ARTICLE IN PRESS

Journal for Nature Conservation 13 (2005) 185—196

KEYWORDSpin tools;AdvancedclassificatiStructuralassessmenModelling;Spatio-temscale

1617-1381/$ - sdoi:10.1016/j.

�CorrespondE-mail addr

www.elsevier.de/jnc

Multiscale GIS tools for site management

Stefan Langa,�, Tobias Langankeb

aCentre for Geoinformatics (Z_GIS), University of Salzburg, Hellbrunnerstrasse 34, 5020 Salzburg, AustriabUniversity of Nottingham, Centre for Environmental Management, School of Geography, University Park, NottinghamNG7 2RD, United Kingdom

Received 18 November 2004; accepted 9 February 2005

S

on;

t;

poral

ee front matter & 200jnc.2005.02.003

ing author. Tel.: +43 6esses: stefan.lang@sb

SummaryThe concept of the EU Fifth Framework project SPIN (Spatial Indicators for EuropeanNature Conservation) not only included methodological research on case-studies foradvanced classification methods and spatial indicators, but also explicitly incorpo-rated the development of more generic GIS tools. These tools were to provide atangible framework for the applied questions of site management across differentscales and include tools for documentation, visualisation, implementation, changedetection and advanced classification. A survey among SPIN partners and end-usersrevealed a high potential of the various tools when being cautiously integrated withinthe site assessment work flow. In this paper, we focus on operational and technicalissues while, in this respect, complementing the other contributions in this issue. Allof the tools presented can also be applied outside the narrow focus of monitoring forNatura 2000, and their wider application potential in nature conservation issummarised.& 2005 Elsevier GmbH. All rights reserved.

Introduction

The purpose of this article is to give a condensedoverview of the tools and models developed withinthe SPIN project as a tangible and ready-to-use outcome. The paper provides substantialinformation of the product palette to categorisethe outcomes and to describe their specificrequirements, scale-dependency, usability andavailability.

5 Elsevier GmbH. All rights rese

62 8044 5262; fax: +43 662 8044 5g.ac.at (S. Lang), tobias.langank

Quantitative tools for site assessment

Objectivity and transparency in the process ofintegrity assessments of Natura 2000 sites can besupported by quantitative methods (Blaschke,2001), if being applied cautiously. Aiming for thistarget and relying on the latest remote sensing andGIS technology the SPIN project (Spatial Indicatorsfor European Nature Conservation, EVG1-CT-2000-00019) has developed a set of tools and methods

rved.

25.e@nottingham.ac.uk (T. Langanke).

ARTICLE IN PRESS

S. Lang, T. Langanke186

that shows high potential in this respect. The suiteof tools, models and adopted methods (belowsummarised by the term ‘tools’) provides a rangeof supportive links for the work flow of habitatintegrity assessment, as well as site managementand monitoring. Although this process chain wasalready envisaged in the initial concept, it wasrefined and extended throughout the project,finally covering most of the elements of thisprocess. It was intended to provide tangible meansto support the tasks of nature conservationauthorities and, at the same time, offer reprodu-cible and transferable methods. For a subset of thisprocess chain Langanke, Blaschke, and Lang (2004)have already shown the relevance of a consistentset of tools and methods. It was also emphasisedthat the quality of data sets produced in the earlystages of the work flow are vital for the meaningfulapplication of any subsequent analysis.

Following end-user feedback early in the projectit became clear that the large variety of manage-ment and monitoring tasks could only be addressedby a variety of tools. The tools are categorised andcharacterised according to their purposes andsupportive potential as well according to severaloperational aspects. All tools have been testedusing the SPIN case study sites and developed instrong collaboration and feedback loops with end-users.

For both the tool development and their usage inthe site assessment context a certain theoreticalunderstanding of scale is required. Below, weattempt to give a condensed overview of somescale issues relevant in this context.

Spatio-temporal scales of human-inducedimpacts

Human-induced impacts and disturbances onNatura 2000 sites take effect in a wide range oftemporal and spatial scales and may produce acomplex pattern of responses (Forman, 1995; DeLeo & Levin, 1997). For example effects fromglobal climatic change may take place withinseveral decades but will alter the physical appear-ance of a site only in a subtle form, leading to themigration of some species from the initial sites tothe surroundings. Even the complete loss of aspecies suitable ‘‘climate space’’ sensu Hossell,Ellis, Harley, and Hepburn (2003) for a certaincountry or region is possible. These changes may bedetected, for example, by repeated ground surveysor extended time series of aerial photographs orother very-high resolution earth observation (EO)data. On the other hand disturbances could be very

obvious in the short-term. The expansion of aresidential area or an industrial site into aprotected area takes place within a couple ofmonths and is detectable even on medium-spatial-resolution satellite imagery. Direct disturbances,which in fact require the very protection of a site,show a closer link to the scale domain of humanactivities; the respective disturbance patternsoccur on a certain scale range (such as logging,harvesting, agricultural cultivation, nutrition). Inthis respect the human impact will influence aNatura 2000 site up to a certain buffer depth andhuman activities of the same intensity thereforemay have different effects when affecting sites ofdifferent sizes: The percentage of an area con-cerned will be much higher for a small site than fora larger one. Likewise when considering remainingundisturbed habitats of edge-sensitive speciesbeing exposed to disturbances, the relative corearea is a function of the actual size (see Langanke,Rossner, Vrscaj, Lang, & Mitchley, 2005). In generalthe scale and the relevance of human impactsdepends on the size of the affected area, theremaining undisturbed area, the frequency of theimpact or disturbance, the resilience of the habitattype or species under consideration and otherfactors (see Miles et al., 2001).

Relevance of a multi-scale approach

Indicating the conservation status of a Natura2000 site and quantitatively assessing changesrequires the interpretation of an observed patternat a certain scale. The pattern being observed isconsidered to adequately represent the site underinvestigation, but it is actually determined by theextent and the grain applied (O’Neill, DeAngelis,Waide, & Allen, 1986; Wiens, 1989; Turner, Gardner,& O’Neill, 2001). Spatial extent refers to theoverall size of the study area and is usually directlyinfluenced by the extent of the available datamaterial. This especially applies to satellite ima-gery or other EO data that are always provided in arectangular shape and therefore either needs to bemasked or mosaiced. Grain reflects the resolutionof the study, i.e. the individual sample size. Inmany cases this is equal to the ground sampledistance of primary data sources (image data), butcould differ if data sources were merged. Usually,for practical reasons the increase of extent impliesa decrease in grain size. This is reflected in thecharacteristics of primary data sources. For exam-ple a Landsat scene covers an area of 141 km2 witha maximum spatial resolution of 15m, whereas anIKONOS scene covers only 11 km2 but provides a 1m

ARTICLE IN PRESS

Multiscale GIS tools for site management 187

resolution. Both grain and extent, when beingchanged, will usually lead to a remarkable increaseor decrease in spatial variance. Changing thesedimensions therefore has implications for e.g.values of structural indicators based on landscapemetrics, as demonstrated by Wu (2004) andBlaschke and Petch (1999).

Natura 2000 sites, as any ecosystem, are char-acterised by a specific configuration of habitatsthat show steep gradients, which, under certainconstraints regarding generalisation and discretisa-tion, can be used for distinct boundaries that canbe mapped. Applying a certain window of spatialand temporal scale, spatial structures have con-stant frequency behaviour due to stabilising fac-tors. These frequencies are considered to bediscontinuously distributed showing a limited num-ber of time patterns and spatial structures andleading to an organisation in nested systems (Allen& Starr, 1982; Farina, 1998). Wu and Loucks (1995)have used the metaphor of a ‘scaling ladder’ toillustrate the notion of thresholds of adjacent scaledomains in horizontal and vertical dimensionswithin a continuous scale spectrum. Change dy-namics reflect this frequency behaviour of systemsin the form of discontinuous stages and changingsystem behaviours from one stage to another.According to scale or hierarchy theory it is assumedthat a series of scales is inherent in any portion ofland (landscape) no matter what the actual size.O’Neill et al. (1986) thus have postulated that inany study at least three nested scales should beconsidered. This three-level concept is based onthe assumption that the hierarchical conditions andprocess-related interdependencies are only suffi-ciently represented when being investigated in atleast three nested scale-domains. The level ofinterest (‘focal level’) is: (a) constrained by theoverarching conditions of the upper level (providingsignificance) and (b) the specific higher detailcomponents of the lower level that provideexplanations (Turner et al., 2001). In general aflexible yet meaningful set of representations isneeded for representing landscape complexity.Burnett and Blaschke (2003) use the term ‘candi-date discretisations’ for the set of possible scaledrepresentations when working on imaged data thatinherently implies several scale levels.

Difficulties in finding the ‘correct’ scale of thefocal level are encountered when pan-Europeanissues are addressed, as in the case of Natura 2000site assessment. In this case, shall we set the focallevel to the site or to the level of single habitats?Both arguments could be justified by investigatingcontrolling mechanisms and constraints schemes.But even if defining the focal scale at the site level

the fixation of the appropriate scale remains achallenge when assessing Natura 2000 sites all overEurope, due to the high variance of sites that rangein size from several hundred hectares to hundredsof square kilometres.

Methods

Three-level approach

Habitat integrity assessment and the monitoringof ecosystem status on a European-wide levelcannot merely rely on quantifications taken at theEuropean or national scale. It has been one of thecentral foci of the SPIN project to complementaggregated statements on site status with detailedassessments of site dynamics on a local scale level(Weiers, Bock, Wissen, & Rossner, 2004). Guided bythe notion of applying a suite of appropriate scales(Turner et al., 2001) and, at the same time,reflecting user requirements from all relevantinstitutional levels, SPIN research has been con-ducted using a multiple-scale approach on all testareas. Classification, analysis and indicator appli-cation have been performed on three differentscale levels, on a regional level with overviewscales of around 1:250 000, at sub-regional levelswith scales ranging from 1:250 000 to 1:100 000 andlocal level with scales of 1:50 000 reaching down to1:25 000. Likewise the classification scheme hasbeen chosen in order to reflect the detectabilityand interpretability of classes in the respectivescale of investigation. A matching of correspondinglevels in the EUNIS class hierarchy and the spatialscale of classification (EUNIS level 1 3 regionallevel, etc.) could not be accomplished in all studyareas. In some cases EUNIS classes have beenmerged to a combined class at the local scale.

Tool description

The tools that have been developed can begrouped into five categories: (1) change detection;(2) advanced classification; (3) documentation andselection; (4) implementation and application; (5)scenarios and modelling. Each category coversspecific phases within the site assessment andmonitoring process. Figure 1 illustrates how thedescribed tools are embedded in an optimisedoverall work flow.

For categorisation and a short description, aswell as operational aspects (such as information oninstitution and principal investigator, accessibility

ARTICLE IN PRESS

Figure 1. The use of SPIN tools in an optimised site assessment work flow. Transparency and objectivity is the aim in allsteps of the evaluation process. The Change Tool highlights changes and sets the focus of investigation; advancedclassification methods provide a reliable base map; V-LATE is used for the actual assessment step after suitableindicators are selected. Finally modelling tools are used for developing and communicating scenarios and modellingfuture trends which may take place within the sites under investigation.

S. Lang, T. Langanke188

and documentation, and additional software re-quirements) see Table 1.

Change detectionA picture on change dynamics over time is one of

the prerequisites for a targeted site management.Change detection therefore marks the startingpoint in the site assessment chain. Yet, change assuch is a relative concept and depends on sitespecific thresholds to distinguish real and relevantchanges from a natural variability, seasonalchanges or image noise. The Change Tool indicateschanges over time on a regional to sub-regionalscale and highlights hot spots for further investiga-tion. It operates on medium-resolution image dataof the Landsat type that reflect changes visible on arelatively aggregated spatial scale. Methodologi-cally the Change Tool uses a pre-classificationchange detection technique based on principalcomponent analysis; for the interpretation ofresults a simple threshold operation on changeprobability layers can be performed to createcategorical change classes (for further discussionon specific potential and limitations, see Kleinod,Wissen, & Bock, 2005).

Advanced classificationThe suite of advanced classification tools enables

the user to obtain evidence on habitat quality,structure and habitat distribution within a Natura2000 site. The tools operate on medium- to high-resolution image data sets. A variety of approacheshave been employed that reflect state-of-the-artimage processing techniques for this type of multi-spectral data. All tools have been rigidly tested onvarious case-study sites, changing scales and the-matic resolution and comparing the results. Whenused under appropriate conditions, they providereliable categorical base maps for specific purposesin a semi-automated and transparent way. Some ofthese methods support the integration of externalexpert knowledge in the classification process. Thisis a major advantage, as the interpretation ofhabitat integrity and the conservation status withinNatura 2000 sites needs a complex understanding ofthe relevant issues that can often only be capturedby using the knowledge of local experts. Sinceexpert knowledge about certain characteristics ofthe classes is sometimes intuitive and vague,softened rules based on fuzzy logic can be used tocontrol the classification process. Other methodsare based on machine-learning algorithms ratherthan on production systems. These are supposed to

ARTIC

LEIN

PRES

S

Table 1. SPIN tools, methods, models — operational aspects and tool description (after Klug et al., 2004. modified and complemented)

ID Name (Acronym) and Institution (PI) Type Accessibilitya Docu Platform specsb Short description

Tool Method/model

Freedownload

Restricteddownload

Internaluse

Built-in help/manual

Publication

Change detection1 Change Tool (DLR-DFD,

michael.wissen@dlr.de)| | | | ERDAS Imagine 8.* The Change Tool is an automated pixel

oriented change detection approach basedon selected dual date spectral bands and/or indices to indicate and flag changedareas by their probability of change.

Advanced classification2 Kernel-based Reclassification Tool (NOA,

kontoes@space.noa.gr, ikeram@cc.uoa.gr)| | | None Kernel based re-classifier takes into

account the spatial arrangement andfrequency of spectral labels present withina predefined square kernel. It examineslabels of adjacent pixels within the squarekernel and calculates an adjacency-eventmatrix, accounting for the spatialarrangement and frequency of the labels.Criterion for pixel re-labelling is the degreeof match between the adjacency eventmatrix and the template matrices producedduring training.

3 Case-based Reasoning Tool (IGUT,kalle.remm@ut.ee)

| | | None Case based approach claims to be auniversal tool for modelling statisticalrelationships and giving predictions basedon any training data. CBR approach prefersraw data to generalisations. Case-basedecological mapping relies on theassumption of finding a species (or otherphenomenon) in locations similar to thosewhere this species has already beenregistered. The CBR approach consists oftwo stages, machine learning of weights offeatures and training instances andprediction by the most similar exemplars.

4 LocalStatistics (IGUT, kalle.remm@ut.ee) | | | | None The program is developed mainly forcomparing different indices of spatialpattern and different kernel size in case ofdifferent explanatory variables anddependent variables. The comparisons ofcharacteristics of spatial pattern in imagesand maps are needed in order to find thebest combinations of source data, kernelsize and pattern descriptors with thehighest indicator values.

Multiscale

GIS

toolsfor

sitemanagem

ent189

ARTIC

LEIN

PRES

S

Table 1. (continued )

ID Name (Acronym) and Institution (PI) Type Accessibilitya Docu Platform specsb Short description

Tool Method/model

Freedownload

Restricteddownload

Internaluse

Built-in help/manual

Publication

5 Object-relationship Habitat Modelling(Z_GIS,tobias.langanke@nottingham.ac.uk,stefan.lang@sbg.ac.at)

| | This method utilises object-based imageanalysis for fine-scaled mapping of habitatstructures. By encoding expert knowledgein a fuzzy rule system, the process ofclassifying becomes more objective andtransparent. The geometry of the semi-automated delineated objects has beenproduced by a rule set that can be fullyreconstructed, if needed. As a productionsystem the outcome strictly depends on theunderlying object relationship model, but isalso flexible for adaptations by slightlychanging the parameterisation.

Documentation and selection6 Metadata Tool (NOA, sifakis@space.noa.gr,

ikeram@cc.uoa.gr)| | Oracle Forms Server 6.5.1 The tool has been providing the central

data pool for EO and non-EO data for theentire SPIN consortium. In detail this meansto provide comparable data sets to allpartners by application of common pre-processing methods (radiometric,atmospheric and geometric correction), toprovide baseline land cover maps forfurther use by the partners, to develop andmaintain metadata bases for satellite andground truth data.

7 Indicator Database for Scientific Exchange(IDEFIX) (Z_GIS, hermann.klug@sbg.ac.at)

| | | | MS Access required Main scientific objectives of IDEFIX are togive an overview of existing landscapemetrics; to compare different metrics toshow the possibilities, limits and behaviourof the metrics as well as their potential toquantify structural changes and conditionsin spatial data; to help to interpret metricsthat are available and implemented invarious software packages; to set up anappropriate selection process of metricsand to develop semi-automatic proceduresfor indicator selection.

8 Metadata Application for Environmental GISData (MEnDat) (CSES, borut.vrscal@bf.uni-lj.si)

| | | MS Access 2000 required.ArcView 3.x or ArcGIS 8.xoptional

MEnDat is a stand-alone menu drivenapplication, which incorporates thedatabase system of data describingenvironmental threats, impacts and mutualdependencies. The database offers links toGIS maps—data layers using commercial GIS

S.Lang,

T.Langanke

190

ARTIC

LEIN

PRES

S

software or the maps can be previewedusing incorporated MEnDat map preview.

Implementation and application9 Vector-based Landscape Analysis Tools

Extension (V-LATE) (Z_GIS,stefan.lang@sbg.ac.at,dirk.tiede@sbg.ac.at)

| | | | ArcGIS 8/9.x The tool performs structural assessment onthe basis of vector data and contains a setof structural indicators to addressstructure-related ecological issues. Itaddresses the main aspects of structuralpattern analysis in seven categories (area-,form-, interior-, edge-, proximity-,diversity- and subdivision analysis).

10 Interactive Metrics Tool (IMT) (Z_GIS,stefan.lang@sbg.ac.at)

| | | | MS Powerpoint, Excel, ArcView3.x

The IMT provides a collection of commonlyused landscape metrics for the utilisation ofvector-based landscape structure analysison any level. Besides its analyticalfunctionality it provides insight into thecalculation procedure and gives access toimplemented formulas through its openstructure.

Scenarios and modeling11 Model of spontaneous afforestation (SFI,

andrej.kobler@gozdis.si)| | | Idrisi 32 This scenario-based spatial analysis tool

rests upon a spatially explicit model of atype of land cover transformation process.Calculations of three following targetvariables is supported: afforestation indexAI; afforestation amount AA; afforestationproximity AP. These variables can beanalysed in regression models andcompared amongst themselves or betweendifferent time frames. Furthermore policyvariables like stockbreeding intensity canbe processed.

12 Habitat Model Brown Bear (SFI,andrej.kobler@gozdis.si)

| | |

13 Habitat Model Capercaillie (ESAP,anne.jacquin@esa-purpan.fr)

| |

aThe development of the tools has been financially supported by the SPIN project (contract number EVG1-CT-2000-00019). However in some cases download from the internet may berestricted due to various reasons. Please contact the respective person (principal investigator, PI) for further details.bIncludes comments on specific platform and software requirements (if not stated differently, the tools work in a Microsoft Windows environment).

Multiscale

GIS

toolsfor

sitemanagem

ent191

ARTICLE IN PRESS

S. Lang, T. Langanke192

incrementally improve the outcome by adaptivelearning, every time new cases are processed.Satellite imagery (or aerial photography) is usuallynot limited to the site boundaries; image dataprovide seamless data on the surrounding area,taken at the same time and with the same spectralproperties. Information about the area surroundinga site is essential for an integrated assessmentaccording to the pressure-state-response concept(for an overview of advanced habitat classificationmethods, see Bock, Xofis, Mitchley, Rossner, &Wissen, 2005).

For the purpose of combining data sets fromdifferent sources, which may be used as explana-tory variables, the LocalStatisitcs tool has beendeveloped. The tool provides means to optimise theintegration of different data types and comparesdifferent indices of spatial pattern with differentkernel sizes (for more details, see Remm, 2005).

Documentation and selectionThis category of tools comprises two databases

giving a content-related framework on the other-wise technical solutions provided by SPIN. Thedatabases support the end-user with a collection ofuseful background information regarding the ap-propriate usage of the tools and the variety ofpotential indicators. At the same time we aim toprevent users from un-reflected usage of some ofthe tools or methods by providing essential in-formation and recommendations e.g. on the usageof structural and functional indicators.

IDEFIX (indicator database for scientific ex-change) provides relevant information to deter-mine the applicability of specific metrics in theframework of Natura 2000 and therefore maycontribute to a standardisation and automation ofparts of the monitoring obligations. In forming aninventory of landscape metrics the database facil-itates recherche and assists in interpretation. Itmay at the same time also increase awareness onthe potential of structural indicators for supportingdaily tasks in nature conservation. By investigatingthe database, users will get more familiar with theunderlying metrics, their behaviour, biases, andlimitations. Searches for appropriate measures arefacilitated by features stored in the databaselinked to specific ecological issues.

The basic objective of the functional indicatordatabase MEnDat (metadata application for envir-onmental GIS data) is to provide additional non-technical information on GIS data sets. By linkingdescriptive data sets to the corresponding GISlayers, the relationships and mutual dependenciesbetween environmental processes and respectivedata sets become clearer to the user (for further

discussion on the usage of structural and functionalindicators, see Langanke et al., 2005).

Implementation and applicationAmong the tools of this category vector-based

landscape analysis tools extension (Figure 2) (V-LATE) has already been widely and successfullyused internationally for various purposes in thecontext of landscape structural assessment. Morethan 500 downloads within 6 months indicate aremarkable interest in this ready-to-use ArcGISplug-in. Complemented by the interactive metricstool (IMT), a didactical tool for learning andteaching, V-LATE offers a set of structure-relatedmeasures that cover profound aspects of thequantitative assessment of habitat structure, suchas core area, fragmentation and integration. Itdirectly works on polygon vector data rather thanon raster data sets. This tool can therefore readilybe used for most of the existing data sets in natureconservation (which tend to be available in vector,rather then in raster format), in all cases wherespatially explicit analysis of arrangements is re-garded as being essential (for more details seeLanganke et al., 2005). Within the assessmentchain, V-LATE is meant to be used for a spatialanalysis of classification results. The method ofclassification itself is (technically) not relevant aslong as vector data are produced. However wetested and developed the tool in the context ofobject-based habitat mapping (Langanke et al.,2005).

Scenarios and modellingTools and methods for scenarios and modelling

help to predict certain developments under varyingconditions or assumptions. They assist in forecast-ing the consequences of persistent pressures on asite or help to demonstrate which effects certainmanagement activities may have. Pressure stateslike spontaneous afforestation (SA) can be analysedin terms of their future behaviour and impact andpotential mitigation strategies can be derived. Onegroup of this category comprises GIS-based modelsfor predicting the likely course of SA of abandonedfarmland and thus do offer decision support inlandscape management. The other group consistsof two habitat suitability models for specific targetspecies, of which both are FFH annex II species(brown bear and capercaillie). The models help tounderstand the interrelationships of pre-conditionsthat are required for the maintenance of suitablehabitats. (For further discussions on habitat model-ing, see Jacquin et al., 2005; Kobler, Cunder, &Pirnat, 2005).

ARTICLE IN PRESS

Figure 2. V-LATE for structural site assessment, implemented in ArcGIS 9.

Multiscale GIS tools for site management 193

Tool survey

A comparative qualitative survey has been con-ducted among the SPIN partners being involved inthe tool development. This survey referred to dataformat requirements for input data and the type ofoutput data being produced. It furthermore col-lected information about the main target scale tobe addressed and incorporates user feedback interms of various critical aspects of user validationfor the respective tools.

Results and discussion

Tools overview

The main results of the tool survey are sum-marised in Table 2. This table basically reveals thefollowing aspects:

�

Scale dependency of the tool in a strict sense,(i.e. the restriction of application to specificscale levels due to the inherent scale of the inputdata), only applies for the Change Tool and the

Brown bear habitat model, a small subset of thetools. All other tools either show scale depen-dency in the sense of an optimum scale (e.g.object-based habitat modelling shows its poten-tial especially in fine-scaled applications) or theydo not imply any scale restriction. But even iftools are applicable in any scale, the results andproduced figures may have different meanings indifferent scales.

�

All tools require a specific type of input data andlikewise produce certain output data types.Some of the tools use a combination of inputdata types and/or produce several types ofoutput data. V-LATE for example requires (la-belled) polygon data as input, and producespolygon data, figures and tables according to theselected analysis function. It has to be noted thatthe meaningfulness of results that are producedby any of the tools is strictly dependent on thequality of the input data. The propagation oferrors (relating to interpretation accuracy, geo-metrical accuracy or semantic aspects) is ob-vious, yet sometimes disregarded when usingquantitative tools.

�

Validation from a user’s perspective impliesthree components: (1) the first concerns the

ARTIC

LEIN

PRES

S

Table 2. SPIN tools, methods, models—input and output data, scale dependency and user validity

ID Inputa Output Scaleb Validationc

Medium res.image data

High res.image data

Rasterdata

Vectordata

Text Tabulardata

Images Rasterdata

Vectordata

Figures Tables Reports Regional Sub-regional

Local Needs extra dataor parameterisation

Is supportive Is alreadyin use

Relies onuser input

Expertknowledge

Supportneeded

1 | | | | | | | |2 | | | | | | |3 | | | | | | | |4 | | | | | | |5 | | | | | | | | | |6 | | | | | | |7 | | | | | |8 | | | | | | | | |9 | | | | | | | | | |

10 | | | | | | | | |11 | | | | | |12 | | | | | |13 | | | | | | | |

aInput data comprise data on which the tool or method can be applied or data to feed a model.bScale refers to the target scale for which the tool is designed or optimised.cValidation reflects general experiences of partners when introducing or discussing the tools with end-users as well as feedback from user surveys.

S.Lang,

T.Langanke

194

ARTICLE IN PRESS

Multiscale GIS tools for site management 195

usability of the tools, their supportiveness andpotential for various tasks related to Europeannature conservation. All tools were regarded asbeing supportive or having a high potential. Thisis not to be seen mutually exclusive, it ratherreflects the fact that the purpose or immediatebenefit has sometimes to be projected (‘has highpotential’), rather than seen (‘is supportive’).Both categories are merged together in the Table2. The second component emphasises that inmost cases the user must actively contributetowards using the tool in an appropriate way.Parameterisation or the collection of additionaldata (usually species data) is sometimes re-quired. Likewise in some cases expert knowledgemust be explicitly integrated. (3) Finally theindicator databases have to be subsequentlyfilled with content; the user input about experi-ences and knowledge is vital for the tools.

Usability of tools in the context of specificmanagement tasks

Baskerville (1997) pointed out the need for apolicy-relevant outcome of ecological research, inother words for research not remaining at a meredescriptive or explaining level. Results are ex-pected to facilitate management tasks and tosupport informed decisions before interventionstake place. This ‘reflection-before-action’ theoremhas also guided the tool design phases in SPIN,providing a range of models, methods and tools foradaptive management and a better forecast of thesystem behaviour on the respective test sites. Eachof the developed tools has a clear component ofmanagement support; thus the suite of tools coversconsiderable parts of the entire assessment chain.

Whereas tools may be of high potential for (moretechnically oriented) users, others who are deeplyinvolved in everyday business may be overstrainedand only see additional workload. The followingsection tries to summarise the major advantagesand challenges of the tools in the context of habitatintegrity assessment.

The Change Tool supports basic monitoring tasksthrough its high repeatability and cost-effective-ness (data costs against area being covered), andtherefore is regarded as being a valuable instru-ment for identifying hot spots for further investiga-tion. Flagging hot spots of change or extractingspecific gradual changes in particular bands forselected areas is an efficient instrument formonitoring tasks, planning and realisation mappingactivities and the assessment of site managementactivities. On the other hand, interpretation of the

results requires a certain level of expertise inimage processing. At the momentary stage thechange tool is optimised for pre-classificationchange detection medium spatial resolution EOdata.

Indicator applications can seldom be based on agiven and comparable data basis that already existsacross the case study areas. Therefore it isnecessary to derive a categorical map base, e.g.from EO data using accurate and transferablemethods that implement existing site knowledgeas much as possible. Research within SPIN thereforeprovides a methodology to cover the whole workflow starting from a set of EO data, throughadvanced classification and mapping to the devel-opment and application of spatially explicit indica-tors on the derived data sets. The remaininglimitations here are only to a lesser degree oftechnical nature, but concern instead the inter-pretation of indicator values. This is especiallyrelevant in the case of structural indicators thatrely in their interpretation on partly unknownpattern-process relationships.

Because of their generic character the GIS toolsand methodologies discussed in this paper are notrestricted to site management in the Natura 2000context, but show potential for a wider scope ofapplications. Biodiversity conservation outside pro-tected sites is of increasing importance and can besupported by many of the tools introduced.

Conclusions

Within the SPIN Project we have provided atoolbox of GIS tools and methodological approachesto support monitoring and management in natureconservation in a very practical sense. Some of thetools have already gained a high interest from theCommission, especially all those supporting changedetection and providing indicators for spatio-structural arrangement of habitats. We haveprovided methodological guidance in advancedhabitat classification, greatly facilitated the selec-tion and calculation of landscape metrics for usersof polygon vector data and added a dynamicdimension with a habitat modelling component.However some of the indicators developed needfurther refinement and research before they can beoperationally implemented in the work flow of theend-users on different user levels. These openresearch questions and indicator refinement con-cern the different tools to varying degrees. Itexplicitly refers to the following issues: (a) specificsoftware and/or technical knowledge is necessary

ARTICLE IN PRESS

S. Lang, T. Langanke196

in some cases, (b) interpretation of obtained valuesmight be difficult especially for using some indica-tors derived using V-LATE. Future developmentscould also focus on the development of more user-friendly and very specific applications for certaintypes of user requirements. This would be mostpromising if closely referring to an emergingde-facto standard for Natura 2000 monitoringmethodology.

Acknowledgements

The work conducted has been co-financed by theEU project SPIN (Spatial Indicators for EuropeanNature Conservation, Contract No. EVG2-2000-0512). The collaboration between the SPIN teamand the various end user institutions is kindlyacknowledged. We thank all SPIN partners for theirrapid response on the tool survey questionnaire.

References

Allen, T. F., & Starr, T. B. (1982). Hierarchy perspectivefor ecological complexity. Chicago: University ofChicago Press.

Baskerville, G. L. (1997). Advocacy, science, policy, andlife in the real world. Conservation Ecology, 1(1), 9.[online] Available from the Internet. URL: http://www.consecol.org/vol1/iss1/art9 (11/2004)

Blaschke, T. (2001). Environmental monitoring andmanagement of protected areas through integratedecological information systems — an EU perspective.In C. Rautenstrauch, & S. Patig (Eds.), Environmentalinformation systems in industry and public adminis-tration (pp. 75–100). Hershey, London: Idea GroupPublishing.

Blaschke, T., & Petch, J. (1999). Landscape structure andscale: Comparative studies on some landscape indicesin Germany and the UK. In M. Maudsley, & J. Marshall(Eds.), Heterogeneity in landscape ecology: Patternand scale (pp. 75–84). Bristol, UK: IALE.

Bock, M., Xofis, P., Mitchley, J., Rossner, G., & Wissen, M.(2005). Object oriented methods for habitat mappingin multiple scales: Case studies from NorthernGermany and North Downs. GB. Journal for NatureConservation, 13, 75–89.

Burnett, C., & Blaschke, T. (2003). A multi-scalesegmentation/object relationship modelling metho-dology for landscape analysis. Ecological Modelling,168, 233–249.

De Leo, G. A., & Levin, S. (1997). The multifacetedaspects of ecosystem integrity. Conservation Ecology,1(1), 3 [online] Available from the Internet URL:http://www.consecol.org/vol1/iss1/art3 (11/2004).

Farina, A. (1998). Principles and methods in landscapeecology. London: Chapman & Hall.

Forman, R. T. T. (1995). Land mosaics: The ecology oflandscapes and regions. Cambridge: Cambridge Uni-versity Press.

Hossell, J. E., Ellis, Z. E., Harley, M. J., & Hepburn, I. R.(2003). Climate change and nature conservation:Implications for policy and practice in Britain andIreland. Journal for Nature Conservation, 11, 67–73.

Jacquin, A., Cheret, V., Denuxa, J. P., Gaya, M., Mitchley,J., & Xofis, P. (2005). Habitat suitability modelling ofCapercaillie (Tetrao urogallus) using Earth Observationdata. Journal for Nature Conservation, 13, 161–169.

Kleinod, K., Wissen, M., & Bock, M. (2005). Detectingvegetation changes in a wetland area in NorthernGermany using earth observation and geodata. Jour-nal for Nature Conservation, 13, 115–125.

Klug, H., Langanke, T., Lang, S., Keramitsoglou, I.,Kontoes, H., Haffner, C., Sifakis, N., Vrscaj, B.,Persolja, J., Kobler, A., Cunder, T., & Hocevar, M.(2004). GIS management and decision tools, SPINproject deliverable (Contract No. EVG2-2000-0512).

Kobler, A., Cunder, T., & Pirnat, J. (2005). Modellingspontaneous afforestation in Postojna area. Slovenia.Journal for Nature Conservation,, 13, 127–135.

Langanke, T., Blaschke, T., & Lang, S. (2004). An object-based GIS/remote sensing approach supporting mon-itoring tasks in European-wide nature conservation.Proceedings of the first Mediterranean conference onearth observation, Beograd (pp. 252–254).

Langanke, T., Rossner, G., Vrscaj, B., Lang, S., & Mitchley,J. (2005). Selection and application of spatial indica-tors for nature conservation at different institutionallevels — experiences from the SPIN project. Journal forNature Conservation, 13, 101–114.

Miles, J., Cummins, R. P., French, D. D., Gardner, S., Orr,J. L., & Shewry, M. C. (2001). Landscape sensitivity:an ecological view. Catena, 42, 125–141.

O’Neill, R., DeAngelis, D., Waide, J., & Allen, T. (1986). Ahierarchical concept of ecosystems. Princeton: Prin-ceton University Press.

Remm, K. (2005). Correlations between forest standdiversity and landscape pattern in Otepaa NaturePark. Estonia. Journal for Nature Conservation,, 13,137–145.

Turner, M., Gardner, R., & O’Neill, R. (2001). Landscapeecology in theory and praxis. Pattern and processes.New York: Springer.

Weiers, S., Bock, M., Wissen, M., & Rossner, G. (2004).Mapping and indicator approaches for the assessmentof habitats at different scales using remote sensingand GIS methods. Landscape and Urban Planning, 67,43–65.

Wiens, J. A. (1989). Scaling of ‘landscapes’ in landscapeecology, or, landscape ecology from a beetle’sperspective. Landscape Ecology, 3, 87–96.

Wu, J. (2004). Effects of changing scale on landscapepattern analysis: Scaling relations. Landscape Ecology,19, 125–138.

Wu, J., & Loucks, O. L. (1995). From balance-of-nature tohierarchical patch dynamics: A paradigm shift inecology. Quarterly Review of Biology, 70, 439–466.