Test of application of centroid and importance sampling in a point-3P forest inventory

Abstract

The volume of 66.8 ha of mature native eucalypt forest in southeast New South Wales was estimated by point-3P sampling using three methods: detailed dendrometry, centroid sampling and importance sampling. The derived resource estimates were 11 790 ± 1026, 11 150 ± 959 and 11 520 ± 1014 m 3, respectively. The dendrometry required for centroid and importance sampling was 82% and 84% less respectively than that required for detailed measurement. Compared with detailed dendrometry, the centroid method underestimated the mean bole volume of second-stage sample trees by 6.9%, whereas importance sampling underestimated by only 2.4%. However, the bole volumes derived from centroid sampling were more closely related to those derived from detailed dendrometry ( r 2 = 0.976) than those derived from importance sampling ( r 2 = 0.935), and the two regressions were significantly different. More importantly, the centroid method proved superior to importance sampling in terms of both the mean-square error and tolerance interval. The reason for this apparent anomaly is discussed. All three methods were equally satisfactory for predicting the assortment of log products (saw-log, pulp, cull). The advantages and disadvantages of applying centroid and importance sampling to standing sample trees are listed.