Top height estimation in lodgepole pine sample plots

Top height definitions are often based on the heights of a certain number of the largest trees per unit area, such as the largest 100/ha. Recognizing that results vary with the extent of the reference area, this area is specified in the British Columbia definition, basing top height on the largest tree in a 0.01-ha plot. The problem is how to estimate top height when data is available for larger plots, without the information needed to subdivide them into 0.01-ha subplots. The usual largest 100/ha overestimates the correct value, and we find that the bias can be substantial. We evaluate two alternatives for natural lodgepole pine stands, using data from 0.04- and 0.08-ha sample plots. The improved estimators considerably reduce bias, although some bias due to spatial size autocorrelations remains. Autocorrelation was found to be predominantly positive, and some implications for growth and yield prediction are mentioned.