Airborne laser scanning (LiDAR) is a popular method for operational forest assessment. In the course of this development, classic regression methods have been replaced by nonparametric or Bayesian methods. Accurate estimates with current methods require hundreds of sample plots, which is costly. The authors propose replacing most of these with plots collected during earlier projects in different, but similar, forests. However, using such replacement plots requires resolution of two problems. The first is overcoming differences between different LiDAR scanners, scanning parameters, and scanning condidons, and the second is avoiding bias. The authors propose a method that resolves both problems and uses only approximately 50 new plots. The method is shown to provide satisfactory estimates for total forest parameters and to be robust against random variation in the selection of the new plots. [PUBLICATION ABSTRACT]
展开▼
