Correcting for nondetection in estimating forest characteristics from single-scan terrestrial laser measurements

Abstract

A problem in the single-scan setup of terrestrial laser scanning is that some trees are shaded by others and therefore not detected in the scan. A basic estimator for forest characteristics such as tree density or basal area is based on the visible area of a scanner. However, simply compensating for nondetection by the visible area may result in considerable bias even in Poisson forests, especially if the detection of a tree depends on its size. We propose a new estimator that is a generalization of the visible area based estimator. Most importantly, the new estimator allows different detection rules; for example, full or partial visibility of a tree can be required for detection. By a simulation study, it is shown to work adequately in different types of simulated and empirical forests with different detection rules.