園林景觀中英文對(duì)照外文翻譯文獻(xiàn)

中英文對(duì)照外文翻譯文獻(xiàn)(文檔含英文原文和中文翻譯)Effectsoftopographyandsurfaceroughnessinanalysesoflandscapestructure–AproposaltomodifytheexistingsetoflandscapemetricsAbstractTopographyandreliefvariabilityplayakeyroleinecosystemfunctioningandstructuring.However,themostcommonlyusedconcepttorelatepatterntoprocessinlandscapeecology,theso-calledpatch-corridor-matrixmodel,perceivesthelandscapeasaplanimetricsurface.Asaconsequence,landscapemetrics,usedasnumericaldescriptorsofthespatialarrangementoflandscapemosaics,generallydonotallowfortheexaminationofterraincharacteristicsandmayevenproduceerroneousresults,especiallyinmountainousareas.Thisbriefmethodologicalstudyprovidesbasicapproachestoincludereliefpropertiesintolarge-scalelandscapeanalyses,includingthecalculationofstandardlandscapemetricsonthebasisof“true”surfacegeometriesandtheapplicationofroughnessparametersderivedfromsurfacemetrology.Themethodsaretestedfortheirexplanatorypowerusingneutrallandscapesandsimulatedelevationmodels.Theresultsrevealthatareaanddistancemetricspossessahighsensitivitytoterraincomplexity,whilethevaluesofshapemetricschangeonlyslightlywhensurfacegeometriesareconsideredfortheircalculation.Insummary,theproposedmethodsprovetobeavaluableextensionoftheexistingsetofmetricsmainlyin“rough”landscapesections,allowingforamorerealisticassessmentofthespatialstructureKeywords:patch-corridor-matrixmodel,neutrallandscapes,digitalelevationmodels,relief,roughnessparameters1.Introduction–The“3D-issue”inlandscapeecology“3D”hasbecomeafrequently-usedterminmanyfieldsofscience,eveninecology.3D-visualisationand3D-graphicshaveundergoneenormousadvancementsintherecentyears.Realisticvisualisationsoflandscapesorcitiesaregainingimportanceinspatialplanningprocesses,forexamplewhentheimpactofconstructionprojectsistobeclarifiedorwhenthedynamicsoflandscapechangeovertimearetobedemonstrated.Also“3D-GIS”isbeginningtoemerge.However,“3D-analysis”inlandscapeecologyandtheexaminationof“3D-patterns”arestillsomewhatneglected,eventhoughelevationandlandsurfacefeaturescanberegardedaskeyelementsinmanyecologicalprocesses.Thus,fromalandscapeecologicalperspective,thereisaneedfor3D-analysisintermsoftheexaminationandcharacterisationofthetopographyoflandscapesandspecificterrainfeatures.Previouspublicationshavetriedtohighlightthenecessitytoincorporateaspectsofthethirddimensionintolarge-scalelandscapeanalyses.Manyauthorshavepointedoutthattopographyisafactorwhichmayplayakeyroleinecosystemfunctioningandstructuring,andwhichinmanycasesisnotsufficientlytakenintoaccount.Theconnectionbetweensurfacecharacteristicsandboththespeciesrichnessandcompositioninvascularplants(asshowninBurnettetal.1998;Davis&Goetz1990;Sebastiá2004)isawell-knownfactthathasfrequentlybeenusedforthedesignofbiodiversitydistributionmodels(e.g.Bolstadetal.1998).Theimpactofreliefonthedifferentiationofanecosystemasawholeandonparticularecologicalfunctionssuchassoilmoisture,temperaturedistribution,thebalanceofsolarirradiation,ormicroclimatehasbeendescribedindetailaswell(Bailey2004;Oke1978;Swansonetal..Bearingthesewell-studiedlinksbetweenterrainfeaturesandecologicalprocessesinmind,onefactappearstobenoteworthy:Sofar,thewell-establishedpatch-corridor-matrixmodel–assuggestedbyForman(1995)–doesnotexplicitlyconsiderthethirddimensioninitsapproachtodescribethespatialarrangementoflandscapes.Itisgenerallyacceptedthattheacknowledgementofthe“effectofpatternonprocess”(Turner1989)constitutestheself-conceptionofmodernlandscapeecology.Buttheestablisheduseoflandscapemetricsforthecharacterisationofgeometricandspatialpropertiesofcategoricalmappatterns(Mc-Garigal2002)holdsaviewofthelandasa“planimetric”surface–aspectsofthree-dimensionalpatterns(topography,elevation)havenotyetexpandedintothisconcept(seealsoBlaschke&Dr?gu?2003).Alargenumberoflandscapemetricshasbeendescribedindetailandusedforseveralpurposes,includingspatialplanningorecologicalmodelling.Alotofuser-friendlysoftwareproductsforthecomputationofsuchmetricsonthebasisofeithervectororrasterdataareavailable,e.g.FRAGSTATS(McGarigal&Marks1995),LeapII(Schnekenburgeretal.1997)orV-LATE(Lang&Tiede2003).Butinformationabout3D-featureslikesurfaceroughness,landform,orreliefvariabilitywithinlandscapeelements(“patches”)cannotbemadeaccessibleusingthesemeasures.Moreover,onemayevenyielderroneousresultsfromthecalculationoflandscapemetrics,sincethebasicgeometries(area,perimeter)ofpatchesanddistancesbetweenthemaregenerallyunderestimatedinplanimetricobservationsbyneglectingtheunderlyingrelief.Thesediscrepanciesbetweenthepatch-corridor-matrixmodelandtheactualconditionswithinlandscapescanberegardedasamajordrawbackoftheconcept,especiallyinmountainousregionsorinareasexhibitingacomplexterrain.Figure1providesavisualrepresentationoftheeffectsthatreliefmayhaveontheparameteroutputofcommongroupsoflandscapemetrics.Forexample,itisapparentthatforareaordistancemetrics,adefinitetendencytowardshighervaluescanbeexpectedwhensurfacecomplexityistakenintoaccount.Geomorphologyoffersalargesetofparameterstodescribethelandsurfaceandtoclassifythegeorelief(seeDikau&Schmidt1999;Evans1972;Pike2000;Wilson&Gallant2000).However,measuresofcurvature,aspect,slope,orcombinedparameterssuchaswetnessindicesareonlyoflimitedusewhenonetriestocharacterisethespatialpatternoflandscapesusingcategoricalmaps.Thesemeasuresingeneralrelatetocatchmentareasanddiscretelandformelements,ratherthantothe“patches”thatcommonlandscapemetricsareappliedto;acompatibilitybetweentheseapproachescannotbetakenforgrantedineverycase.Figure1:Commonlandscapemetricsgroupsusedinthepatch-corridor-matrixmodelandtheeffectsofunderlyingterrain(upperpartredrawnaccordingtoWiensetal.1993).Othertechniquestoincorporatesurfacefeaturesintolandscapeanalyseshavebeenproposed.Forexample,Beasom(1983)hassuggestedasimplemethodforassessinglandsurfaceruggednessbasedontheintersectionsofsamplepointsandcontourlines.MoreelaborateproposalstoincludetopographiccharacteristicsintoanalysesoflandscapepatternandofvegetationdistributionshavebeenmadebyDorneretal.(2002).Simplemoving-windowalgorithmsforestimationsofthe“concavity/convexity”ofrasterpixelsindigitalelevationmodels(DEM)havebeendevelopedbyMcNab(1992)andBlaszczynski(1997).Whileanapplicationoftheseapproachesforspecialcasestudiesandparticularthematiccontextsmaybeveryvaluable,integrationintothepatch-corridor-matrixmodelhasnotbeenachievedyet.Meanwhile,thetechnologicalprogressinthefieldofremotesensinghasledtoarapidimprovementinthequalityofDEMs.EspeciallyLiDAR(“l(fā)ightdetectionandranging“)measurementsprovidehigh-resolutionelevationdataofthelandsurface.Theycanaccuratelyestimateattributesofvegetationstructureandshouldthereforebeofparticularinteresttolandscapeecologists(Lefskyetal.2002).Firstattemptstoderive3DlandscapemetricsfromLiDARdatahavealreadybeenmadeearlier(e.g.Blaschkeetal.2004).Allthesenotesonthe“3D-issue”inlandscapeecologyandtheshortcomingsintheanalysisofimportantsurfacefeaturesmarkthestartingpointforthestudyathand.Themainpurposeofthispaperistopresentsomebasicprinciplesonhowtosolvetheproblem,basedontherecognitionofthediscretelandunitasacentralconceptinlandscapeecologicalhypotheses(Zonneveld1989).Theterm“3D”isusedinthiscontext,eventhoughdigitalelevationmodelsactuallyrefertoa“2.5D”representationoftherealworld,withonez-valueassociatedwitheachx,y-coordinate.Inmostcases,however,DEMscanbeconsideredassufficienttoprovideanapproximationofthetruesurfaceconditions.Thismethodicallyorientedarticleattemptstorevealandquantifytheeffectsthatthevariabilityandroughnessofthelandsurfacemayhaveontheparametervaluesofcommonlandscapemetricsandtriestopresentafewsuitableworkaroundsforthisissue.Theseincludemodificationalgorithmsforcommonlandscapemetricsaswellastheintroductionofalternativemeasurestocapturesurfaceroughness.Thesemethodsaremainlyexemplifiedusingneutrallandscapemodels.2.Methods–Consideringterraincharacteristicsinthepatch-corridor-matrixmodelTwobasicapproachesforthefirststepstowards3D-analysisoflandscapestructureareproposedinthispaper:Thefirstonecomprisesdifferentcorrectionalgorithmsforstandardarea,shapeanddistancemetrics.Thesecondoneisbasedontheaggregationofheightinformationintheformofsimple“surfaceroughness”parameters.2.1AdjustingstandardlandscapemetricsThesimplestandmostobviousapproachtoincorporatethethirddimensionintolandscapeanalysesistoadjusttheexistingsetofmetricsandtomitigatethesourceoferrorassociatedwiththeplanimetricprojectionofslopes.SuchtechniqueshavebeenproposedearlierbyDorneretal.(2002),whosuggestedtocomputethetruesurfaceareaofeachrastercellinaDEMbythequotientprojectedarea/cos(slope)andtoapproximatethetruedistancesbetweenadjacentcellsbysimpleapplicationofthePythagoreanTheoremusingEuclideandistanceanddifferencesinelevation.However,asystematicintegrationintocalculationalgorithmsoflandscapemetricswasnotpresented.Inthispaper,amoredetailedapproachischosentocalculatetruesurfacearea,basedonthefindingsofJenness(2004).Thetechniqueisbasedonamovingwindowalgorithmandestimatesthetruesurfaceareaforeachgridcellusingatriangulationmethod(Figure2).Eachofthetrianglesislocatedinthree-dimensionalspaceandconnectsthefocalcellwiththecentrepointsofadjacentcells.ThelengthsofthetrianglesidesandtheareaofeachtrianglecaneasilybecalculatedbymeansofthePythagoreanTheorem.Theeightresultingtrianglesaresummeduptoproducethetotalsurfaceareaoftheunderlyingcell.Thismethodispreferredsinceitcanbeexpectedtoprovidemoreaccurateresults;incontrasttotheapproachmentionedabove,alleightneighboursofthepixelofinterestareincludedinthecalculation,insteadofonlytheonedefiningtheslopeangle.Figure2:Methodtodeterminetruesurfaceareaandtruesurfaceperimeterofpatches.Truesurfaceareaofthefocalrastercellisobtainedbyaddingtheeightshadedtriangles,truesurfaceperimeterbysummationoftheeightboldlinesegments(figureredrawnaccordingtoJenness2004).Additionalcomputationstepshavetobeconductedtoobtainthetruesurfaceareanotonlyforeachrasterpixelbutforeachpatchinalandscapeinordertoincludethesenewgeometryvaluesintothecalculationofcommonlandscapemetrics.Arasterfilecontainingthepatchstructureoftheconcerninglandmosaicisoverlaidwiththecorrespondingelevationmodel.Thensurfaceareavaluesofthepixelsrepresentingeachpatcharesummedup.Equalresolutionandextentofthepatchfileandtheelevationmodelarepresumed.Jenness’methodisalsoadaptedinordertocalculaterealisticsurfaceperimetersofeachpatch.Thisisdonebysimplyaddingupthelinesegmentsformingthesurfaceedgeoftherasterpixels(seeboldlinesinFigure2)incasetheyarepartofthepatchboundary.Amoreintricateprocedureisneededforthecalculationofthetruesurfacedistancesbetweenpatchesofthesameclass,thatisthe3D-equivalenttothe“EuclideanNearestNeighbour”measureasusedintheFRAGSTATS-setofmetrics(seeMcGarigaletal.2002).Thequestioncanbereferredtoasaso-called“shortestpathproblem”,forwhichvarioussolutionsaredescribedinliterature,eachofthemhavingitsassetsanddrawbacks(e.g.Cormenetal.2001).Inthepresentcase,aweightedgraphG(V,E)isconstructed,witheachrastercellrepresentingonevertexVandeachconnectionlinebetweenthecellsformingoneedgeEofthegraph.TheweightassociatedtoeveryedgeiscalculatedbyusingthePythagoreanTheoremtoapproximatethe3D-distancebetweencentrepointsofadjacentrastercells.Afterthesesteps,asuitablealgorithmneedstobeappliedtothegraphinordertodeterminetheshortestpathbetweenabordercellofthefocalpatchandtheclosestbordercelloftheclosestpatchofthesameclass.Inthepresentcase,aformoftheDijkstra-algorithmischosen,asitisexpectedtoprovidegoodestimatesfortheshortestpath(Chen2003).Thismethodisbasedonanundirectedcircularsearchprocedure.Consideringtheproblem,thevastcomputationeffortbecomesevident:fora1000x1000DEM,theconstructedgraphconsistsof1*106vertices,aggregatedtoformanumberof“nodes”(definedbythepatchespresent),andapprox.4*106edges.Thisimpliesthatatrade-offbetweencomputationtimeandcalculationresultshastobemade,withtheDijkstra-methodprovidinganacceptablecompromisebetweenthesetwofactors.Onthebasisofthesetruesurfacegeometries,anumberofbasiclandscapemetricscanbecalculatedandbecomparedtotheirplanimetric2D-equivalents.ThesemetricsarelistedinTable1.Table1:Selectedstandardmetrics,calculatedusingboth2D-and3D-geometries.2.2CharacterisingsurfaceroughnessAsoutlinedinthefirstchapter,surfaceroughnessmaybeacriticalissueinassessinganumberofecologicalfunctions,notablyclimaticconditionsorerosionprocesses.Therefore,simpleandstraightforwardmeasurestocaptureroughnesscharacteristicsareneededtohelpimprovetheaccuracyoflandscapeanalyses.Themostself-evidentapproachinthiscontextmaybetosimplycalculatetheratiooftruesurfacearea(asdescribedintheprevioussection)andplanimetricarea.Thismayprovideafirstestimateoftheoveralldeviationofthepatchsurfacefromaperfect2D-plane.Completelyplanepatchesconsequentlyresultinanarearatio-valueof1.Otherconceptsforthecharacterisationofsurfacefeaturessuchasroughnessareprovidedbysurfacemetrology(Stoutetal.1993).Thisscientificfielddealswiththecharacterisationofmanufacturedsurfaces(forexampleopticlenses)onamicroscopicscale.Whenthesemeasuresaretransferredtoalargerscale,theymaybeapplicabletoecologicalproblemsandanalysesoflandscapestructureaswell.Theindex“AverageSurfaceRoughness”(Ra)appearstobethemost-frequentlyusedparameterfromthissetandatthesametimetheonewiththeleastcomputationeffort.Raisusuallycalculatedasthemeanabsolutedepartureofapatch’selevationvaluesfromthemeanplane.Unlikethe3D/2Darearatio,thisindexisnotdimensionlessbutmaintainstheunitsoftheDEM.Therefore,itcanbeconsideredasanabsolutemeasureofsurfaceroughness.AmodificationofRaisRq,the“Root-Mean-SquareDeviationoftheSurface”,whichisastandardisedversionoftheformer.Theseandothermeasuresforthecharacterisationofthelandsurfaceusingsurfacemetrology-indicesaregiveninTable2,eventhoughnotallofthemareexplicitlycoveredindetailinthisstudy.RaandRqwerechosenforthisstudysincetheyarewidely-usedindisciplineslikematerialsscienceandarerathereasilyinterpretable.TheimplementationofthemethodsdescribedwascarriedoutusingboththeMATLABpackage(Math-Works2005)andanArcGIS-extensionprogrammedinC#usingthe.NET-environmentandtheArcObjectsclasslibraries(ESRI2005).2.3CasestudyusingneutrallandscapemodelsAtthispoint,acoupleofquestionsmayarise:Isitactuallynecessarytoincludeelevationandtopographyrespectivelyintoanalysesoflandscapestructure?Isthereanysignificantdifferenceatallbetweenthe2D-and3D-formsoflandscapemetrics?Dosimpleroughnessparameterstellusanythingaboutthereliefvariability?And,whichmayevenbethemostimportantone:istheadditionalcomputationeffortworthbeingcarriedout?Inordertohelpansweringthesequestionsandtomakevalidstatementsabouttherelevance,sensitivityandexplanatorypoweroftheproposedmethods,theabovementionedindiceswereappliedtoasetofneutrallandscapemodelsandsimulatedDEMs.Neutrallandscapemodelshaveprovedtobeavaluablemeansfortherepresentationofrealisticconditionsorforthereflectionofextremestatesoflandscapesystems(Gardner&Urban2007;Gardneretal.1987;Lietal.2004).Thisturnedouttobeuseful,asneutrallandscapesallowtomirrorlandscapesectionsofdifferentstructuring,whereassimulatedelevationmodelsmayreflectvariableheterogeneityoftheunderlyingterrain.Inthegivencase,thesoftwareSimmap(Saura&Martínez-Millán2000)wasusedtocreatelandscapeswithanextentof1000x1000rastercells(withanassignedhorizontalresolutionof1x1m)andthreelanduseclassesofequalsurfacepercentages.Theinitialprobabilitypwasalteredtoproducetwodifferenttypesoflandscapestructuring.Similarly,theprogrammeLandserf(Wood2005)wasappliedtoproduceelevationmodelsofvariousreliefvariabilities.Moreprecisely,theparameter“fractaldimension”(FD)wasalteredtoyieldthreeDEMsofincreasing“roughness”.DetailsaboutthetestlandscapesandtheDEMscanbederivedfromFigure3.Table2:Examplesofsomesimpleindicestoderiveinformationaboutsurfacecharacteristicsandtheircalculationformulae(PrecisionDevices1998).Figure3:CombinationsofneutrallandscapemodelsandsimulatedDEMs.3.Results–TheeffectoftopographyonselectedlandscapemetricsThesixlandmosaic/elevationmodel-combinationsweresubjectto3D-landscapeanalysisaccordingtotheoutlinedtechniques.Thearithmeticmeanoftheindexvalueswascalculatedforallthepatchespresentinthelandscapesinordertoillustratetheeffectoftheunderlyingreliefintheexaminedsituations(planimetricconditionsaswellasDEMswithfractaldimension2.1,2.5and2.9respectively).TheresultsaredisplayedinthediagramsinFigure4.Somefundamentalfindingscanbenoted.Forthemeanpatchareaandmeanpatchperimeter,thereisacleartrendtowardshighervaluesforincreasingreliefvariability.Thisholdstrueforboththehighlyfragmentedlandscape(p=0.54)aswellasforthemosaicdominatedbyfewerandlargerpatches(p=0.58).ThedifferencesbetweentheplanimetriccaseandeachofthethreesimulatedDEMsprovetobesignificantwhencomparedusingat-testforpairedsamples.Asexpected,acleardependenceofthevaluesontheterrainvariabilityandtheabilityoftheappliedmethodstocapturethiseffectcanbedemonstrated.Forthedistancemeasure“NearestNeighbour”,asimilareffectisevident.Thereisaclearincreaseforthemeandistancebetweennearestneighboursofthesameclasswhenthereliefisbecoming“rougher”andmorevariable.Forthegroupoftheshapemetrics,thefindingsarenotasclearandnotaseasilyinterpretable.FortheFractalDimension(FRAC),thedifferencesbetweenthe2D-versionandits3D-equivalentappliedtothethreeDEMsareratherlowandalmostneglectable,whilestillslightlyincreasingwithterrainroughness.Perimeter-AreaRatio(PARA)showsitstypicalsize-dependency(McGarigaletal.2002),andthereforethisindexhastobecarefullyinterpretedduetothegrowingmeanpatchsizewithincreasingterrainroughness.Thus,adefinitestatementabouttheeffectsofterrainontheoutputofthisparametercanhardlybemade.Allinall,thisparameterappearstobelargelyindependentofterrainroughness.AstheShapeIndex(SHAPE)correctsforthesizeproblemofthePerimeter-AreaRatioindex,itmaybethemostinterestingonetohaveacloserlookatwithinthegroupoftheshapemetrics.ThedifferencesofthemeanShapeIndexforthefourreliefsituationsexaminedseemtoberathersmallforthe2D-caseandthefirsttwoelevationmodelswithanabruptriseforthemostvariableelevationmodel.Thisagainisthecaseforbothofthelandscapemosaicsconsidered.Ofcourse,thisriseisuptoacertainextentproportionaltotheincreaseinmeanpatcharea(seeabove),asSHAPEtendstoincreasewithgrowingarea,evenifperimeterincreasesforthesamefactoratthesametime.ThiscanbederivedfromthecalculationformulaforSHAPE.Finally,thetwosimpleroughnessparameterscalculated,AverageSurfaceRoughness(Ra)andRoot-Mean-SquareDeviationoftheSurface(Rq),wereappliedtohetestlandscapes.Theresultsindicateacleardependencyoftheparameteroutputsontheunderlyingreliefwithaverysimilarbehaviourofthetwoindicesandsimilaroutputsforthetwodifferentlandscapemosaics.4.Discussion–OntherelevanceoftheproposedmethodsThisshortmethodologicalexaminationissupposedtoclarifytheeffectoftopographyandsurfaceroughnessonafewcommonlandscapemetrics.Moreover,somebothsimpleandfundamentalapproachestoconsidertheseeffectsarepresented.Patchareaandperimeterexhibitastrongconnectednesstothevariabilityoftheunderlyingterrain.Theeffectsmaynotbeasdistinctunderreal-worldconditions.ButthesimulatedlandscapemodelsandDEMsclearlydemonstratethataconsiderationoflandscapemosaicsaspurelyplanimetricsurfacesandtheircharacterisationusing2D-landscapemetricsmaynotbesufficientineverycase,especiallywhenterrainishighlyvariable.Whenonetriestocharacteriselandscapesinthesecases,theapplicationofcorrectedmetricsasproposedinthispapermaybeadvisable.Thesameholdstruefordistancemeasures.Thesemetricsmayhaveacriticalrelevancee.g.inspecies-centredhabitatanalyses.Ascanbeseenfromtheresultspresentedhere,theeffectofthereliefonthe“true”surfacedistancesbetweenpatchesshouldnotbeneglectedinroughterrain.Thiseffectmaybeexaggeratedbytheapplicationofthesimulatedlandscapes,evenifthepurposeistorevealthefundamentalrelationshipbetweendistanceandtopographyandtoprovideatechniquetoimprovethecalculationofsuchdistancemeasures.StatementsregardingshapeindiceslikePARA,FRACorSHAPEarenotasconcise.Thesemetricsdoreacttotheterrain,butabsolutedifferencesbetweentheexaminedreliefsituationstendtobelowandthetrendisnotasobviousforallofthesemeasuresasisthecaseforarea,perimeteranddistancemeasures.Onereasonforthismaybethesimplefactthatinthecalculationalgorithmsforthesemetrics,parametersofthepatchgeometriesappearbothaboveandbelowthefractionline.Therefore,whendividingforexample3D-perimeterby3D-area(bothhavinglargervaluescomparedtotheir2D-equivalents),thedifferencesbetweenthe2Dandthe3D-approachmaysimplylevelouttoacertainextent.Finally,theresultsrevealthattheanalysisofsurfaceroughnessmayserveasavaluableinstrumenttoprovidehighlycondensedinformationaboutthetopographiccharacteristicsofpatches.AsbothRaandRqarecloselyconnectedtotheinitialroughnessparameteroftherespectiveDEMs(i.e.theirfractaldimensionFD),theycanberegardedasagoodextensionofcommonlandscapemetricstowardsthethirdspatialdimension,especiallyastheircalculationalgorithmsareratherstraight-forwardandcanbeeasilyintegratedintothepatch-corridor-matrixmodeloflandscapes.Moreover,theresultsfromtheseparametersareeasilyinterpretable.Toassesstheinfluenceoflandscapeconfigurationandpatchstructuringonthemetricsoutput,thetwochosenmosaicswithaninitialprobabilitypweresupposedtoserveasarepresentationofdifferentstructuralconditions.Itturnsoutthatthegeneraltrendsinindexbehaviourforincreasingfractaldimensionoftheunderlyingreliefaregenerallythesame.Thedistancetothenearestneighbourinthesameclasstendstobelargerforp=0.58,becauseonaveragelargerpatchesofotherclasseshavetobecrossed.Thisindicatesthattheapplicationofthecorrectionalgorithmfordistancecalculationsmaybeparticularlyvaluableincoarse-grainedlandscapeswithlargereliefvariability.Asidefromthesefindings,preliminarystudiescarriedoutapplyingtheproposedmethodstoreal-worlddatahaveshownthatlargepatchesingeneralleadtosomemitigationofterraineffectsonlandscapemetrics,asoftenlandscapeelementscompriseboth“flat”and“rough”areas.Thiscircumstanceisnotreflectedtothesamedegreebytherelativelyhomogeneoussimulatedelevationmodelsusedinthisstudy,wherethequantificationofterrainroughnessratherthangenerallandscapeconfigurationwasthemainfocus.Theresultssuggestthattheproposedmethodsmayexhibitalargepotentialformanyecologicalproblems.Sinceespeciallymeasuresforhabitatareaandhabitatisolationorfragmentationarekeyvariablesinmanyspecies-centredanalyses,theusageofcorrectionalgorithmsforthesegeometriesappearstoofferthepossibilityofimprovedresults(forexamplesdealingwiththesemeasuresseeBennett2003;Fahrig1997;Kraussetal.2005;vanDorp&Opdam1987).Figure4:Diagramsdisplayingthearithmeticmeansand95%-confidenceintervalsaroundthemeanvaluefortheselectedindices;eachindexwascalculatedforthetwotestlandscapes(displayedinblueandgreenrespectively)combinedwitheachofthethreeelevationmodelsaswellastheplanimetriccase.5.ConclusionsandoutlookThefindingspresentedinthispaperindicatethatthepatch-corridor-matrixmodelastheprevailingconcepttoperceiveanddescribelandscapesmaynotsufficeincaseswheretopographicandmorphologicfeaturesofthelandsurfaceneedtobetakenintoaccount.Astopographyplaysacrucialroleinmanyecologicalprocesses,simplemethodsandtechniquesforitsassessmentareneeded.Weproposesomestraightforwardapproachesthatenablelandscapeecologiststoaccountfortheeffectsofreliefandlandformintheiranalyses.Thesuggestedframeworkfortheadjustmentofstandardlandscapemetrics,thusconvertingthemto3D-metrics,maybeappliedt