiology). In this context geo-information about the methane cycle, gas richdeposits, fault zones, distribution of earth qua-kes, and source rocks for oil and gas werecompiled. Furthermore, data about the use ofthe seafloor as pipelines, platforms, naturalconservation areas, and sand and gravelmining were compiled. The aggregation ofheterogenous geodata obtained from veryvarious sources required a rather laborious har-

Table 1: Overview about some of the parameters compiled within the MarGIS database. (BFA/IFOE:Federal Research Centre for Fisheries, Institute for Fishery Ecology; BFA/ISH: Federal Research Centrefor Fisheries, Institute for Sea Fisheries, BSH: Federal Maritime and Hydrographic Office; EC: EuropeanCommunity project 98/021; CEFAS: Centre for Environment, Fisheries and Aquaculture Science; GFS:National Ground Fish Surveys; ICES: International Council for the Exploration of the Sea; IFMHH:Institute for Marine Research; SBS/UWB: School of Biological Sciences, University.

93

monization procedure. This was one prerequi-site for the integration of data and metadatainto the MarGIS_GDB linked to the Geo-Information-System ArcGIS.

Whereas the development of a data basemodel suited for incorporation of heterogene-ous data, supporting analysis by GIS and ableto be operate during cruises was one objecti-ves of the project, it was not designed as along-term archive. Such specific purposes arethe domain of world data centres as PAN-GAEA. For example, PANGAEA provides therequired infrastructure and includes specificconcepts for referencing data tables or cita-tions by DOI´s (Digital Object Identifiers).

Geostatistical analysis for identification ofprovinces at the seafloor Objectives as characterisation of seafloor habi-tats by benthos biology or computation ofgeochemical budgets as sediment oxygendemand or release of CH4 from the seafloorrequires extrapolation of measurements obtai-ned at distinct sites or along tracks to areas.For these purposes geostatistical methodswere applied. Originally coming from geologi-cal research and applied to estimate mineralresources and reserves (Krige 1951; Matheron1965, 1971), geostatistics are nowadays beingused in various terrestrial and marine fields ofresearch. As far as marine research is concer-ned geostatistical instruments were applied byvarious scientific disciplines, e.g. pollution rese-arch (Poon et al. 2000), geology (Chihi et al.2000; Pehlke, 2005), biology (Harbitz &Lindstrøm 2001; Jelinski et al. 2002), or mari-ne geochemistry Schlüter et al. (1998) orJerosch et al. (subm.). Compared to determini-stic procedures like IDW (Inverse DistanceWeighted Method) geostatistical methods takeinto account the degree of spatial autocorrela-tion when predicting measurements.Geostatistics can be subdivided into two wor-king steps: variogram analysis and kriging pro-cedures. With variogram analysis the autocor-relation structure of the underlying spatial pro-cess is examined and modelled. Variogram

maps can be used to detect directional depen-dencies or so called anisotropies in the datafield. The variogram models are used to predictmeasurement values by chosen kriging proce-dures (e.g. ordinary kriging). In the projectMarGIS geostatistical analyses were carried outfor ecologically relevant biotic and abioticparameters as: grain size (0-20 µm, 0-63 µm,20-63 µm und 63-2000 µm), bottom watertemperature, salinity, and concentrations ofdissolved constituents as silicic acid, nitrate,phosphate, ammonia, or oxygen measured ator near the sea floor (Fig. 3). The measurementdata were analysed with regard to three diffe-rent areas of interest and grouped into diffe-rent time intervals: the Exclusive EconomicZone (EEZ) of the North Sea (four three monthsintervals aggregated over a six-year-periodfrom 1995 to 2000), the entire North Sea(summer and winter months aggregated overa three year period from 1997 to 2000) andthe western part of the Baltic Sea (summer andwinter months aggregated over a three yearperiod from 1995 to 2000). Quality assuranceand assessment of the derived maps were con-ducted by cross-validation and standard errormaps. Additionally, if the semivariancesdisplayed on the variogram map indicatedanisotropies in the data field, different rangesfor different directions were compared witheach other. Dependent on existing spatialtrends and skewed value distributions eitherordinary, universal or lognormal kriging wasapplied to spatially extrapolate the measure-ment data to raster maps. The quality of esti-mation was documented in terms of chosenkey values derived from the results of cross-validation. Figure 3 depicts results of geostati-stical analyses. For example the ammoniumconcentrations measured at the sea floor ofthe North Sea in the summer months betweenthe years of 1998 and 2000 (Fig. 3D). Theresults of variogram analysis reveals a distinctautocorrelation structure with a low nugget-sillratio, indicative of low small-scale variability'sas well as strong spatial dependencies of themeasurement values. With the help of thevariogram map anisotropies in 53.3° directionwere detected, resulting in a searching ellipse

94

in the following ordinary kriging calculations.These maps are considered as preliminaryresults, which has to be discussed with chemi-cal oceanographers.

Calculation of ecological sea floor provinces. On basis of the geostatically estimated surfacemaps ecological sea floor provinces were cal-culated by means of multivariate statisticalmethods. For the EEZ of the North Sea as well as the entire North Sea predictive habitat map-ping was performed with help of the decision

tree algorithm Classification and RegressionTrees (CART). Predictive habitat maps calcula-ted for the EEZ of the North Sea as well as theentire North Sea rely on punctual data on eightbenthic communities collected at 184 sites wit-hin the German Bight and the bordering cen-tral North Sea (Rachor & Nehmer 2003).Predictive habitat mapping can be defined asthe development of a numerical or statisticalmodel about the relationship among environ-mental variables (data on bottom water mea-surements on salinity, temperature, dissolvedcomponents as silicic acid, oxygen, phosphateand nitrate, or sedimentological data) and

Figure 3: Examples for geodata compiled and mapped by geostatistical techniques for the North Sea.A: Digital bathymetric map of the North Sea, including the Exclusive Economic Zones. B: Distributionof sites where NO3 concentrations in bottom water are available. C: Sediment map compiled andcomputed for the North Sea and part of the Baltic Sea. D: Example for suggested ammonium con-centration in bottom water (summer months between 1998 and 2000) derived by variogram analysisand kriging.

95

benthic biological communities. The methodsused for predictive mapping vary widely fromstatistical approaches (including geostatistics)to more complex methods, such as expertsystems, and decision tree analysis (Kelly et al.2005). In the MarGIS project CART was app-lied to derive a classification model for theeight benthic communities investigated byRachor & Nehmer (2003). CART is applied invarious scientific disciplines to uncover hiddenstructures in complex data matrices. CART is aso called tree-growing algorithm that producesdecision trees to predict or classify the outco-me of a certain feature (= target variable) froma given set of meaningful predictor variables(Breiman et al. 1984). A major advantage ofthis technique is its ability to model non-addi-tive and non-linear relationships among inputvariables. In contrast to most of the classifica-tion techniques as, e.g. cluster analysis or clas-sical regression analysis, CART handles verylarge sets of mixed, i.e. both categorical andparametric data without prior transformationof scale dignity. The central goal of the CART-algorithm is to produce homogenic classeswith respect to the features of the target vari-able. Whether the target variable is of metric,ordinal or nominal scale dignity, differentimpurity measures exist. The Gini-index is com-monly used when the target variable is cate-gorical, although other options exist (Steinberg

and Colla 1995). CART does not make anyassumptions on the distribution of the dataand is extremely robust with respect to specialcases as outliers or rare biotopes.Decision treeswere computed to predict the occurrence ofbenthic communities from the intersectedabiotic grid data. One decision tree each wascomputed for two sets of predictors and timeintervals: the geostatistically estimated rasterdata for the German EEZ of the North Sea aswell as the entire North Sea. Figure 4 showsthe nodes of the decision tree for the EEZ interms of histograms where each bar is repre-sentative for one of the eight communities.The results of the CART analysis resulted in adecision tree grown in nine binary splits lea-ding to 10 endnodes or classes, respectively. Ascan be seen each decision tree starts with oneroot node containing all observations of thesample. By following the dendrogram from upto down it can be observed that the portion ofeach benthic communities increases stepwise.This leads to nine endnodes in which one ofthe eight communities is dominant (portion >75%). Since each of these end nodes is defi-ned by a set of decision rules, the tree can beapplied to predict the occurrence of benthiccommunities at places where no such informa-tion is available. Each of the resulting spatialunits may therefore be described with respectto the possibility of the occurrence of one of

Figure 4: Decision tree for the occur-rence of eight benthic com-munities derived by Rachor & Nehmer (2003)

96

the eight communities. This possibility of theoccurrence of each community can be derivedfrom its percentage in the corresponding end-node. This was done for both the EEZ and theentire North Sea resulting in two differenthabitat maps. The preliminary habitat map ofthe entire North Sea is shown in Figure 5. Allhabitat classes were described with help of sui-table statistical measures.

Distribution of Geodata via an InternetMap ServerThe traditional way for distributing geoinfor-mation – single, printed maps or atlases- issupplemented by computer-based techniqueswhich are often accessible via the internet. The internet has evolved very rapidly from delive-ring essentially static information into a verydynamic information resource. In the last yearsthe demand of spatial web-based information has been strongly increased, that means todeliver interactive, »clickable maps« on-the-flyto suit the demands of the user. In order to fol-low this requirement the large GIS manufactu-

rers – and meanwhile the open source group,as well - are developing special software toolscalled Map Server technology for managingdynamic geospatial information. Nowadays itis state-of-the-art in the GIS community towork in or to provide a Map Server environ-ment. Up to now, several web based InternetMap Server (IMS) were implemented to provi-de information about terrestrial environments.In contrast, in the marine context these deve-lopments seem to be just at the beginning.One goal of the MarGIS project was to design and to install a pertinent web-based userfriendly system for the dissemination of marinegeo-information.Figure 6 provides an overviewhow the different software components, thespatial elements, and the data model interact.In this case a combination of a client-serverdatabase and Map Server technology areresponsible for the web-based spatial analyses.The dynamic presentations of the maps in theinternet and the availability of the metadataare realized by tools as the ArcIMS (Fig. 7). Thisstructure allows to view the results of a spatialdata query in the internet in conjunction withthe most current internet browsers. For theclient the basic functions as Pan, Zoom, Query,Find and the possibility of printing are imple-mented. The Internet Map Server ArcIMS is lin-ked to an Apache Server with a Tomcat asServlet Engine. The various geo-data formats(as: shapes, coverages, grids, raster images) andthe metadata are stored in a MSSQL geodatab-ase and the ArcSDE software works as a gate-way. The processing chain for the user is: WebBrowser as client via ArcIMS and by ArcSDE tothe data.The ArcIMS Servlet Connector isresponsible for the communication between theWeb Server and the Application Server. TheApplication Server organizes the process of theincoming queries and transfers these to theSpatial Server. Different services as Image-,Feature-, Metadata-, Query-, Geocode- orExtract-Services are carrying out on theApplication Server. MarGIS uses Image-, Query-and Metadata-Services. The image service crea-tes raster data in formats as jpg, png or gif.Whenever a user zooms or pans, a query is sentto the image service with the coordinates of the

Figure 5: Example for a classification scheme derived byCART analysis for the seafloor of the North Sea.

97

new map window. The image service generatesthe new image with the new coordinates andsend the URL of the image back to the client.The communication between client and applica-tion server is realized by ArcXML. The client pro-duces and analyzes the requests and responsesof the server with the help of a java script.Besides maps and data, the Metadata (accor-ding to ISO 19115) assigned to each informa-tion layer is distributed via the Map server aswell. To tackle the problem of presenting largedata sets, MarGIS uses a web-based viewerwhich allows a clear presentation of information(Fig. 7). On the basis of cascading the informa-tion, the different layers are grouped in thema-tic blocks as for example geology, temperature,chemistry, fishery data, salinity and others. Withthe help of pull downs the user can select theappropriate layer (Fig. 7A). To each informationlayer metadata information (ISO 19115) areavailable (Fig. 7B). The document of the meta-data information is also located on the databa-se server which optimize the updating process.There are different possibilities of querying thedata: as data to a certain object (point, line,polygon, pixel, grid), to a certain region or withthe help of SQL commands. A result of such arequest is presented in figure 7A for the salinitylayer. Information layers which include quantita-tive data for example the concentration of nitra-te, descriptive statistical parameters as box andwhisker plots, average, histogram and otherscan be directly generated.

As expected, the recherché for possible datasources and data acquisition was a rather labo-rious task. Very different data types and for-mats of data tables were received. Harmoniza-tion of these data and integration into a geo-database was a rather time consuming step.Digital maps and other aggregated digitalinformation describing the marine environ-ment of the North Sea or the Baltic Sea are stillvery scarce or available only for rather smallsub-regions. Without the very good coopera-tion with different research institutes, federalauthorities, and scientists from various rese-arch disciplines such a compilation would benearly impossible (see the Acknowledge-ments). Although elaborated Marine DataModels were developed by research groupsand Federal Maritime and HydrographicOffices these had to be modified to suit ourspecific research needs. For our requirements,the close link to the geo-object concept causessome overhead for integration of data into anMDM. For example, integration of chemical orgeological raw data measured in the laborato-ry and associated metadata, describing analyti-cal methods and quality assessment, in anMDM is a multi-step process which requiressome experience with GIS. For our purposes,we used only a sub-set of the geo-object func-tionality to ensure data integration, export ofgeodata for statistical analysis, and visualisa-tion of concentration profiles without therequirement of specific software.Within theproject different statistical techniques were

Figure 6: System structure of the web-based Marine Geo-Information-Services.

applied for identification of provinces of theseafloor. Intention was to come up with a»best practise« for such a spatial analysis.Obviously, no general recommendation canbe given for analysis of heterogeneous geo-data. From our perspective and experience,the combination of variogram analysis andkriging and application of CART provides rat-her good and robust results for identificationof provinces at the seafloor.The data recher-ché showed that national and internationalprograms compiled an impressive set of mari-ne geodata. Especially the amount of infor-mation on sedimentology, benthic and pela-gic biology or chemical oceanography provi-des a very good overview about the environ-ment of the North Sea and Baltic Sea.Nevertheless, for studies about seasonal vari-

ations or long term trends only a few timeseries stations are available, where chemicalor biological data were measured over longerperiods. Furthermore, data on dissolved con-stituents relevant for studies on global chan-ge and the carbon cycle as CH4 or DMS arevery limited and accessible only for a few sub-regions. Other data sets required for ecosy-stem modelling, describing sediment-waterexchange of oxygen, nutrients, and tracegases or »admixture rates« as eddy diffusioncoefficients are scarce. From our perspective,measurement of such data sets are a pre-requisite for upcoming research projectsaiming to model and predict the impact ofglobal change on coastal environments. Acompilation of available data and a geostati-stical approach might support optimisation of

98

Figure 7: A: Geodata distributed via theInternet Map Server (IMS) anddisplayed in a standard web-browser.The data points are sites of salinitymeasurements in bottom water. Thedata retrieval results in colour-codeddata points mapped onto the bathy-metric map and as a table providingdetailed information about eachmeasurement. The data are servedfrom the MarGIS_GDB geodatabase.B: Metadata, according to ISO19115, which is closely linked toeach information layer. The metada-ta are accessible via the internetmap server. ConclusionsThe objecti-ves of the MarGIS project to identifyspatial entities –provinces- at theseafloor by combination of bioticand abiotic data and geostatisticalanalysis was achieved by a three stepprocess (Fig. 7): 1. compilation ofheterogeneous data describing thegeological, chemical, physical andbiological environment of the lowerwater column and seafloor, 2. appli-cation and comparison of geostati-stical and multivariate-statisticalmethods for data aggregation, and3. distribution of data and maps viaan internet maps server.

99

future research programs ensuring coverageof representative spatial entities at the sea-floor. A topic closely related to the distribu-tion of compiled geodata to other researchgroups our via the internet map server is thecopyright issue. For sure, data which werereceived on a partnership basis could not bedelivered to third parties. For geodata compi-led from journals and published reports oraggregated geoinformation (e.g. maps wederived by multi-variate statistic or geostati-stics) the issue of copyright seems not such aswell defined. In MarGIS all data sources arecited in and available as metadata which areclosely linked to the geodata. Nevertheless,

we decided in some cases to reduce the infor-mation content of the geodata e.g., byreclassification of thematic maps or by provi-ding colour codes instead of data valuespriorto integration into the Internet Map Server.Unfortunately, in our as well as other applica-tions the potential of IMS to provide complexdata sets from geodatabase systems for rese-arch and the general public is limited. Thenumerous discussion in workshops revealedthe need for a general agreement or otherregulations covering copyright issues relatedto distribution of geodata via the internet.

Figure 8: Schematic workflowof the MarGIS project startingfrom (I.) data recherché forgeological, chemical, biologi-cal and oceanographic rawdata and aggregate informa-tion about the bottom waterand seafloor of the North Seaand parts of the Baltic Seaand Norwegian continentalmargin. These geoinformationwere integrated into a geo-database. (II.) Maps for thedifferent parameters werederived by geostatistical ana-lysis and converted to rasterdata. Geodata and result ofthe spatial analysis are distri-buted via an Internet MapServer (III.)

100

AcknowledgementsWe are very grateful to several scientist andinstitutes for their support of the objectives of MarGIS. On an institutional level we would liketo thank especially scientist from followinginstitutes and organisations: BSH: FederalMaritime and Hydrographic Office; BfN:Federal Nature Conservation Agency; MUDAB:Marine Environmental Database of BSH andUBA; ICES: International Council for theExploration of the Sea; IFM-HH: Institute ofMarine Research, University of Hamburg;IfÖ/BFA-Fi: Institute for Fishery Ecology /Federal Research Centre for Fisheries; ISH/BFA-Fi: Institute for Sea Fisheries / Federal Research Centre for Fisheries; GEUS: Geological Surveyof Denmark and Greenland; BAW: FederalInstitute for Waterway Engineering; BGS:British Geological Survey; BODC: BritishOceanographic Data Centre; DOD: GermanOceanographic Data Centre; RIKZ: NationalInstitute for Coastal and Marine ManagementSBS/UWB: School of Biological Sciences,University of Wales, Swansea and Bangor;TNO-NITG: Netherlands Institute of AppliedGeo-science TNO - National Geological; WSD: North Directorate for Water and NavigationNorth Region; NWSD North West Directoratefor Water and Navigation North West Region; Danish Energy Authority; DEAL Data registry;Nederlands Instituut voor Geowetenschappen TNO, Utrecht, NL; UBA: Federal EnvironmentalAgency. For their interest and support wewould like to thank especially Dr. E. Rachor(AWI), Prof. Dr. S. Ehrich (BFA Hamburg), Dr. U.Brockmann (University Hamburg) and collea-gues from the BSH (Federal Maritime andHydrographic Office). Furthermore we are gra-teful for the initiative of the BMBF/DFG»Informationssysteme im Erdmanagement«and the stimulating efforts of Prof. Dr. R. Bill(University Rostock)

ReferencesBreiman, L., Freidman, J.H., Olshen, R.A.,Stone, C.J. (1984): Classification and Regres-sion Trees, Wadsworth.

Chihi, H., Galli, A., Ravenne, C., Tesson, M., DeMarsily, G. (2000): Estimating the Depth ofStratigraphic

Units from Marine Seismic Profiles UsingNonstationary Geostatistics, Natural resourcesresearch 9 (1), pp. 77-95.

Harbitz, A., Lindstrom, U. (2001): Stochasticspatial analysis of marine resources with appli-cation to minke whales (Balaenoptera acutoro-strata) foraging: A synoptic case study from thesouthern Barents Sea, Sarsia, 86, pp. 485-501.

Jelinski, D.E., Krueger, C.C., Duffus, D.A.(2002): Geostatistical analyses of interactionbetween killer whales (orcinus orca) and recre-ational whale-watching boats, Appl. Geogr.22, pp. 393-411.

Jerosch, K., Schlüter, M. Foucher, J.P., Allais, A.-G., Klages, M., and Edy, C. (subm): Spatialdistribution of mud flows and chemoautotro-phic communities affecting the methane cycleat Håkon Mosby Mud Volcano.

Klages, M., Thiede, J., Foucher, J.-P., 2004. Theexpedition ARK XIX/3 of the Research Vessel»Polarstern« in 2003, Berichte zur Polar- undMeeresforschung, 488: 346 S.

Kelly, A., Powell, D., Riggs, R.A. (2005):Predicting potential natural vegetation in aninterior northwest landscape using classifica-tion tree modeling and a GIS, Western Journalof Applied Forestry, 20 (2), pp. 117-127

Kresse, W. and Fadaie, K. (2004): ISOStandards for Geographic Information: 322 S.,Berlin Heidelberg (Springer).

101

Krige, D.G. (1951): A statistical approach tosome basic mine evaluation problems on thewitwatersrand, J. Chem. Metall. Min. Soc. S.Africa 52 (6), pp. 119-139

Lembo, G.; Silecchia, T.; Carbonara, P.;Spedicato, M.T. (2000): Nursery areas ofMerluccius Merluccius in the Italian Seas and inthe east side of the Adriatic Sea, Biol. Mar.Medit. 7 (3), pp. 98-116.

Matheron, G. (1965): Les variables régionali-sées et leur estimation. Masson, Paris

Matheron, G. (1971): The theory of regionali-zed variables and its application. Fontaine-bleau.

Pehlke, H. (2005): Prädiktive Habitatkartierungfür die Ausschließliche Wirtschaftszone (AWZ)der Nordsee, Diplomarbeit, Universität Vechta,Institut für Umweltwissenschaften.

Poon, K.-F., Wong, R.W.-H., Lam, M. H.-W.,Yeung,H.-Y., Chiu,T. K.-T. (2000): Geostatisticalmodelling of the spatial distribution of sewagepollution in coastal sediments, Water Research34 (1), pp. 99-108.

Rachor, E., Nehmer, P. (2003): Erfassung undBewertung ökologisch wertvoller Lebensräumein der Nordsee, Abschlußbericht für das F undE-Vorhaben FKZ 899 85 310 (Bundesamt fürNaturschutz), Alfred-Wegener-Institut fürPolar- und Meeresforschung, Bremerhaven.

Schlüter, M., Rutgers van der Loeff, M. M.,Holby, O. and Kuhn, G. (1998). Silica cycle insurface sediments of the South Atlantic: Deep-Sea Research I, 45, 1085-1109.

Steinberg, D., Colla, P. (1995): CART. Tree-structured Non-Parametric Data Analysis,Salford Systems San Diego, Ca.Wright, D.,Bartlett, D. (2000): Marine and CoastalGeographical Information System: 320 S.,London (Taylor & Francis)

Author’s Index

AArndt O. . . . . . . . . . . . . . . . . . . . . . 74Azzam R. . . . . . . . . . . . . . . . . . . . . . . 2

BBär W . . . . . . . . . . . . . . . . . . . . . . . 32Brand S.. . . . . . . . . . . . . . . . . . . . . . 32Breunig M.. . . . . . . . . . . . . . . . . . . . 32Butenuth M. . . . . . . . . . . . . . . . . . . 52

CChrist I. . . . . . . . . . . . . . . . . . . . . . . 12

DDannowski R.. . . . . . . . . . . . . . . . . . 74

FFritsche U. . . . . . . . . . . . . . . . . . . . . 88

GGösseln G.v.. . . . . . . . . . . . . . . . . . . 52

HHäußler J. . . . . . . . . . . . . . . . . . . . . 32Hecker J.-M. . . . . . . . . . . . . . . . . . . 74Heipke C.. . . . . . . . . . . . . . . . . . . . . 52Hübner S . . . . . . . . . . . . . . . . . . . . . 12

JJerosch K. . . . . . . . . . . . . . . . . . . . . 88

KKandawasvika A. . . . . . . . . . . . . . . . 32Kappler W.. . . . . . . . . . . . . . . . . . . . . 2Kersebaum K.-C. . . . . . . . . . . . . . . . 74Kiehle C. . . . . . . . . . . . . . . . . . . . . . . 2Kipfer A. . . . . . . . . . . . . . . . . . . . . . 32Klien E. . . . . . . . . . . . . . . . . . . . . . . 12Köberle A. . . . . . . . . . . . . . . . . . . . . 88Kunkel R. . . . . . . . . . . . . . . . . . . . . . . 2

LLipeck U. . . . . . . . . . . . . . . . . . . . . . 52Lutz M. . . . . . . . . . . . . . . . . . . . . . . 12

MMäs S. . . . . . . . . . . . . . . . . . . . . . . . 32Meiners H.G. . . . . . . . . . . . . . . . . . . . 2Michels I. . . . . . . . . . . . . . . . . . . . . . 74Morchner C. . . . . . . . . . . . . . . . . . . 88

PPeesch R . . . . . . . . . . . . . . . . . . . . . . 88

RReinhardt W. . . . . . . . . . . . . . . . . . . 32

102

SSchätzl P . . . . . . . . . . . . . . . . . . . . . . 74Schlüter M. . . . . . . . . . . . . . . . . . . . . 88Schröder W. . . . . . . . . . . . . . . . . . . . 88Sester M. . . . . . . . . . . . . . . . . . . . . . 52Staub G. . . . . . . . . . . . . . . . . . . . . . . 32Steidl J. . . . . . . . . . . . . . . . . . . . . . . . 74

TThomsen A. . . . . . . . . . . . . . . . . . . . 32Tiedge M. . . . . . . . . . . . . . . . . . . . . . 52

VVetter L. . . . . . . . . . . . . . . . . . . . . . . 88

WWaldow H.v. . . . . . . . . . . . . . . . . . . . 74Wang F. . . . . . . . . . . . . . . . . . . . . . . 32Wendland F. . . . . . . . . . . . . . . . . . . . . 2Wiesel J. . . . . . . . . . . . . . . . . . . . . . . 32Witte J. . . . . . . . . . . . . . . . . . . . . . . . 12

Author’s Index

103

Notes

Information Systems in Earth Management

From Science to Application

Results from the First Funding Period(2002-2005)

GEOTECHNOLOGIENScience Report

No. 8

Information Systems in Earth Management

ISSN: 1619-7399

In Germany the national programme »Information-Systems in Earth Management«has been initiated in 2002 as part of the R&D-Programme GEOTECHNOLOGIEN.Between 2002 and 2005 six joint projects have been funded with about 4 MillionEuro by the Federal Ministry of Education and Research. All projects were carriedout in close cooperation with various national and international partners from aca-demia and industry.

This report highlights the scientific results from this funding period addressing thefollowing objectives:

- Semantical and geometrical integration of topographical, soil, and geological data - Rule based derivation of geoinformation - Typologisation of marine and geoscientifical information - Investigations and development of mobile geo-services- Coupling information systems and simulation systems for the evaluation of trans-

port processes

The GEOTECHNOLOGIEN programme is funded by the Federal Ministry for

Education and Research (BMBF) and the German Research Council (DFG)

No.

8In

form

atio

nSy

stem

sin

Earth

Man

agem

ent

GEO

TECH

NO

LOG

IEN

Scie

nce

Repo

rt