Abstract
Light detection and ranging (lidar) has been successfully used to describe a wide range of forest metrics at local scales. However, little research has tested the general applicability of this technology to describe commercially important stand dimensions, such as total stem volume (V), at national levels across broad environmental gradients.
Using an extensive national data set covering the spatial extent of Pinus radiata plantation forests in New Zealand, the key objectives of this study were to (1) develop regression models to best describe V for P. radiata from lidar metrics and (2) investigate whether these relationships could be improved using coincident environmental and stand-level information. Development of relationships between lidar metrics and forest volume are of particular importance for P. radiata, as this species constitutes approximately 90% of the 1.8 Mha plantation resource.
Using lidar mean height and the percentage of lidar ground returns, the initial model (model 1) accounted for 85% of the variance in V. Addition of stand stocking (number of stems ha−1), measured within the plots, to the model (model 2) significantly (p < 0.001) improved predictions, with R 2 increasing to 0.86 and the root mean square error declining from 80.1 m3 ha−1 to 71.6 m3 ha−1. For both models, partial responses show V to be most sensitive to lidar mean height, which was included in the model as a second-order polynomial.
Although environmental variables are established determinants for V, their inclusion did not significantly improve either model 1 or 2. Residual values for both models showed little apparent bias when plotted against stand-level information or a wide array of environmental variables, supporting the general applicability of these relationships.
Acknowledgements
We are very grateful to the Ministry for the Environment (MfE) for providing permission to use the data described in this paper. We acknowledge the assistance of Mark Kimberley for providing the data to us in a usable form and for advice on analysis. We thank Marie Heaphy for constructing