Key points are not available for this paper at this time.
Abstract Optimisation of economic return from forests requires that comprehensive forest inventory data be available to support the design of harvesting strategies. Such inventory data can potentially be obtained by remote sensing. This study investigates the accuracy with which wood volume (m3 ha -1) in a plantation forest can be calculated from Landsat TM data at the pixel and foreststand spatial scales. Wood volumes were estimated from regression analysis, nonparametric line-fitting, and an N-dimensional K-nearest-neighbour classification scheme. At the pixel scale, relations between Landsat data and measured wood volume were found to be significant but weak, with r2 values of 0.3, and with correspondingly poor estimates of wood volume (root-mean-square errors rmses of 100 m3 ha -1). By averaging the pixel-scale estimates, wood volume estimates of acceptable accuracy were obtained for forest-stand areas of about 40 ha (rmses of 46 m3 ha -1). Parametric regression performed slightly better overall than non-parametric line fitting techniques for estimating wood volume. Estimates of similar accuracy to those obtained by regression were also given by NK-classification at the pixel-scale, provided K was large ( 15), although the classifier produced biased results at the forest-stand scale. It is concluded that Landsat TM only provides an acceptable data source for estimating wood volumes in plantation forests for areas of about 40 ha and larger. The very low dynamic range in the Landsat data is probably a significant factor limiting its use for inventory at more detailed scales.
Trotter et al. (Tue,) studied this question.