Improving the Efficiency and Precision of Tree Counts in Pine Plantations Using Airborne LiDAR Data and Flexible-Radius Plots: Model-Based and Design-Based Approaches

Abstract

This paper explores and develops design-based and model-based methods which are suited to sampling strategies developed for LiDAR-assisted plantation inventories. Much of the model-based theory is either recent or adapted from other areas of sampling. The design-based theory extends and adapts previous work to the present situation. The methodology is developed around the increasing utility and precision of LiDAR as a sampling tool for operational forest inventory. Flexible-radius plots, as a means of optimizing the sampling effort, are examined from a sampling perspective. Mixed models are also employed to model the residual variance using specified correlation structures and this includes predictors which utilize local trend such as those employed in kriging. In the design-based setting, model-assisted estimators are used, including regression and ratio estimators. A plot-based survey of a young, single-aged stand located within a Pinus radiata plantation in the northern tablelands of New South Wales is used to illustrate the theory. Model covariates are obtained from airborne laser scanning (LiDAR) data.