A new sampling technique for estimating the mean height

A new sampling technique for easily estimating arithmetic mean height without knowledge of number of trees is proposed. This sampling technique was created by applying the geometrical probability to sampling space that is composed of total number of trees and height, HM, which is greater than the maximum tree height in the forest stand. As a result of a theoretical study, an estimator, HM(lambda), the unbiased estimator of arithmetic mean height was developed, where (lambda) is a discrete random variable that can take the value 1 if a sample tree is counted; otherwise it is 0. This estimator was tested for spatial patterns that are partly clumped, such as an aggregated distribution, and that are partly uncrowded, such as a random distribution and a regular distribution. Computer simulation was performed by modeling a forest stand and sampling an actual forest stand. This sampling technique was also found to be efficient for aggregated spatial patterns because P (mean value of absolute rate of error) and CVy (coefficient of variation of an estimator yj in die jth computer simulation) obtained by using this sampling technique were stable in four stands with different spatial patterns if 40 points were sampled per stand. This sampling technique may also be effective for stands of partly clumped spatial patterns because it is less time-consuming and have high precision.