Construction Models For Estimating DBHOB And Total Tree Height Based On Intelligent Harvester Head Data

Harvester as new generation equipment,greatly improve harvesting efficiency.It can cut trees into logs based on product specifications which set by people.Harvester head as one important part of harvester,it can measure and collect massive data,such as small end overbark diameter of log,log length,log volume and so on.These data can be combined with other data from different sources for relevant research.However,harvester head data does not include total tree height(TH)and DBHOB,which commonly used in forest models.Therefore,based on the features of a harvester head dataset and a stump height dataset,using 325 1 sample trees of taper measurements of Pinus radiata to construct experimental data,then building models to estimate DBHOB and TH for individual trees in harvester head data.The first step for DBHOB estimate model is to choose 30 heights for each tree,from 0.05 m to 2.95 m,the interval is 0.1 m,through PCHIP method to estimate diameters at the 30 heights.A system model includes 30 equations which represent relationships between DBHOB and diameter at the 30 heights,respectively.Using ordinary least squares method to estimate parameters of each equation,total obtain 30 sets of conversion factors.Next,choose another 29 heights for verify this model,through PCHIP method to estimate diameters at the 29 heights.Using three methods to estimate conversion factors at the 29 heights,curve fitting method,linear interpolation and nearest interpolation,respectively.The result showed that:using linear interpolation method to obtain conversion factors and to estimate DBHOB is best.In addition,average relative error and total relative error of DBHOB estimate model are all in±5%,can be applied to the actual production.For estimating total tree height,this paper proposed two models.One is based on Varjo's model,using overbark diameter replace underbark diameter,to meet the actual situation of data recorded in harvesting operations.The other is based on iterative method.The comparison result showed that:the modified Varjo's model is more accurate than the iterative model as its MSEP is smaller,can be used preferentially in practical application.On the other hand,the modified Varjo's model is less flexible,it cannot be used if missing same harvesting information,while the iterative model can be used normally.