Predicting Attributes of Regeneration Forests Using Airborne Laser Scanning

Forest inventory attributes including dominant height (H-d), mean tree height (H-t), total number of stems (N-t), and number of dominant stems (N-d), were predicted in regeneration forests using metrics from airborne laser scanning (ALS) data. The study was conducted in Norway, using field data from 48 sample plots and 173 validation plots distributed in 19 regeneration stands. The stand-level accuracies, obtained in terms of relative root mean square error were 17%, 28%, 47%, and 18% for H-d, H-t, N-t, and N-d, respectively. None of the predicted attributes differed significantly from the field-measured values at stand level. Furthermore, the influences of echo categories, ground thresholds, low tree height, transformation of response variable and stratification by site productivity, and species composition on the prediction errors (residuals) were assessed. All factors except selection of echo categories and site productivity affected the residuals of one or more of the forest attributes. Thus, in addition to documenting the accuracy obtained for tree heights and number of stems in regeneration stands, the study elucidates methodological issues in ALS-aided regeneration forest inventories.