Influences of mean top height definition and sampling method on errors of estimates in New Zealand’s forest plantations

Abstract

Background: A study was undertaken of 51 stand inventories to compare two alternative mean top height (MTH) calculation methods prevalent in New Zealand, and to evaluate the consequences of creating height versus diameter at breast height (H-D) curves at a stand-level during inventories as opposed to fitting H-D curves at a plot-level. Methods: The dataset was separated into two groups; one with plots having less than 6 heights measured and one with more than 5 heights measured. MTH was calculated using all possible combinations of the two calculation methods and with H-D curves either at a stand-level or a plot-level. Graphs were prepared to compare the 4 alternative MTH estimation techniques for all plots. In addition standard deviations of MTH between plots were calculated within stands, and then these were compared for different MTH calculation methods using interleaved histograms and with a mixed effects analysis of variance. Results: Results showed that the two MTH calculation methods were almost identical so long as H-D curves were fitted at a plot-level, but they differed substantially when curves were fitted at a stand-level. In addition, fitting H-D curves at a stand-level reduced independence of samples, resulting in substantial decreases in estimated standard deviations in MTH within samples, thereby artificially reducing confidence intervals around sample estimates. Conclusions: Inventory estimates of MTH were found to depend on calculation method, and so a standard definition is required. In addition, H-D curves fitted at a stand level undermined the assumption that sampling units were independent, and thereby reduced estimated variation between plots by up to 69%, depending on MTH calculation method. Forest inventory procedures in New Zealand’s forest plantations should be redesigned to enable accurate definition of confidence intervals around sample estimates, and to facilitate the use of inventories for estimating variation in productivity across landscapes.