Statistic tools for forest inventory auditing in Eucalyptus spp. stands

Abstract

In order to audit the accuracy of data collection, which may influence the estimated stock of timber in forest stands, this study aimed to evaluate the use of the statistical tools t-test of Student at the significance levels of 95% and 99%, and graphical analysis of the residuals in the auditing of forest inventories. Six forest inventory plots were randomly selected from a total of 120 plots, and the variables circumference at 1.30 m high (CBH) and total height of the trees were measured. To estimate the volumes, models proposed by Trorey to the total height and Meyer for the individual volume were fitted; which presented respectively: adjusted coefficient of determination values (R2aj%) of 47.30% and 98.71% and standard error of estimate (Syx%) of 8.06% and 4.35%. After checking normality and variance equality through the test of Kolmogorov-Smirnov and F test, respectively, values of CBH, total height and timber volume estimated in the forest inventory were confronted with the audited measures. It could be observed that the estimated volumes and CBH values did not differ statistically, whereas the total height values showed significant difference for the t-test on two plots at the level of 95%, and in one plot at the level of 99%. By graphical analysis of residuals in the total height measurements, overestimation trends in two plots were identified and underestimation in one also. The plots that showed statistical differences were the same that showed trends in the measurements of the total height, proving the effectiveness of the methods.