基于多代理系统的湿地时空分布多尺度模拟研究(IJISA-V4-N9-4)

I.J.IntelligentSystemsandApplications,2012,9,29-38PublishedOnlineAugust2012inMECS()DOI:10.5815/ijisa.2012.09.04Copyright©2012MECSI.J.IntelligentSystemsandApplications,2012,9,29-38WetlandsSpatial-TemporalDistributionMulti-ScaleSimulationUsingMulti-AgentSystemHuanYuCollegeofEarthSciences,ChengduUniversityofTechnology,Chengdu,[e-mail]tainHazardsandEnvironment,ChineseAcademyofSciences,Chengdu,[e-mail]NortheastInstituteofGeographyandAgroecology,ChineseAcademyofSciences,Changchun,[e-mail];[e-mail]—Thesimulationofwetlandlandscapespatial-temporaldistributionnotonlycanrevealthemechanismsandlawsoflandscapeevolution,butachievethesustainablelanduseaswellasprovidesupportsforwetlandconservationandmanagement.Inthisreport,theinlandfreshwaterwetlandsintheSanjiangPlainofChinawereselectedforwetlandlandscapechangingprocesssimulationstudies.Resultsshowedthatbothvisualeffectsofsimulationandpredictionweregoodandthetotalaccuracyco-efficiencyofpointstopointswasalsosignificantlyhigh(above82%),whichdemonstratedthefeasibilityandeffectivenessofwetlandlandscapespatial-temporaldistributionsimulationusingMulti-AgentSystem(MAS).Scalesexertedinfluenceonvisualeffects,simulationaccuraciesandstatisticsoflandscapeindex.ScaleeffectswereobviousduringsimulationprocessusingMAS.Itwasdemonstratedthat60mwasthebestscaleforsimulation.Itwasshownthatcontagionindexlineswereexponentialdistributionwhileaccuracylineswerelognormaldistributionwiththescalerising,whichprovidedareferenceforscaleeffectassessmentandsimulationscaleselection.IndexTerms—Multi-AgentSystem,Multi-Scale,Wetlands,Spatial-TemporalDistribution,SimulationI.IntroductionTransformationoftheEarthsurfacereceivesmuchattentionbygeophysicistsworldwide,asaprocessitself;landcoverchangehashugeimpactsonecosystems,biodiversity,andtheglobalclimate[1].Wetlandsareintegralpartsoftheglobalecosystemastheycanpreventorreducetheseverityoffloods,feedgroundwater,andprovideuniquehabitatsforfloraandfauna[2].However,underdualpressuresofnaturalfactors(suchasclimatechange)andthegrowingintensityofhumanactivities/disturbance,wetlandareashrinksdrasticallyinrecentdecades,ecologicalfunctionsdegradeobviously,whichleadtoaseriesofecologicalandenvironmentalproblems[3].Closeattentionshouldbepaidtotheacceleratingreduction,lossorchangeofwetlandcoveragebeforeitgetsevenworse.Landscapedigitalreconstructionandspatial-temporaldistributionsimulationbasedonmulti-periodregionallandcoveragecanhelptounderstandthemechanismsandlawsoflandscapesuccession,torecognizetherelationshipbetweenhumanactivitiesandlandscapechanges,thenpredictthefuturetrendofthelandscape,andultimatelyprovidestrategiestodecision-makerforwetlandconservationandmanagement.Asatoolforanalyzingcomplexnaturalsystems,multi-agentmodelisespeciallysuitableforsimulationofvariouspoliciesandimpactonspecificenvironmentofindividualdecision.Conclusionresultedfromsuchsimulationstudiescanprovidedecisionsupportformanagementanddevelopmentofnaturalresourcesandsustainableutilization[4].Currently,themulti-agenttechnologyhasbeenwidelyusedinsimulationofurban,agriculturallandusechange,andnaturalresourcemanagement[5-11].Thesuitabilityforwetlandlandscapechangesimulationshouldbetestedaswell.TheinlandfreshwaterwetlandsintheSanjiangPlain,NortheastofChinawasselectedforthissimulationstudy,inordertoverifythefeasibilityandeffectivenessofwetlandlandscapespatial-temporalevolutionsimulationusingMASandprovideanewtechnicalideaandmethodforwetlandlandscapechangesimulationandprotection,experimentofwetlandlandscapeshrinkingprocesssimulationwascarriedoutbasedonmulti-temporallanduseandotherancillarygeographicdata.Inaddition,thescaleproblemsarewidespreadinnature.Manyphenomenaandprocessesinvolvemulti-scalefeaturesorscaleeffects.Atthesametime,observations(measurements)ofthephenomenonor30WetlandsSpatial-TemporalDistributionMulti-ScaleSimulationUsingMulti-AgentSystemCopyright©2012MECSI.J.IntelligentSystemsandApplications,2012,9,29-38processarealsoondifferentscales(resolutions),thescalesareoftenunavoidableespeciallyinthefieldofgeosciences.Scalesinthefieldofgeoscienceshavemanyconnotations.Fromtheremotesensingpointofview,scaleisthemeasurementofspacescope,intervaltimeandspectralwidth[12].Insurveyingandcartography,thescaleisusuallybeingdescribedastheratioofdistanceinmapandtheactualdistanceexpressed.Nonetheless,fromecologypointofview,scalereferstotheecosystemsizeordynamictimeinterval,whichisalwaysexpressedassizeandmagnitude[13,14].Inthisresearch,wedefinethescaleasthesmallestareaunitandtimeintervalofvariousdatasourcesinthecourseofspatial-temporalevolutionsimulation.Inthisreport,MAStechnologyisintroducedintothesimulationofwetlandlandscapechangesprocessthroughtheresolutionchangestoachievemulti-scaledataacquisition.Thescaleeffectsofsimulationareobserved,withwhichareferenceforscalechoicewillbeconcluded.Thispaperisorganizedasfollows:InSections2,thedetailsituationofstudyareawillbeintroduced.Dataandmethodswillbepre