Measurement variability error for estimates of volume change

Abstract

Using quality assurance data, measurement variability distributions were developed for attributes that affect tree volume prediction. Random deviations from the measurement variability distributions were applied to 19 381 remeasured sample trees in Maine. The additional error due to measurement variation and measurement bias was estimated via a simulation study for various components of volume change. In comparison with sampling error, the error due to measurement variation was relatively small. When biases in measurements had contradictory effects on the calculation of individual tree volume, there was little additional error, however, systematic biases produced substantial error increases. The proportion of measurement variation error attributable to diameter at breast height and tree species classification was small relative to that attributable to bole (merchantable) height and percent cull attributes, which composed the preponderance of uncertainty due to measurement variation. The greatest impacts were associated with the accretion component, which was subject to measurement variation and bias at both the initial and subsequent measurements.