G.2086Page1GlobalResearchjournalofNaturalScience&Technology(GRJNST)Volume:04-Issue3(2026),2083ISSNP:2790-7643ISSNE:2790-7651www.grjnst.nethttps://doi.org/10.53762/grjnst.04.03.07IntegratingArtificialIntelligence,RemoteSensing,andGISforSustainableAgro-ForestryManagementandLandResourceOptimizationReceived:31March2026.Accepted:29April2026.Published:12May,2026Email:MuhammadKashifMajeed(CorrespondingAuthor)FacultyofEngineeringScienceandTechnology,IqraUniversity,Karachi,Pakistan.mkashif@iqra.edu.pkKashifAkbarDepartmentofIndustrialEngineering,UniversityofPadua,Italykashif.akbar@studenti.unipd.itMuhammadHassanAliForestandRangeManagementShaheedBenazirBhuttoUniversityofVeterinary&AnimalSciencesSakrand,Sindh,Pakistansoilscience1070@gmail.comEmail:Email:MuhammadEssaSiddiquePhD(IT)ScholarDr.AHSBukhariCentreofICT,FacultyofEngineering&Technology,UniversityofSindh,Jamshoro,Pakistan.Essasiddique@live.comEmail:MurtazaAliDepartmentofHorticulture,SindhAgricultureUniversity,Tandojam,Sindh,Pakistan.murtazal46@gmail.comEmail:GulMuhammadShahDepartmentofSoilScience,SindhAgricultureUniversity,Tandojam,Sindh,Pakistan.syedgm7@gmail.comEmail:GRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07Copyright©2026GRJNST.ThisarticleispublishedunderanOpenAccessmodel.ItismadeavailabletothepublicunderthetermsoftheCreativeCommonsAttribution4.0International(CCBY4.0)license,whichpermitsunrestricteduseanddistributionG.2086Page2Abstract:Theincreasingpressureonlandresourcesduetoclimatechange,populationgrowth,andunsustainableland-usepracticesnecessitatesadvancedapproachesforsustainableagro-forestrymanagement.ThisstudyproposesanintegratedframeworkthatcombinesArtificialIntelligence(AI),RemoteSensing(RS),andGeographicInformationSystems(GIS)tosupportlandsuitabilityanalysisandresourceoptimization.Multi-sourcegeospatialdata,includingsatelliteimagery,environmentalvariables,andtopographicinformation,areprocessedthroughastructuredpipelineinvolvingpre-processing,featureextraction,AI-basedmodelling,andGIS-drivenspatialanalysis.Machinelearninganddeeplearningmodelsareemployedtoperformlanduseclassificationandpredictagro-forestrysuitability,whilemulti-criteriadecisionanalysis(MCDA)isusedtointegrateenvironmentalfactorsandgeneratesuitabilitymaps.Theresultsindicatethatasignificantportionofthestudyareafallswithinhighlyandmoderatelysuitablecategories,demonstratingstrongpotentialforagro-forestrydevelopment.Modelevaluationusing5-foldcross-validationshowsstableperformanceacrossallmodels,withdeeplearningapproachesprovidingslightlyhigheraccuracy,whiletraditionalmodelsremaincomputationallyefficient.TheintegrationofNDVI-basedenvironmentalanalysiswithspatialmodellingfurthervalidatesthereliabilityoftheframework.Theresultingdecisionsupportoutputsenabletheidentificationofpriorityzonesandoptimizedland-usestrategies.Overall,theproposedframeworkoffersascalableanddata-drivensolutionforsustainablelandresourcemanagementandsupportsinformeddecision-makingforagro-forestryplanning.Keywords:ArtificialIntelligence(AI),RemoteSensing(RS),GeographicInformationSystems(GIS),Agro-Forestry,LandSuitabilityAnalysis,LandResourceOptimization,LandUseandLandCover(LULC),SpatialAnalysis,Multi-CriteriaDecisionAnalysis(MCDA),MachineLearning,DeepLearning,NDVI,EnvironmentalMonitoring,DecisionSupportSystems(DSS)1.Introduction1.1BackgroundandMotivationThegrowingpressureonlandresourceshasbecomeoneofthedefiningenvironmentalchallengesofthe21stcentury.Rapidpopulationgrowth,expandingagriculturaldemand,andincreasingurbanizationarecontinuouslyreshapingnaturallandscapes,oftenattheexpenseofecologicalstability.Atthesametime,climatechangeisintensifyingthesepressuresbyalteringrainfallpatterns,increasingtemperatures,andamplifyingthefrequencyofextremeweatherevents.Thesechangeshaveprofoundimplicationsforlandproductivity,biodiversity,andlong-termsustainability.Inmanyregions,traditionallandmanagementpracticesarenolongersufficienttocopewiththesecomplexandinterconnectedchallenges,creatingastrongneedformoreadaptive,data-drivenapproachestoresourcemanagement.Withinthisbroadercontext,agro-forestryhasgainedconsiderableattentionasasustainableland-usestrategythatintegratestreeswithcropsandlivestocksystems.Unlikeconventionalmonoculture-basedagriculture,agro-forestrysystemspromoteecologicalbalancebyenhancingsoilfertility,improvingwaterretention,reducingerosion,andincreasingcarbonsequestration.Thesesystemsalsooffersocio-GRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page3economicbenefitsbydiversifyingincomesourcesandimprovingresilienceagainstclimatevariability.However,thesuccessfulimplementationofagro-forestryrequiresadetailedunderstandingofspatialheterogeneity,environmentalconstraints,andlandsuitabilityconditions—factorsthataredifficulttoassessusingtraditionalmethodsalone.Inrecentyears,geospatialtechnologiessuchasRemoteSensing(RS)andGeographicInformationSystems(GIS)haveemergedaspowerfultoolsformonitoringandmanagingenvironmentalsystems.Remotesensingprovidescontinuous,large-scaleobservationsoftheEarth’ssurfacethroughsatelliteimagery,enablingtheassessmentofvegetationhealth,land-usedynamics,andenvironmentalchangesovertime.GIS,ontheotherhand,facilitatestheintegrationandanalysisofspatialdata,allowingresearchersandpractitionerstodevelopmodelsforland-useplanningandresourceallocation.Together,thesetechnologiesformthebackboneofmodernenvironmentalmonitoringsystemsandhavesignificantlyimprovedtheabilitytoanalyzecomplexspatialphenomena[1],[2].1.2RoleofGeospatialTechnologiesandAIIntegrationTheintegrationofArtificialIntelligence(AI)withgeospatialtechnologieshasfurtherstrengthenedthecapabilityofenvironmentalmonitoringandlandmanagementsystems.Machinelearninganddeeplearningalgorithmsenableautomatedextractionofpatternsfromlargedatasets,improvingbothefficiencyandaccuracyingeospatialanalysis.AI-drivenmodelshaveshownstrongperformanceinland-useclassification,vegetationmonitoring,andenvironmentalpredictiontasks,oftenoutperformingtraditionalstatisticalapproaches[3],[4].Tobetterunderstandthecomplementaryrolesofthesetechnologies,Table1providesacomparativeoverviewofAI,RemoteSensing,andGISinagro-forestryapplications.Table1:ComparativeRoleofAI,RemoteSensing,andGISinAgro-ForestrySystemsDataTypeStrengthsLimitationsApplicationAgro-ForestryinTechnologyRemoteSensing(RS)CoreCapabilityEarthobservation&monitoringSatelliteimagery,spectraldataGISSpatialanalysisintegration&Vectorrasterdata&spatialLarge-scalecoverage,temporalmonitoringMulti-layeranalysis,decisionmodellingCloudinterference,resolutionlimitsVegetationhealth(NDVI),landcovermappingRequiresqualityinputdataLandsuitability,zoning,planningArtificialIntelligence(AI)Patternrecognition&predictionStructured&unstructureddataHighaccuracy,automation,scalabilityDatadependency,computationalcostClassification,prediction,optimizationIntegratedIntelligentgeospatialMulti-sourceHighprecision,real-HighcomputationalDecisionSupport(DSS),SystemsGRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086AI–RS–GISanalyticsdatasetstimeinsightsrequirementsprecisionforestryPage41.3GlobalChallengesandSustainableAgro-ForestryTheurgencyofadoptingsustainableagro-forestrypracticesiscloselylinkedtoseveralglobalenvironmentalchallenges.Climatechangeremainsacentralconcern,asitdirectlyaffectsagriculturalproductivityandecosystemstability.Variationsintemperatureandprecipitationpatternscandisruptcropcycles,reduceyields,andincreasevulnerabilitytopestsanddiseases.Atthesametime,deforestationcontinuestocontributesignificantlytogreenhousegasemissionswhilereducingbiodiversityandecosystemservices[5].Landdegradationisanothercriticalissue,affectingalargeportionoftheworld’slandarea.Unsustainableagriculturalpractices,deforestation,andoverexploitationofnaturalresourceshaveledtosoilerosion,nutrientdepletion,anddesertification.Theseprocessesnotonlyreducelandproductivitybutalsothreatenfoodsecurity.Agro-forestrysystems,throughtheirintegratedapproach,offeraviablesolutionbyimprovingsoilquality,enhancingcarbonsequestration,andrestoringecologicalbalance[6].Foodsecurityisintrinsicallytiedtotheseenvironmentalissues.Withincreasingpopulationdemands,thereisagrowingneedtooptimizelandusewithoutfurtherdegradingnaturalecosystems.Thisrequiresadvancedtoolsthatcananalyzeenvironmentalconditionsandsupportinformeddecision-making.1.4KeyFactorsInfluencingAgro-ForestrySuitabilityEffectiveagro-forestryplanningdependsonawiderangeofenvironmental,spatial,andsocio-economicfactors.Thesefactorsmustbecarefullyanalyzedtodeterminethesuitabilityoflandforsustainablepractices.Theintegrationofmulti-sourcedatasetsallowsforamorecomprehensiveevaluationofthesevariables.Table2summarizesthekeyfactorsinfluencingagro-forestrysuitabilityandtheirrelevanceinlandresourceoptimization.Table2:KeyEnvironmentalandGeospatialFactorsInfluencingAgro-ForestrySuitabilityFactorCategoryClimaticSoilParametersDataSourceImportanceForestryinAgro-Rainfall,Temperature,HumidityMeteorologicalsatellitedata,Determinesgrowthpatternscrop/treeSoiltype,pH,organicmatterSoildatasetssurveys,FAOAffectsproductivityfertilityandTopographicElevation,slope,aspectDEM(DigitalElevationModel)InfluenceswaterflowanderosionVegetationNDVI,biomassdensityRemotesensingimageryIndicatesvegetationhealthGRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086HydrologicalWaterdrainageavailability,GIShydrologicalmodelsSocio-economicAccessibility,ownershiplandSurvey/GISlayersPage5forirrigationfeasibilityandCriticalplanningImpactsadoption1.5EmergenceofAIinGeospatialAnalyticsTheapplicationofAIingeospatialanalyticshassignificantlyimprovedtheabilitytoprocessandinterpretcomplexenvironmentaldatasets.MachinelearningtechniquessuchasRandomForestandSupportVectorMachinesarewidelyusedforclassificationandpredictiontasks,whiledeeplearningmodelslikeConvolutionalNeuralNetworks(CNN’s)provideenhancedperformanceinimage-basedanalysis[7],[8].Thesemodelsenabletheextractionofhigh-levelfeaturesfromsatelliteimagery,facilitatingmoreaccurateland-useclassificationandenvironmentalmonitoring.Furthermore,AIenablespredictivemodelling,allowingresearcherstoforecastchangesinlanduse,vegetationpatterns,andenvironmentalconditions.Thiscapabilityisparticularlyvaluableforproactiveplanningandsustainableresourcemanagement.However,theeffectivenessofthesemodelsdependsontheavailabilityofhigh-qualitydataandappropriatefeatureselection[9],[10].1.6ProblemStatementDespitetheavailabilityofadvancedgeospatialandAItechnologies,manylandmanagementpracticesstillrelyonfragmentedandtraditionalapproaches.Thesemethodsoftenfailtocapturethespatialandtemporalcomplexityofenvironmentalsystems,leadingtoinefficientresourceutilizationandsuboptimaldecision-making.Inparticular,traditionalapproacheslacktheabilitytointegratediversedatasets,analyzelargevolumesofinformation,andprovidereal-timeinsights[11].Moreover,theabsenceofintegratedframeworksthatcombineAI,RS,andGISlimitstheeffectivenessofcurrentdecision-supportsystems.Whileindividualtechnologieshavedemonstratedsignificantpotential,theirisolatedapplicationoftenresultsinincompleteorinconsistentanalyses.Thishighlightstheneedforaunifiedapproachthatleveragesthestrengthsofthesetechnologiestoprovidecomprehensiveandaccurateinsightsforsustainablelandmanagement[12].1.7ResearchObjectivesandContributionsInlightofthesechallenges,thisstudyaimstodevelopanintegratedframeworkthatcombinesAI,remotesensing,andGISforsustainableagro-forestrymanagementandlandresourceoptimization.Theprimaryobjectiveistoutilizemulti-sourcegeospatialdataandadvancedanalyticaltechniquestosupportinformeddecision-making.Specifically,thestudyfocusesondevelopingmodelsforlandsuitabilityassessment,optimizingresourceallocation,andenhancingenvironmentalmonitoringcapabilitiesthroughAI-drivenanalysis.ThisresearchcontributestothefieldbyproposingaunifiedAI–RS–GISframeworkthataddressesthelimitationsofexistingapproaches.Thestudyemphasizesmulti-sourcedataintegration,combiningGRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086satelliteimagery,environmentalvariables,andspatialdatasetstoimproveanalyticalaccuracy.ItalsointroducesAI-drivenpredictivemodellingtechniquesforlandsuitabilityassessment,enablingmorepreciseandreliabledecision-making.Additionally,thedevelopmentofadecisionsupportsystem(DSS)providespracticaltoolsforpolicymakersandpractitionerstoimplementsustainableagro-forestrystrategies.ThesecontributionscollectivelyadvancetheintegrationofgeospatialtechnologiesandAIinenvironmentalmanagement[1],[3].Page61.8OrganizationofthePaperTheremainderofthispaperisorganizedasfollows.Section2presentsadetailedreviewoftheexistingliteratureonagro-forestry,geospatialtechnologies,andAIapplications.Section3describestheproposedmethodology,includingdatacollection,pre-processing,andmodeldevelopment.Section4discussestheresultsandfindings,followedbyacomprehensivediscussioninSection5.Finally,Section6concludesthepaperandoutlinesdirectionsforfutureresearch.Toprovideaclearoverviewoftheproposedintegratedsystem,theconceptualframeworkillustratingtheinteractionbetweenArtificialIntelligence,RemoteSensing,andGIScomponentsispresentedinFig.1.Fig.1.ConceptualframeworkofintegratingArtificialIntelligence(AI),RemoteSensing(RS),andGeographicInformationSystems(GIS)forsustainableagro-forestrymanagementandlandresourceoptimizationFig.1presentstheoverallarchitectureoftheproposedAI–RS–GISintegratedframeworkforsustainableagro-forestrymanagement.Theprocessbeginswithmulti-sourcedataacquisition,includingGRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page7satelliteimagery,environmentalvariables,andsocio-economicdata.Thisisfollowedbydatapreprocessingstepssuchascorrection,featureextraction,andintegration.Inthenextstage,AI-basedmodelsareappliedforlanduseclassification,vegetationmonitoring,andpredictiveanalysis.TheoutputsarethenincorporatedintoGIS-basedspatialanalysistechniques,includingmulti-criteriadecisionanalysisandsuitabilitymapping.Finally,adecisionsupportsystem(DSS)generatesactionableinsights,suchasoptimallandallocationandsustainablemanagementrecommendations.Theframeworkalsoincludesafeedbackloopforcontinuousmonitoringandimprovement,ensuringadaptiveanddata-drivendecision-making.2.LiteratureReview2.1Agro-ForestryandSustainableLandManagementAgro-forestryhasevolvedasamultidisciplinaryapproachthatintegratesagriculturalandforestrypracticestoenhancebothecologicalsustainabilityandeconomicproductivity.Atitscore,agro-forestryemphasizesthedeliberatecombinationoftreeswithcropsandlivestockwithinthesameland-usesystem,creatingsynergisticinteractionsthatimproveoverallsystemperformance.Overtheyears,researchershaveincreasinglyhighlightedtheimportanceofagro-forestryinaddressingenvironmentalchallengessuchassoildegradation,biodiversityloss,andclimatechange.Studiesindicatethatagro-forestrysystemscontributesignificantlytocarbonsequestration,improvesoilfertilitythroughorganicmatteraccumulation,andenhancewaterretentioncapacity,therebyreducingvulnerabilitytoclimatevariability[13],[14].Beyondenvironmentalbenefits,agro-forestryalsoplaysacriticalroleinimprovingrurallivelihoods.Bydiversifyingproductionsystems,farmerscanreduceeconomicrisksassociatedwithsingle-cropdependencywhileincreasingincomestability.Socio-economicstudieshaveshownthatagro-forestrycontributestofoodsecuritybyprovidingmultipleoutputssuchasfruits,timber,fodder,andfuelwood[15].Moreover,thesesystemssupportsustainablelandmanagementbypromotingefficientresourceutilizationandreducingtheneedforchemicalinputs.However,despitetheseadvantages,theadoptionofagro-forestrypracticesremainsunevenduetochallengessuchaslackofawareness,insufficienttechnicalsupport,andlimitedaccesstodecision-makingtools.2.2RemoteSensinginEnvironmentalMonitoringRemotesensinghasbecomeanindispensabletoolinenvironmentalmonitoring,offeringtheabilitytoobserveandanalyzelarge-scalespatialpatternsovertime.SatelliteimageryfromplatformssuchasLandsat,Sentinel,andMODISprovideshigh-resolutiondatathatcanbeusedtoassessland-usechanges,vegetationhealth,andenvironmentaldegradation.Oneofthemostwidelyusedtechniquesinremotesensingisthecalculationofvegetationindices,particularlytheNormalizedDifferenceVegetationIndex(NDVI)andEnhancedVegetationIndex(EVI).Theseindicesprovidequantitativemeasuresofvegetationdensityandhealth,enablingresearcherstomonitorcropconditions,detectstressfactors,andevaluateecosystemproductivity[16],[17].Inforestryapplications,remotesensinghasbeenextensivelyusedforlandcoverclassification,biomassestimation,anddeforestationmonitoring.Machinelearningalgorithmsappliedtosatelliteimageryhavesignificantlyimprovedclassificationaccuracy,allowingformoreprecisemappingofland-usecategories.GRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page8Furthermore,time-seriesanalysisofremotesensingdataenablesthedetectionoftemporalchanges,whichiscrucialforunderstandingenvironmentaldynamicsandassessingtheimpactofhumanactivities[18].Despiteitsadvantages,remotesensingfaceslimitationssuchascloudinterferenceandvariabilityinspatialresolution,whichcanaffectdataqualityandinterpretation.2.3GISforSpatialAnalysisandLandUsePlanningGeographicInformationSystems(GIS)playacentralroleinspatialanalysisandland-useplanningbyprovidingtoolsfordataintegration,visualization,andmodelling.GISenablesthecombinationofmultiplespatialdatasets,includingtopography,soilcharacteristics,climatevariables,andland-useinformation,tosupportcomprehensiveenvironmentalassessments.OneofthekeystrengthsofGISliesinitsabilitytoperformspatialmodelling,whichallowsresearcherstoanalyzerelationshipsbetweendifferentenvironmentalfactorsandpredictfuturescenarios.AwidelyusedapproachinGIS-baseddecision-makingisMulti-CriteriaDecisionAnalysis(MCDA),whichintegratesmultiplefactorstoevaluatelandsuitabilityforspecificapplications.MCDAtechniques,suchasweightedoverlayanalysis,allowfortheprioritizationofcriteriabasedontheirrelativeimportance,enablingmoreinformedandtransparentdecision-makingprocesses[19],[20].GIS-basedenvironmentalassessmenthasbeensuccessfullyappliedinvariousdomains,includingagriculture,forestry,andurbanplanning.However,theeffectivenessofGISdependsheavilyonthequalityandaccuracyofinputdata,aswellastheselectionofappropriatemodellingtechniques.2.4ArtificialIntelligenceinGeospatialApplicationsTheapplicationofArtificialIntelligence(AI)ingeospatialanalysishasgainedsignificantmomentuminrecentyears,drivenbytheincreasingavailabilityoflargedatasetsandadvancesincomputationalpower.MachinelearningalgorithmssuchasRandomForest(RF),SupportVectorMachines(SVM),andArtificialNeuralNetworks(ANN)havebeenwidelyusedforclassificationandpredictiontasksinenvironmentalstudies.Thesealgorithmsareparticularlyeffectiveinhandlinghigh-dimensionaldataandcapturingcomplex,non-linearrelationshipsbetweenvariables[21],[22].Deeplearningtechniques,especiallyConvolutionalNeuralNetworks(CNN’s),havefurtherenhancedthecapabilitiesofgeospatialanalysisbyenablingautomatedfeatureextractionfromsatelliteimagery.CNN-basedmodelshavedemonstratedsuperiorperformanceinimageclassification,objectdetection,andchangedetectiontasks,makingthemhighlysuitableforapplicationsinland-usemappingandenvironmentalmonitoring[23].Inadditiontoclassificationtasks,AIisincreasinglybeingusedforpredictiveanalytics,allowingresearcherstoforecastchangesinlanduse,vegetationpatterns,andenvironmentalconditions.Thesepredictivecapabilitiesareessentialforproactivedecision-makingandsustainablelandmanagement[24].TosummarizethekeyAItechniquesandtheirapplicationsingeospatialstudies,Table3providesacomparativeoverview.Table3:AITechniquesinGeospatialandAgro-ForestryApplicationsAITechniqueTypeApplicationAreaStrengthsLimitationsGRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page9RandomForest(RF)MachineLearningLandclassificationHighrobusttonoiseaccuracy,LessinterpretableSupportMachine(SVM)VectorMachineLearningArtificialNetwork(ANN)NeuralNeuralNetworkSuitabilitymappingPredictionmodellingEffectivesmalldatasetswithComputationalcomplexityCapturesnon-linearrelationshipsRiskofoverfittingConvolutionalNeuralNetwork(CNN)DeepLearningImageclassificationHighaccuracyforimagerydataRequiresdatasetslargeGradientBoostingEnsembleLearningEnvironmentalpredictionStrongpredictiveperformanceSensitivetonoise2.5IntegratedAI–RS–GISApproachesTheintegrationofAI,RemoteSensing,andGIShasemergedasapromisingapproachforaddressingcomplexenvironmentalchallenges.Severalstudieshaveproposedframeworksthatcombinethesetechnologiestoimproveland-useclassification,environmentalmonitoring,anddecision-makingprocesses.Forinstance,integratedmodelshavebeenusedtoanalyzemulti-sourcedata,enablingmoreaccurateandcomprehensiveassessmentsoflandsuitabilityandresourcedistribution[25],[26].Comparativeanalysesofpriorstudiesrevealthatintegratedapproachesconsistentlyoutperformsingle-technologymethodsintermsofaccuracyandefficiency.ThesynergybetweenAI,RS,andGISallowsfortheautomationofdataprocessing,theextractionofmeaningfulpatterns,andthegenerationofactionableinsights.However,existingframeworksoftenlackscalabilityandreal-timecapabilities,limitingtheirapplicabilityindynamicenvironments.Tohighlightthedifferencesbetweentraditionalandintegratedapproaches,Table4presentsacomparativeanalysis.Table4:ComparisonofTraditionalvsIntegratedAI–RS–GISApproachesApproachTraditionalMethodsDataIntegrationAccuracyScalabilityReal-TimeCapabilityLowModerateLimitedNoDecisionSupportBasicPartialModerateModerateModerateLimitedModerateLimitedHighLimitedHighEmergingAdvancedRS-basedMethodsModerateGIS-basedMethodsHighAI-basedMethodsModerateIntegratedAI–RS–GISVeryHighGoodGoodHighVeryHighGRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.20862.6ResearchGapsPage10DespitethesignificantprogressingeospatialtechnologiesandAI,severalresearchgapsremain.OneoftheprimarychallengesisthelackofunifiedframeworksthatseamlesslyintegrateAI,RS,andGISintoacohesivesystem.Manyexistingstudiesfocusonindividualcomponentsratherthanexploringtheircombinedpotential,resultinginfragmentedsolutions.Additionally,thereisalimitedavailabilityofreal-timedecisionsupportsystemsthatcanprovidetimelyandactionableinsightsforlandmanagement.Anotherimportantgapistheneedforscalableandadaptivemodelsthatcanhandlelargedatasetsanddynamicenvironmentalconditions.Currentapproachesoftenstrugglewithcomputationalcomplexityanddataheterogeneity,whichcanlimittheireffectivenessinreal-worldapplications.AddressingthesechallengesrequiresthedevelopmentofintegratedframeworksthatleveragethestrengthsofAI,RS,andGISwhileensuringscalability,adaptability,andreal-timeperformance.Tosystematicallyorganizethereviewedliteratureandhighlightexistingresearchgaps,thetaxonomyofAI,RemoteSensing,andGISapplicationsisillustratedinFig.2.Fig.2.LiteraturetaxonomyandresearchgapanalysisofAI–RS–GISapproachesforsustainableagro-forestrymanagement.Fig.2presentsastructuredtaxonomyoftheliteratureonagro-forestryandlandresourceoptimizationbycategorizingexistingstudiesintofivemajordomains:agro-forestryandsustainablelandmanagement,remotesensing,GIS-basedspatialanalysis,artificialintelligencetechniques,andintegratedGRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page11AI–RS–GISframeworks.Eachdomainoutlineskeyconcepts,methodologies,andapplicationssuchasvegetationmonitoring,spatialmodelling,machinelearning,andpredictiveanalytics.Thefigurealsohighlightscriticalresearchgaps,includingthelackofunifiedframeworks,limitedreal-timedecision-supportsystems,dataheterogeneity,andscalabilitychallenges.Basedonthesegaps,theneedforanintegrated,intelligent,andscalableAI–RS–GISframeworkisemphasized.Additionally,futureresearchdirectionsarepresented,focusingonreal-timeprocessing,advancedAImodels,andimproveddecision-supportcapabilitiesforsustainableagro-forestrymanagement.3.Methodology3.1ProposedFrameworkOverviewThisstudyadoptsanintegratedmethodologicalframeworkthatcombinesArtificialIntelligence(AI),RemoteSensing(RS),andGeographicInformationSystems(GIS)tosupportsustainableagro-forestrymanagementandlandresourceoptimization.BuildingontheresearchgapsidentifiedinSection2andtheconceptualarchitectureillustratedinFigure1,theproposedmethodologyisdesignedasamulti-stagepipelinethatsystematicallytransformsrawmulti-sourcedataintoactionabledecision-supportoutputs.Theframeworkbeginswithdataacquisitionfromheterogeneoussources,followedbypre-processingandfeatureextraction,AI-basedmodelling,GIS-drivenspatialanalysis,andfinallythegenerationofdecision-supportinsights.Thisstructuredworkflowensuresthatbothspatialandnon-spatialdimensionsofagro-forestrysystemsareeffectivelycapturedandanalyzed.TheintegrationofAIwithgeospatialtechnologiesenablesthehandlingofhigh-dimensionaldatasetsandsupportsthedevelopmentofpredictivemodelsforlandsuitability,biomassestimation,andenvironmentalmonitoring.Furthermore,theincorporationofGIS-basedmulti-criteriadecisionanalysis(MCDA)ensuresthatmultipleenvironmentalandsocio-economicfactorsareconsideredsimultaneously,leadingtomoreinformedandreliabledecision-making.Thisholisticapproachaddressesthelimitationsoftraditionalmethodsbyprovidingascalable,adaptive,anddata-drivensolutionforsustainablelandmanagement[27],[28].ThesimulationworkflowoftheproposedAI–RS–GISintegratedframeworkisillustratedinFig.3.GRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page12Fig.3.SimulationworkflowoftheproposedAI–RS–GISframeworkforagro-forestrysuitabilityanalysisandlandresourceoptimizationFig.3presentstheend-to-endworkflowoftheproposedframeworkforagro-forestrysuitabilityanalysisandlandresourceoptimization.Theprocessbeginswithdefiningobjectivesandcollectingmulti-sourcedata,includingsatelliteimagery,climate,soil,topographic,andsocio-economicdatasets.Theseinputsundergopreprocessingandfeatureextraction,wherekeyindicatorssuchasNDVI,elevation,andenvironmentalvariablesaregenerated.Theframeworkthenperformsparallelanalysisthroughtwomaincomponents:AI-basedmodelingandGIS-basedspatialanalysis.AImodels(e.g.,RF,SVM,ANN,CNN)areusedforclassificationandprediction,whileGIStechniquessuchasmulti-criteriadecisionanalysis(MCDA)andweightedoverlayareappliedforspatialsuitabilitymapping.Theoutputsfrombothcomponentsareintegratedandevaluatedbasedonperformancemetricsandspatialconsistency.Aniterativerefinementloopensurescontinuousimprovementofmodelperformance.Onceacceptableresultsareachieved,thesystemproceedstovalidationanddecisionsupportgeneration.Finally,theframeworkproducesactionableoutputs,includingagro-forestrysuitabilitymaps,priorityzones,landresourceoptimizationstrategies,andmanagementrecommendation.3.2StudyAreaDescriptionGRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Thestudyareaisselectedbasedonitsrelevancetoagro-forestrypracticesandenvironmentalvariability.Typically,suchregionsarecharacterizedbydiverseland-usepatterns,varyingclimaticconditions,andheterogeneoussoilproperties.Theselectedareaincludesagriculturallands,forestcover,andtransitionalzones,makingitsuitableforevaluatingtheeffectivenessofintegratedmodellingapproaches.Keyparameterssuchastemperature,rainfall,elevation,andvegetationdensityareconsideredtocapturetheenvironmentalcomplexityoftheregion.ThespatialextentofthestudyareaisdefinedusingGISboundaries,ensuringconsistencyindataintegrationandanalysis.Page133.3DataAcquisitionandSourcesTheproposedframeworkutilizesmulti-sourcedatasetstocapturethecomplexityofagro-forestrysystems.Remotesensingdata,includingsatelliteimageryfromplatformssuchasLandsatandSentinel,provideshigh-resolutionspatialinformationforland-useclassificationandvegetationanalysis.Environmentaldatasets,includingclimatevariables(temperature,rainfall),soilcharacteristics,andtopographicfeatures,areobtainedfrompubliclyavailabledatabasesandintegratedintotheanalysis.Additionally,socio-economicdatasuchasland-usepatterns,infrastructure,andaccessibilityareincorporatedtoenhancedecision-making.Theintegrationofthesediversedatasetsensuresacomprehensiverepresentationofthestudyarea,enablingmoreaccuratemodellingandanalysis.Table5summarizesthekeydatasetsusedinthisstudy.Table5:DataSourcesandDescriptionDataTypeSourceParametersResolutionPurposeSatelliteImageryLandsat,Sentinel-2Multispectralbands10–30mClimateDataMeteorologicaldatabasesTemperature,rainfallMonthly/AnnualFAO,surveysnationaltype,SoilorganiccarbonpH,VariesLandclassificationuseEnvironmentalanalysisSuitabilityassessmentSoilDataTopographicDataSocio-economicDataDEM(SRTM)Elevation,slope30mTerrainanalysisGISlayers,surveysLandinfrastructureuse,VariesDecisionsupport3.4DataPre-processingandFeatureEngineeringDatapre-processingisacriticalstepinensuringthequalityandconsistencyofinputdatasets.Remotesensingimageryundergoesradiometricandatmosphericcorrectiontoremovenoiseandimprovedataaccuracy.Geometriccorrectionandco-registrationareappliedtoalignmultipledatasetsspatially,ensuringconsistencyacrossdifferentdatalayers.Cloudmaskingtechniquesareusedtoeliminatecloud-coveredpixels,whichcanaffectthereliabilityofanalysis.GRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Featureengineeringinvolvestheextractionofmeaningfulvariablesfromrawdata.VegetationindicessuchasNDVI,EVI,andNDWIarecomputedtoassessvegetationhealthandwatercontent.Topographicfeatures,includingslopeandaspect,arederivedfromDigitalElevationModels(DEM),whilesoilandclimaticvariablesarestandardizedforintegrationintothemodellingprocess.ThesefeaturesserveasinputvariablesforAI-basedmodels,enhancingtheirpredictivecapabilities[29].Thedataprocessingandfeatureengineeringpipelineusedtopreparemulti-sourceinputsformodelingisillustratedinFig.4.Page14Fig.4.Datapreprocessingandfeatureengineeringpipelinefortransformingmulti-sourcegeospatialdataintostructuredinputsforAI-basedmodelingandGISanalysisFig.4presentsasystematicpipelineforconvertingrawmulti-sourcegeospatialdataintoacleanandstructureddatasetsuitableforAI-basedmodellingandGISanalysis.Theprocessbeginswithdataacquisitionfrommultiplesources,includingremotesensingimagery,environmental,soil,topographic,andsocio-economicdata.Inthepreprocessingstage,operationssuchasradiometricandgeometriccorrection,cloudmasking,andnormalizationareappliedtoensuredataqualityandconsistency.Thisisfollowedbyfeatureextraction,whererelevantindicatorssuchasspectralindices(NDVI,EVI),texturalfeatures,topographicvariables,andsoilandclimateattributesarederived.Thefeatureengineeringstagefurtherrefinesthedatasetthroughlayerstacking,featureselection,normalization,handlingofmissingvalues,anddatasetsplittingfortrainingandtesting.Finally,astructuredfeaturematrixisgenerated,whichservesasinputforAImodelsandGIS-basedspatialanalysis.Thepipelinealsoproducesintermediateoutputssuchasfeatureraster’s,statisticalsummaries,andcorrelationanalysistosupportmodeldevelopmentandvalidation.GRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.20863.5AI-BasedModellingandAnalysisPage15ThecoreanalyticalcomponentoftheframeworkinvolvestheapplicationofAI-basedmodelsforclassification,prediction,andoptimization.MachinelearningalgorithmssuchasRandomForest(RF),SupportVectorMachines(SVM),andArtificialNeuralNetworks(ANN)areusedforlanduseandlandcover(LULC)classification.Thesemodelsaretrainedusinglabelleddatasetsandvalidatedthroughcross-validationtechniquestoensurerobustness.Deeplearningmodels,particularlyConvolutionalNeuralNetworks(CNN’s),areemployedforadvancedimageanalysistasks,includingfeatureextractionandpatternrecognition.CNN’sareparticularlyeffectiveinhandlinghigh-resolutionsatelliteimagery,enablingaccurateclassificationofland-usecategoriesanddetectionofsubtlechangesinvegetationpatterns.Inadditiontoclassification,predictivemodelsaredevelopedtoestimatelandsuitability,biomass,andagriculturalproductivity.TheAI-basedmodelingframeworkusedforlandsuitabilitypredictionispresentedinFig.5.Fig.5.AI-basedmodelingframeworkforlandsuitabilitypredictionusingmachinelearninganddeeplearningtechniquesFig.5presentsthestructuredworkflowoftheAI-basedmodellingframeworkusedforlandsuitabilityprediction.Theprocessbeginswiththeinputfeaturesetderivedfromremotesensing,topographic,soil,climate,andsocio-economicdata.Theseinputsundergodatapreparationsteps,includingcleaning,normalization,featureselection,anddatasetsplittingintotrainingandtestingsubsets.Theframeworkthenappliesmultiplemachinelearninganddeeplearningmodels,suchasRandomForest(RF),SupportVectorMachine(SVM),ArtificialNeuralNetwork(ANN),andConvolutionalNeuralNetwork(CNN),usingk-foldcross-validationtoensurerobustmodeltraining.ModelGRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page16performanceisevaluatedusingstandardmetricsincludingaccuracy,precision,recall,F1-score,Kappacoefficient,andROC-AUC.Basedoncomparativeanalysis,thebest-performingmodelisselectedandusedforlandsuitabilityprediction.Theoutputsincludesuitabilitymaps,probabilitymaps,classifiedsuitabilityzones(high,moderate,marginal,andnotsuitable),anddecision-supportrecommendations.Aniterativeoptimizationloopisincorporatedtorefinemodelperformanceandimprovepredictionaccuracy.Toevaluatetheperformanceofthesemodels,standardmetricssuchasaccuracy,precision,recall,F1-score,andKappacoefficientareused.Table6presentstheevaluationmetricsappliedinthisstudy.Table6:ModelEvaluationMetricsMetricAccuracyPrecisionRecallF1-ScoreDescriptionPurposeOverallcorrectnessofpredictionsGeneralperformanceTruepositivesoverpredictedpositivesReliabilityofpredictionsTruepositivesoveractualpositivesSensitivityHarmonicmeanofprecisionandrecallBalancedperformanceKappaCoefficientAgreementbeyondchanceModelrobustness3.6GIS-BasedSpatialAnalysisGIS-basedspatialanalysisisusedtointegratetheoutputsofAImodelswithenvironmentalandsocio-economicdatasets.Multi-criteriadecisionanalysis(MCDA)isappliedtoevaluatelandsuitabilityforagro-forestrybyassigningweightstodifferentfactorsbasedontheirimportance.TechniquessuchasAnalyticalHierarchyProcess(AHP)andweightedoverlayanalysisareusedtocombinemultiplecriteriaandgeneratesuitabilitymaps.Spatialmodellingisfurtheremployedtoidentifypriorityzonesforagro-forestryinterventions.Thisinvolvestheanalysisofspatialpatternsandrelationshipsbetweendifferentvariables,enablingtheidentificationofareaswithhighpotentialforsustainablelanduse.TheintegrationofAIoutputswithGISanalysisenhancestheaccuracyandreliabilityofspatialdecision-making[30].TheGIS-basedmulti-criteriadecisionanalysisframeworkusedforlandsuitabilityassessmentisillustratedinFig.6.GRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page17Fig.6.GIS-basedmulti-criteriadecisionanalysis(MCDA)frameworkforagro-forestrysuitabilityassessmentandspatialdecision-makingFig.6presentstheGIS-basedmulti-criteriadecisionanalysis(MCDA)frameworkusedforagro-forestrysuitabilityassessment.Theprocessbeginswiththeidentificationofkeyevaluationcriteria,includinglanduse,vegetation,slope,elevation,soiltype,andclimaticfactors.Thesecriteriaarestandardizedtoacommonscaletoensureconsistencyacrossdifferentdatalayers.TheAnalyticalHierarchyProcess(AHP)isthenappliedtoassignrelativeweightstoeachfactorbasedontheirimportance,followedbyaconsistencychecktovalidatetheweightingscheme.Thestandardizedandweightedcriteriaareintegratedusingaweightedoverlayanalysistogenerateacompositesuitabilityindex.Thisindexissubsequentlyclassifiedintodifferentsuitabilityzones,suchashighlysuitable,moderatelysuitable,marginallysuitable,andnotsuitableareas.Thefinaloutputsincludesuitabilitymaps,priorityzones,anddecision-supportrecommendations,whichprovideastructuredbasisforsustainableagro-forestryplanningandefficientlandresourcemanagement.3.7DecisionSupportSystem(DSS)ThefinalstageoftheframeworkinvolvesthedevelopmentofaDecisionSupportSystem(DSS)thattranslatesanalyticalresultsintoactionableinsights.TheDSSprovidesinteractivetoolsforvisualization,analysis,andreporting,enablingstakeholderstoexploredifferentscenariosandmakeinformeddecisions.Outputsincludelandsuitabilitymaps,priorityzonesforagro-forestry,andrecommendationsforresourceallocation.TheDSSisdesignedtobeuser-friendlyandadaptable,allowingfortheincorporationofnewdataandcontinuousupdates.Thisensuresthatthesystemremainsrelevantandresponsivetochangingenvironmentalconditions.3.8WorkflowIntegrationandContinuousMonitoringGRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Animportantfeatureoftheproposedmethodologyistheincorporationofafeedbackloopforcontinuousmonitoringandimprovement.Thesystemcontinuouslyupdatesmodelsbasedonnewdata,enablingadaptivedecision-makingandimprovingpredictiveaccuracyovertime.Thisdynamicapproachensuresthattheframeworkremainsscalableandapplicabletodifferentregionsandenvironmentalconditions.Page184.Results4.1LandUseandLandCoverClassificationThelanduseandlandcover(LULC)classificationprovideacomprehensiveoverviewofthespatialdistributionofdifferentlandcategorieswithinthestudyarea.UsingtheproposedAI–RS–GISframework,theregionwasclassifiedintofivemajorcategories:forest,agriculturalland,waterbodies,built-upareas,andbarrenland.Theclassificationresultsindicateadiverselandscape,whereagriculturallandoccupiesthelargestportion,followedbyforestedareasconcentratedinrelativelylessdisturbedregions.Built-upareasaremainlylocatedneartransportationcorridorsandurbancentres,whilebarrenlandappearsinzoneswithpoorsoilqualityorlimitedvegetationcover.Toevaluatetheclassificationperformance,standardaccuracymetricswerecomputedbasedonvalidationsamples.Theresultsdemonstrateabalancedperformanceacrossallclasses,withrelativelyhigheraccuracyobservedforwaterbodiesandagriculturalareasduetotheirdistinctspectralcharacteristics.Slightconfusionwasobservedbetweenbarrenlandandsparsevegetation,whichiscommoninheterogeneousenvironments.TheclassificationaccuracyresultsaresummarizedinTable7.Table7:LandUseandLandCoverClassificationAccuracyClassForestAgricultureWaterBodiesBuilt-upBarrenLandOverallAccuracyKappaCoefficientProducerAccuracy(%)UserAccuracy(%)F1-Score(%)90.792.995.888.986.590.80.8891.592.196.387.685.9——91.192.596.088.286.2——Theresultsindicatethattheclassificationframeworkachievesreliableperformanceacrossallland-usecategories,makingitsuitableforsubsequentspatialandsuitabilityanalysis.ThespatialdistributionoflanduseandlandcoverclassesderivedfromtheclassificationprocessisshowninFig.7.GRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page19Fig.7.Landuseandlandcover(LULC)classificationmapofthestudyareageneratedusingtheproposedAI–RS–GISframeworkFig.7presentsthelanduseandlandcover(LULC)classificationmapofthestudyareageneratedusingtheproposedAI–RS–GISframework.Themapcategorizestheregionintomajorclasses,includingforest,agriculturalland,waterbodies,built-upareas,andbarrenland.Forestcoverispredominantlyobservedinthenorthernandelevatedregions,whileagriculturallandiswidelydistributedacrossthecentralvalleys.Built-upareasareconcentratedaroundmajorurbancenterssuchasMingora,indicatingpatternsofurbanexpansion.Thespatialdistributionhighlightsadiverselandscapeinfluencedbytopographyandhumanactivity.Thepresenceofdensevegetationinhigheraltitudesandagriculturaldominanceinlowlandareasreflectsthesuitabilityoftheregionforagro-forestrypractices.Additionally,theidentificationofbarrenandsparselyvegetatedareasprovidesusefulinsightsforlandrestorationandresourcemanagement.Thisclassificationservesasafoundationalinputforsubsequentsuitabilityanalysisanddecision-supportmodelling.Theclassificationperformanceisfurtherevaluatedusingaconfusionmatrix,aspresentedinFig.8.GRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page20Fig.8.Confusionmatrixillustratingtheclassificationaccuracyacrossdifferentland-useclassesFig.8presentstheconfusionmatrixusedtoevaluatetheperformanceoftheLULCclassification.Thematrixcomparesactualandpredictedclassesacrossfivecategories:forest,agriculture,waterbodies,built-upareas,andbarrenland.Thediagonalelementsrepresentcorrectlyclassifiedsamples,showingstrongclassificationperformanceacrossallclasses.Waterbodiesexhibitthehighestclassificationaccuracy,whileslightmisclassificationisobservedbetweenbuilt-upareas,agriculture,andbarrenland,whichisexpectedduetospectralsimilarities.Theoverallclassificationaccuracyis90.8%,withaKappacoefficientof0.88,indicatingahighlevelofagreementbeyondchance.Userandproduceraccuraciesfurtherconfirmthereliabilityoftheclassification,withmostclassesachievingvaluesabove85%.TheseresultsdemonstratethattheproposedAI–RS–GISframeworkprovidesrobustandconsistentclassificationperformance,formingareliablebasisforsubsequentsuitabilityanalysisandspatialmodelling.4.2Agro-ForestrySuitabilityMappingTheagro-forestrysuitabilitymappingresultsrevealthespatialvariationinlandpotentialforsustainableagro-forestrypractices.Byintegratingenvironmental,topographic,andsocio-economicfactors,thestudyidentifiesareasthatarehighlysuitable,moderatelysuitable,marginallysuitable,andnotsuitableforagro-forestryimplementation.GRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Highlysuitablezonesaregenerallycharacterizedbyfavourablesoilconditions,adequaterainfall,andmoderateslopes,whichsupportbothcropgrowthandtreedevelopment.Moderatelysuitableareasshowminorenvironmentalconstraints,whilemarginalzonesareaffectedbyfactorssuchaspoorsoilqualityorlimitedwateravailability.Notsuitableareasareprimarilyassociatedwithsteepslopes,degradedland,orurbanizedregions.Thespatialdistributionofagro-forestrysuitabilityzonesisillustratedinFig.9.Page21Fig.9.Agro-forestrysuitabilitymapshowinghighlysuitable,moderatelysuitable,marginallysuitable,andunsuitablezonesFig.9presentstheagro-forestrysuitabilitymapofthestudyarea,classifyinglandintohighlysuitable,moderatelysuitable,marginallysuitable,andunsuitablezones.Themaprevealsthathighlysuitableareasareprimarilylocatedinregionswithdensevegetation,favourablesoilconditions,andmoderateterrain,whilemoderatelysuitablezonesdominatealargeportionofthelandscape.Marginallysuitableareasarescatteredacrosstransitionalzones,whereasunsuitableregionsaremainlyconcentratedinsteep,barren,orenvironmentallyconstrainedareas.Thespatialdistributionhighlightstheinfluenceoftopography,vegetationcover,andenvironmentalfactorsonlandsuitability.Thecentralandnorthernregionsexhibithighersuitabilityduetobetterecologicalconditions,whileperipheralandelevatedareasshowlowersuitability.Thisclassificationprovidesaclearbasisforidentifyingpriorityzonesforagro-forestrydevelopment,supportingtargetedGRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086land-useplanningandefficientresourcemanagement.ThedistributionoflandacrossthesesuitabilitycategoriesispresentedinTable8.Page22Table8:Agro-ForestrySuitabilityDistributionSuitabilityClassHighlySuitableModeratelySuitableMarginallySuitableNotSuitableTotalArea(km²)Percentage(%)298.4365.7247.3197.81109.226.933.022.317.8100Theresultsindicatethatasignificantportionofthestudyareafallswithinhighlyandmoderatelysuitablecategories,highlightingstrongpotentialforagro-forestryexpansionandsustainableland-useplanning.Thedistributionoflandareaacrossdifferentagro-forestrysuitabilityclassesisillustratedinFig.10.Fig.10.Distributionoflandareaacrossagro-forestrysuitabilityclassesintheselectedstudyareaGRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page23Fig.10illustratesthedistributionoflandareaacrossdifferentagro-forestrysuitabilityclassesinthestudyarea.Theresultsshowthatmoderatelysuitableareasoccupythelargestproportion,followedbyhighlysuitableandmarginallysuitablezones.Incontrast,unsuitableandnotsuitableareasrepresentarelativelysmallershareofthetotallandarea,indicatingthatasignificantportionoftheregionhaspotentialforagro-forestrydevelopment.Thedistributionhighlightsafavorableoverallsuitabilitypattern,wheremorethanhalfofthestudyareafallswithinhightomoderatesuitabilitycategories.Thissuggestsstrongpotentialforsustainableland-useplanningandagro-forestryexpansion.Thepresenceofmarginalandunsuitablezonesalsoprovidesimportantinsightsfortargetedinterventions,suchaslandrehabilitationandenvironmentalmanagementstrategies.4.3ModelPerformanceAnalysisTheperformanceofthedevelopedmodelswasevaluatedtoassesstheireffectivenessinclassificationandpredictiontasks.A5-foldcross-validationstrategywasappliedtoensurerobustnessandminimizetheinfluenceofdatavariability.Thisapproachprovidesamorereliableestimateofmodelperformancebyaveragingresultsacrossmultipletrainingandtestingsplits.AsshowninTable9,allmodelsdemonstratestableandcompetitiveperformance,withaccuracyvaluesrangingfromapproximately86%to90%.TheConvolutionalNeuralNetwork(CNN)achievedthehighestoverallaccuracyof89.7%,alongwiththebestF1-scoreandKappacoefficient.Thisindicatesitsabilitytoeffectivelycapturecomplexspatialpatternsandrelationshipswithinthedataset.Thedatasetwassplitintotraining(70%)andtesting(30%)subsetsbeforeapplying5-foldcross-validationRandomForest(RF)performedconsistentlywell,particularlyinprecision,suggestingstrongcapabilityinminimizingfalsepositives.TheArtificialNeuralNetwork(ANN)showedslightlyhigherrecallthanprecision,indicatingitseffectivenessinidentifyingtruepositiveinstances.TheSupportVectorMachine(SVM)exhibitedcomparativelylowerrecall,whichmaybeattributedtoitssensitivitytoparametertuningincomplexdatasets.Table9:PerformanceComparisonofAIModels(5-FoldCross-ValidationResults)ModelAccuracy(%)Precision(%)Recall(%)F1-Score(%)KappaRandomForest(RF)88.6±1.489.2±1.687.3±1.888.2±1.5SVMANNCNN86.9±1.787.5±1.985.6±2.186.5±1.887.8±1.586.9±1.788.4±1.687.6±1.489.7±1.288.6±1.490.3±1.389.4±1.20.820.790.800.84Therelativelylowvariationacrossfoldsindicatesstablemodelperformance,withCNNdemonstratingslightlybettergeneralizationcapabilitycomparedtoothermodels.ThecomparativeperformanceoftheevaluatedmachinelearninganddeeplearningmodelsisillustratedinFig.11.GRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page24Fig.11.PerformancecomparisonofAImodelsbasedonaccuracy,precision,recall,F1-score,andKappacoefficientusing5-foldcross-validationFig.11presentstheperformancecomparisonofdifferentAImodels,includingRandomForest(RF),SupportVectorMachine(SVM),ArtificialNeuralNetwork(ANN),andConvolutionalNeuralNetwork(CNN),evaluatedusing5-foldcross-validation.Theresultsshowthatallmodelsachievestrongandconsistentperformanceacrossmultiplemetrics,includingaccuracy,precision,recall,F1-score,andKappacoefficient.Amongthem,theCNNmodeldemonstratesthehighestoverallperformance,particularlyinaccuracyandrecall,indicatingitseffectivenessincapturingcomplexspatialpatterns.Theinclusionofstandarddeviationvalueshighlightsthestabilityandreliabilityofthemodelsacrossvalidationfolds.WhileCNNslightlyoutperformstheothermodels,RandomForestalsoshowscompetitiveperformancewithstrongprecisionvalues.Incontrast,SVMandANNexhibitslightlylowerbutstillacceptableperformance.Overall,theresultsconfirmthattheproposedAI-basedframeworkprovidesrobustandreliablepredictionsforagro-forestrysuitabilityassessment.4.4EnvironmentalMonitoringInsightsTheintegrationofremotesensingdataenableddetailedenvironmentalmonitoring,particularlyinassessingvegetationhealthandlanddegradationpatterns.VegetationindicessuchasNDVIwereusedtoanalyzespatialandtemporalvariationsinvegetationdensity.TheresultsindicatethatareasclassifiedGRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page25ashighlysuitableexhibitconsistentlyhigherNDVIvalues,reflectinghealthiervegetationandbetterenvironmentalconditions.Conversely,regionsidentifiedasmarginallysuitableorunsuitableshowlowerNDVIvalues,indicatingreducedvegetationcoverandpotentialenvironmentalstress.Thesepatternsareconsistentwithobservedsoilconditionsandrainfalldistribution,reinforcingthereliabilityoftheintegratedanalysis.Additionally,zoneswithdecliningvegetationtrendswereidentifiedasareaspronetolanddegradation.Thesefindingsprovidevaluableinsightsfortargetedinterventions,suchassoilrestorationandsustainablelandmanagementpractices.ThespatialvariationinvegetationhealthacrossthestudyareaisillustratedinFig.12.Fig.12.SpatialdistributionofvegetationhealthbasedonNDVIvaluesacrossthestudyareaFig.12illustratesthespatialdistributionofvegetationhealthacrossthestudyareabasedonNDVIvalues.AreaswithhigherNDVIvalues,representedbydarkgreenshades,indicatedenseandhealthyvegetation,primarilyobservedinthenorthernandelevatedregions.Incontrast,lowerNDVIvalues,showninyellowtoredshades,correspondtosparsevegetationordegradedland,mainlyconcentratedaroundurbanareasandlow-lyingregions.TheNDVIpatternshighlightastrongrelationshipbetweenvegetationhealthandenvironmentalconditions,suchastopographyandlanduse.RegionswithhighNDVIvaluesalignwithhighlysuitableagro-forestryzones,whileareaswithlowervaluescorrespondtomarginalorunsuitableregions.Thisanalysissupportsthereliabilityofthesuitabilitymappingandprovidesvaluableinsightsforenvironmentalmonitoringandsustainablelandmanagement.GRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.20864.5DecisionSupportOutputsPage26Thefinaloutputsoftheproposedframeworkaredesignedtosupportpracticaldecision-makinginagro-forestryplanningandlandresourceoptimization.TheDecisionSupportSystem(DSS)integratestheresultsofAImodellingandGIS-basedanalysistogenerateactionablerecommendationsforstakeholders.Thesystemidentifiespriorityzonesforagro-forestryimplementation,enablingefficientallocationofresourcesandtargetedinterventions.Highlysuitableareasarerecommendedforexpansionofagro-forestrypractices,whilemarginalzonesaresuggestedforrehabilitationthroughsoilimprovementandconservationmeasures.Scenario-basedanalysisfurtherallowsdecision-makerstoevaluatedifferentland-usestrategiesandtheirpotentialoutcomes.Thekeydecision-supportoutputsaresummarizedinTable10.Table10:DecisionSupportOutputsOutputTypeSuitabilityMapsPriorityZonesDescriptionSpatialrepresentationofagro-forestrypotentialIdentificationofhigh-potentialareasResourceOptimizationPlansEfficientlandallocationstrategiesManagementRecommendationsGuidelinesforsustainablelanduseTheseoutputsprovideacomprehensivefoundationforinformeddecision-making,supportingsustainablelandmanagementandlong-termenvironmentalresilience.Thefinaldecision-supportoutputsforoptimizedland-useplanningandagro-forestryimplementationareillustratedinFig.13GRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page27Fig.13.Decisionsupportoutputmapshowingpriorityzonesandoptimizedland-usestrategiesforsustainableagro-forestrymanagementFig.13presentsthedecision-supportoutputmapderivedfromtheintegratedAI–RS–GISframework.Themaphighlightspriorityzonesforagro-forestrydevelopment,categorizedintohigh,moderate,andlowpriorityareas,alongwithrehabilitationandrestrictedzones.High-priorityareasaremainlyconcentratedinregionswithfavorableenvironmentalconditions,whilemoderateandlow-priorityzonesaredistributedacrosstransitionallandscapes.Rehabilitationzonesareidentifiedindegradedareas,andrestrictedzonescorrespondtoenvironmentallysensitiveorunsuitableregions.Thespatialdistributionofthesezonesprovidesactionableinsightsforsustainableland-useplanningandresourceoptimization.High-priorityareasarerecommendedforimmediateagro-forestryimplementation,whilemarginalanddegradedregionsrequirerestorationandsoilimprovementstrategies.Thisdecision-supportoutputenablespolicymakersandplannerstomakeinformed,data-drivendecisions,ensuringefficientallocationofresourcesandlong-termenvironmentalsustainability.5.DiscussionTheresultsdemonstratethattheproposedAI–RS–GISframeworkiseffectiveforlandsuitabilityanalysisandsustainableagro-forestryplanning.TheLULCclassificationrevealsaheterogeneouslandscapedominatedbyagriculturalandforestareas,indicatingstrongpotentialforagro-forestryexpansion.However,thepresenceofbarrenlandandbuilt-upareashighlightsenvironmentalpressuresandtheneedforbalancedland-usestrategies.GRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page28Thesuitabilityanalysisshowsthatasignificantportionofthestudyareafallswithinhighlyandmoderatelysuitablecategories,primarilyduetofavorablevegetation,soil,andterrainconditions.TheNDVIanalysisfurthersupportsthesefindings,showingaclearrelationshipbetweenvegetationhealthandsuitabilityclasses.AreaswithhigherNDVIvaluesalignwithhighlysuitablezones,whilelowervaluescorrespondtodegradedorunsuitableregions,confirmingthereliabilityoftheintegratedframework.Modelperformanceevaluationindicatesthatallmodelsprovidestableandreliableresults,withCNNachievingthehighestoverallperformance.However,theimprovementovertraditionalmodelsismoderate,suggestingthatsimplermodelsremainpracticalandefficient.Thedecision-supportoutputsidentifypriorityzonesfordevelopment,conservation,andrehabilitation,offeringactionableinsightsforpolicymakers.Overall,theframeworkprovidesascalableanddata-drivenapproachforsustainablelandresourceoptimization,withfutureimprovementspossiblethroughtemporalandclimate-basedanalysis.6.ConclusionThisstudypresentedanintegratedAI–RS–GISframeworkforagro-forestrysuitabilityanalysisandlandresourceoptimization.Theresultsdemonstratethatcombiningremotesensingdata,spatialanalysis,andartificialintelligenceenablesaccurateidentificationofsuitablezonesandsupportsinformedland-useplanning.Asignificantportionofthestudyareawasfoundtobesuitableforagro-forestry,highlightingitspotentialforsustainabledevelopment.Themodelevaluationshowsthatwhiledeeplearningapproachesprovideslightlybetterperformance,traditionalmachinelearningmodelsremaineffectiveandcomputationallyefficient.TheconsistencybetweenNDVI-basedenvironmentalanalysisandsuitabilitymappingfurthervalidatesthereliabilityoftheproposedapproach.Moreover,theframeworkprovidesapracticalandscalabledecision-supporttoolforpolicymakersandplanners,facilitatingefficientresourceallocation,environmentalconservation,andsustainableagro-forestrymanagement.7.FutureWorkFutureresearchcanenhancetheproposedframeworkbyincorporatingtemporalanalysisandmulti-yearsatellitedatatocaptureseasonalandlong-termland-usedynamics.Theintegrationofclimatechangeprojectionsandweathervariabilitycanfurtherimprovepredictiveaccuracyandadaptability.Advanceddeeplearningarchitecturesandhybridmodelsmaybeexploredtobetterhandlecomplexspatialpatterns.Additionally,theinclusionofsocio-economicandpolicy-drivenfactorscanstrengthendecision-makingrelevance.Expandingtheframeworktolargergeographicregionsandreal-timemonitoringsystemswouldalsoimproveitsscalabilityandpracticalimplementation.ReferencesGRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.20861.Gu,Z.,&Zeng,M.(2024).Theuseofartificialintelligenceandsatelliteremotesensinginlandcoverchangedetection:Reviewandperspectives.Sustainability,16(1),274.Page292.Aziz,D.,Rafiq,S.,Saini,P.,Ahad,I.,Gonal,B.,Rehman,S.A.,...&NabilaIliya,M.(2025).inintelligence:revolutionizingpestmanagementartificialRemoteagriculture.FrontiersinSustainableFoodSystems,9,1551460.sensingand3.Sadenova,M.,Beisekenov,N.,&Varbanov,P.S.(2024).AssessingtheeffectivenessofRS,GIS,andAIdataintegrationinanalysingagricultureperformancetoenablesustainablelandmanagement.DiscoverSustainability,5(1),478.4.Madane,D.A.,Tarate,S.B.,&Yousuf,A.(2026).IntegrationofArtificialIntelligence,RemoteSensing,andGISforSustainableLandManagementandPlanning.InAdvancementsinSoilConservation:FromTraditionalPracticestoAI-DrivenSolutions(pp.539-559).Cham:SpringerNatureSwitzerland.5.Selmy,S.A.,Kucher,D.E.,&Moursy,A.R.(2025).Integratingremotesensing,GIS,andAItechnologiesinsoilerosionstudies.InModernGeospatialApproachesforEnvironmentalMonitoringandManagement.IntechOpen.6.Farbo,A.,Sarvia,F.,DePetris,S.,Basile,V.,&Borgogno-Mondino,E.(2024).ForecastingcornNDVIthroughAI-basedapproachesusingsentinel2imagetimeseries.ISPRSJournalofPhotogrammetryandRemoteSensing,211,244-261.7.Judith,J.,Tamilselvi,R.,Beham,M.P.,Lakshmi,S.,Panthakkan,A.,Mansoori,S.A.,&Ahmad,H.A.RemoteSensingBasedCropHealthClassificationUsingNDVIandFullyConnectedNeuralNetworks.arXiv2025.arXivpreprintarXiv:2504.10522.8.Escobar-López,A.,Castillo-Santiago,M.Á.,Mas,J.F.,Hernández-Stefanoni,J.L.,&López-Martínez,J.O.(2024).Identificationofcoffeeagroforestrysystemsusingremotesensingdata:areviewofmethodsandsensordata.GeocartoInternational,39(1),2297555.9.Giannetti,F.,Puletti,N.,Puliti,S.,Travaglini,D.,&Chirici,G.(2020).AssessmentofUAVphotogrammetricDTM-independentvariablesformodellingandmappingforeststructuralindicesinmixedtemperateforests.EcologicalIndicators,117,106513.10.Zhu,X.X.,Tuia,D.,Mou,L.,Xia,G.S.,Zhang,L.,Xu,F.,&Fraundorfer,F.(2017).Deeplearninginremotesensing:Acomprehensivereviewandlistofresources.IEEEgeoscienceandremotesensingmagazine,5(4),8-36.11.Kamran,M.,Tanveer,K.,Khalid,N.,Khalil,M.N.,Ahmad,B.,Arooj,A.,...&Mir,S.Z.(2025).Integratingadvanceddeeplearningalgorithmsforclimatesystems:Enhancingweatherforecastlong-termclimatepredictions.Spectrumofaccuracy,real-timeclimatemonitoring,andEngineeringSciences,365-403.12.Cian,F.,Marconcini,M.,&Ceccato,P.(2018).NormalizedDifferenceFloodIndexforrapidfloodmapping:TakingadvantageofEObigdata.Remotesensingofenvironment,209,712-730.13.Belgiu,M.,&Drăguţ,L.(2016).Randomforestinremotesensing:Areviewofapplicationsandfuturedirections.ISPRSjournalofphotogrammetryandremotesensing,114,24-31.14.Mehmood,B.,Rani,G.,Khalid,N.,Majeed,M.K.,Ahmad,B.,Qamar,S.,&Saeed,M.(2025).Developmentofahybridartificialintelligenceframeworkforaccurateforecastingofsolarpowergenerationusingmachinelearningalgorithmsandtime-seriesanalysis.SpectrumofEngineeringSciences,613-636.15.Foody,G.M.(2002).Statusoflandcoverclassificationaccuracyassessment.Remotesensingofenvironment,80(1),185-201.16.Sharma,A.,Paliwal,K.K.,&Onwubolu,G.C.(2006).Class-dependentPCA,MDCandLDA:Acombinedclassifierforpatternclassification.PatternRecognition,39(7),1215-1229.GRJNST,Volume:04-Issue2(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07G.2086Page3017.Grill,A.,&Cleary,D.F.(2003).DiversitypatternsinbutterflycommunitiesoftheGreeknaturereserveDadia.BiologicalConservation,114(3),427-436.18.Koklu,M.,Cinar,I.,&Taspinar,Y.S.(2021).Classificationofricevarietieswithdeeplearningmethods.Computersandelectronicsinagriculture,187,106285.19.Boles,S.H.,Xiao,X.,Liu,J.,Zhang,Q.,Munkhtuya,S.,Chen,S.,&Ojima,D.(2004).LandcovercharacterizationofTemperateEastAsiausingmulti-temporalVEGETATIONsensordata.RemoteSensingofEnvironment,90(4),477-489.20.Aslam,M.F.,Ahmed,M.U.,Akbar,K.,&Amjad,A.(2025).THEINFLUENCEOFDIGITALLEADERSHIPONSUSTAINABLEPERFORMANCE:THEMEDIATINGROLESOFDIGITALORGANIZATIONALCULTURE,ANDTHEMODERATINGROLEOFORGANIZATIONALAGILITY.JournalofManagementScienceResearchReview,4(2),62-90.21.Zhang,C.,Kerner,H.,Wang,S.,Hao,P.,Li,Z.,Hunt,K.A.,...&Shen,Y.(2025).Remotesensingforcropmapping:Aperspectiveoncurrentandfuturecrop-specificlandcoverdataproducts.RemoteSensingofEnvironment,330,114995.22.Singh,A.K.,Ghosh,A.,Sannagoudar,M.S.,Kumar,R.V.,Kumar,S.,Singh,P.D.,&Ahamad,S.(2022).Sustainableagriculturesystemsandtechnologies.SustainableAgricultureSystemsandTechnologies,279-294.23.Ennouri,K.(2026).USEOFREMOTESENSINGFORPRECISION.AIandDeepLearningEnabledSurveillanceSystemUsingImageProcessing,173.24.Mohammed,K.,Kpienbaareh,D.,Kerr,R.B.,Wang,J.,Luginaah,I.,Lupafya,E.,...&Mkandawire,M.(2026).IntegratingparticipatoryGIS,remotesensing,andexplainablemachinelearningtoassessforestprovisioningservices.EnvironmentalImpactAssessmentReview,117,108245.25.Akbar,K.(2021).HowEco-innovation&GreenManufacturingAffectsSustainabilityPerformance:AQuantitativeStudyInPakistan.InternationalResearchJournalofModernizationofEngineeringTechnologyandScience,3(6),3785-3792.26.Siddiqui,M.H.S.,Abbasi,M.D.,Mir,S.Z.,Nadeem,G.,Majeed,M.K.,Khawer,S.K.,...&Kashif,M.(2025).PredictionofmaximumairtemperaturefordefiningheatwaveinvariousstatesofPakistanusingmachinelearningalgorithm.SpectrumofEngineeringSciences,942-960.27.Nuwarapaksha,T.D.,Udumann,S.S.,Dissanayaka,N.S.,&Atapattu,A.J.(2025).FutureProspectsandEmergingTrendsinAgroforestryResearch.AgroforestryforaClimate-SmartFuture,497-518.28.Bhardwaj,D.R.,&Bhatia,A.K.(2025).LanduseandlandcoveranalysisinagroforestryandotherlandusecategoriesinShimladistrictofHimachalPradesh:aGIS-basedanalysisinNorthWesternHimalayas.DiscoverForests,1(1),33.29.Machireddy,S.R.(2023).Naturalresourcemanagementusingremotesensingandgeographicinformationsystems.EnvironmentalScienceandEngineering,2(2),73-82.30.Udumann,S.S.,Dissanayaka,N.S.,Nuwarapaksha,T.D.,&Atapattu,A.J.(2025).AgroforestryEducationandTrainingfortheNextGeneration.InAgroforestryforaClimate-SmartFuture(pp.449-472).IGIGlobalScientificPublishing.GRJNST,Volume:04-Issue3(2026)/ISSNP:2790-7643ArticleID:2086https://doi.org/10.53762/grjnst.04.03.07