简介：Background:Remotesensing-basedmappingofforestEcosystemService(ES)indicatorshasbecomeincreasinglypopular.TheresultingmapsmayenabletospatiallyassesstheprovisioningpotentialofESsandprioritizethelanduseinsubsequentdecisionanalyses.However,themappingisoftenbasedonreadilyavailabledata,suchaslandcovermapsandotherpubliclyavailabledatabases,andignoringtherelateduncertainties.Methods:Thisstudytestedthepotentialtoimprovetherobustnessofthedecisionsbymeansoflocalmodelfittinganduncertaintyanalysis.Thequalityofforestlanduseprioritizationwasevaluatedundertwodifferentdecisionsupportmodels:eitherusingthedevelopedmodelsdeterministicallyorincorporationwiththeuncertaintiesofthemodels.Results:PredictionmodelsbasedonAirborneLaserScanning(ALS)dataexplainedthevariationinproxiesofthesuitabilityofforestplotsformaintainingbiodiversity,producingtimber,storingcarbon,orprovidingrecreationaluses(berrypickingandvisualamenity)withRMSEsof15%–30%,dependingontheES.TheRMSEsoftheALS-basedpredictionswere47%–97%ofthosederivedfromforestresourcemapswithasimilarresolution.Duetoapplyingasimilarfieldcalibrationsteponbothofthedatasources,thedifferencecanbeattributedtothebetterabilityofALStoexplainthevariationintheESproxies.Conclusions:Despitethedifferentaccuracies,proxyvaluespredictedbyboththedatasourcescouldbeusedforapixel-basedprioritizationoflanduseataresolutionof250m~2,i.e.,inaconsiderablymoredetailedscalethanrequiredbycurrentoperationalforestmanagement.TheuncertaintyanalysisindicatedthatmapsoftheESprovisioningpotentialshouldbepreparedseparatelybasedonexpectedandextremeoutcomesoftheESproxymodelstofullydescribetheproductionpossibilitiesofthelandscapeundertheuncertaintiesinthemodels.

简介：Background:Treelinedynamicshaveinevitableimpactsontheforesttreelinestructureandcomposition.ThepresentresearchsoughttoestimatetreelinemovementandstructuralshiftsinresponsetorecentwarminginCehennemdere,Turkey.Afterimplementinganatmosphericcorrection,thegeo-shiftingofimageswasperformedtomatchimagestogetherforaperpixeltrendanalysis.WedevelopedanewapproachbasedontheNDVI,LST(landsurfacetemperature)data,airtemperaturedata,andforeststandmapsfora43-yearperiod.Theforesttreelineborderwasmappedontheforeststandmapsfor1970,1992,2002,and2013toidentifyshiftsinthetreelinealtitudes,andthenprofilestatisticswerecalculatedforeachperiod.Twentysampleplots(10×10pixels)wereselectedtoestimatetheNDVIandLSTshiftsacrosstheforesttimberlineusingper-pixeltrendanalysisandnon-parametricSpearman’scorrelationanalysis.Inaddition,thespatialandtemporalshiftsintreelinetreespecieswerecomputedwithintheselectedplotsforfourtimeperiodsontheforeststandmapstodeterminethepioneertreespecies.Results:Astatisticallysignificantincreasingtrendinallclimatevariableswasobserved,withthehighestslopeinthemonthlyaveragemeanJulytemperature(tau=0.62,ρ<0.00).Theresultantforeststandmapsshowedageographicalexpansionofthetreelineinboththehighestaltitudes(22m–45m)andthelowestaltitudes(20m–105m)from1970to2013.TheperpixeltrendanalysisindicatedanincreasingtrendintheNDVIandLSTvalueswithintheselectedplots.Moreover,increasesintheLSTwerehighlycorrelatedwithincreasesintheNDVIbetween1984and2017(r=0.75,ρ<0.05).CedruslibaniandJuniperuscommunisapp.weretwopioneertreespeciesthatexpandedandgrewconsistentlyonopenlands,primarilyonrocksandsoil-coveredareas,from1970to2013.Conclusion:Thepresentstudyilustratedthatforesttreelinedynamicsandtreelinestructuralchangescanbedetectedusingtwodata