A spatial approach to edge effect modelling for plantation forestry

[摘要] ENGLISH ABSTRACT: One of the major objectives in plantationforestry is toachieveahighlevelofhomogeneityofdistributionanddimensionoftreeswithinthestand.Preciseplantinggeometries,intensivesilvicultureandgeneticselectionareusedtoachievethishomogeneity.However,anaturalvariabilityisstillintroducedbymicro-­‐siteconditionsanddisturbances.Asubstantialsourceofvariationiscausedbyedgeeffectsofneighbouringstandsorotherlanduseforms.Theedgeeffectcausestreesatthestandedgetodevelopdifferentlyfromtreesintheinteriorofthestand.Theoverarchingobjectiveofthisstudyistosimulatetheedgeeffectbasedonaveragestandinteriorvariablesastypicallyreceivedfromanenumerationandspatialinformationonthecurrentandhistoricstandneighbourhood.Withre-­‐introducingthisnaturalvarianceaswellasitsspatialpattern,weexpecttoderiveimprovedplanninginformation.Amajoraimisthusseparatingtheeffectoftheedgeinteractionfromtheotherfactorscontributingtostandvarianceandquantifyingtheresultintermsofstandoutput.Amethodologyisintroducedforquantifyinginteractionatstandedgesbetweenagivenstandanditsneighbouringstandsoveritslifetime.Transferringtheedgeinteractionvaluefromtheedgestoallthetreeswithinthestandisthendonebyapplyinginversedistanceweightinginterpolationfromtheedgestothetreepositionwithinthestand.Onceanedgeinteractionvaluehasbeencalculatedforeachpoint,theextentoftheedgeeffectisquantified.Thespatialextentoftheedgeeffectisderivedempiricallyfromanexistingfullyspatiallymappedstandbymeansofbreakpointregression.Theexpectedvarianceasaresultofedgeinfluenceisthenquantifiedbyproducingasetofmodels,whichcanreproducetheeffectoftheedgeinteractionontreeheight,diameterandvolume.Theedgeeffectistreatedasadynamicinteractionforwhichthetemporalaspectneedstobeconsidered,becausethecurrentspatialstructureofastandisinfluencedbyitscurrentneighbourhood,butalsobythehistoricdevelopmentoftheneighbourhoodinrelationtothestandinquestion.Eachstandthereforeundergoesanedgeeffectwhichiscompletelyuniquetothatstand,withinagiventimeperiod.Forthisreasonthepresentedmethodologyisaspatial-­‐temporalone,aimedatprovidingawayinwhichgrowthandyieldforestmodellingcanbeaugmentedbytheinclusionoftheedgeeffectinapracticalway.Toexplicitlyquantifyedgeeffects,thenaturalvariancehadtobeseparatedintoacomponentexplainedbyedgeeffectandasecondcomponentintroducedbyotherfactorssuchasmicrositeconditionsanddisturbance.Thesecondcomponentistreatedasanunexplainedresidualvariance.Inordertoprovidearealisticsimulationofastandoutputatafiner,treelevel,thissecondstandvarianceneedsnonethelesstobequantified.Thevarianceattributabletofactorsotherthantheedgeeffectismimickedbygeneratingarandomnumberbymeansofaparameterisedstochasticprocessbasedonthevarianceoftheinnerstandregion,whichisbeyondthereachoftheedgeeffect.Inthisway,arealisticspatialpatternofaplantationforeststand,takingintoaccounttheedgeeffectandcombiningitwiththenaturalstandvarianceisachieved.Thisstudy,withinthefieldofplantationforestmanagement,aspirestolanduseoptimizationbothintermsofproductivecapacityestimationandfortheprovisionofinformationforeffectivelandusemanagementplanning.ItmakesuseofopensourcesoftwareresourcesnamelytheRframeworkandQGISandexploresaerialstereophotogrammetryasanoptionfordatacollection.

[发布机构] Stellenbosch University