A method for estimating tree composition and volume using harvester data

This research article introduces a method that can be used to estimate tree composition and volume of arbitrary subdivisions of a logged stand. The method uses spatial data that is generated with a harvester to simulate individual tree locations. The simulation uses two probability density functions: the distance and the angle from the harvester at which the tree is cut. The average estimated volume root mean squared error varied from 4% for 0.4 ha subregions to 29% for 0.03 ha subregions. The stand subdivision method affected the accuracy of volume estimation only in the smallest subregions. Compared with the use of harvester data as such, i.e. without tree location simulation, the improvement in total and species-wise volume estimates varied between 5 and 35%. The data produced by the method can be used as a field data source for remote sensing methods as well as a verification data set for field inventories. However, a question remains over the generality of the model parameters used.