Abstract Sampling is a widespread practice used to generate estimates of common forest parameters. Growth and yield models use these estimates to generate projections necessary for timber management decisions. Informed by a case study in Virginia, USA, we demonstrate through simulation that an optimal amount of information exists for minimizing the difference between the true net present value (NPV) without incurring cost for the information and the estimated NPV obtained through investing monetary resources into sampling. In a 76. 9 ha, 21-year-old loblolly pine (Pinus taeda) stand, 265, 135 m^2 m 2 sample units minimized this difference. Besides the smallest sample unit evaluated (40 m^2 m 2), a consistent 3–4% optimal allowable error was found. Optimal harvest age was on average, within 0. 5 years for sample sizes greater than 25. Simulation assumptions noticeably impacted the results, but all the commonly used sample intensities evaluated were less than optimal across a sensitivity analysis of basal area means and standard deviations. This study demonstrates that information gained from sampling can be viewed as an investment to optimize correct land valuation rather than simply a necessary management cost.
Green et al. (Thu,) studied this question.