摘要
Background:Currently,thecommonandfeasiblewaytoestimatethemostaccurateforestbiomassrequiresgroundmeasurementsandallometricmodels.Previousstudieshavebeenconductedonallometricequationsdevelopmentforestimatingtreeabovegroundbiomass(AGB)oftropicaldipterocarpforests(TDFs)inKalimantan(IndonesianBorneo).However,beforetheuseofexistingequations,avalidationfortheselectionofthebestallometricequationisrequiredtoassessthemodelbiasandprecision.Thisstudyaimsatevaluatingthevalidityoflocalandpantropicalequations;developingnewallometricequationsforestimatingtreeAGBinTDFsofKalimantan;andvalidatingthenewequationsusingindependentdatasets.Methods:Weused108treesamplesfromdestructivesamplingtodeveloptheallometricequations,withmaximumtreediameterof175cmandanother109samplesfrompreviousstudiesforvalidatingourequations.WeperformedordinaryleastsquareslinearregressiontoexploretherelationshipbetweentheAGBandthepredictorvariablesinthenaturallogarithmicform.Results:Thisstudyfoundthatmostoftheexistinglocalequationstendedtobebiasedandimprecise,withmeanrelativeerrorandmeanabsoluterelativeerrormorethan0.1and0.3,respectively.WedevelopednewallometricequationsfortreeAGBestimationintheTDFsofKalimantan.Throughavalidationusinganindependentdataset,wefoundthatourequationswerereliableinestimatingtreeAGBinTDF.Thepantropicalequation,whichincludestreediameter,wooddensityandtotalheightaspredictorvariablesperformedonlyslightlyworsethanournewmodels.Conclusions:OurequationsimprovetheprecisionandreducethebiasofAGBestimatesofTDFs.Localmodelsdevelopedfromsmallsamplestendtosystematicallybias.AvalidationofexistingAGBmodelsisessentialbeforetheuseofthemodels.
出版日期
2017年02月12日（中国期刊网平台首次上网日期，不代表论文的发表时间）
