Variance estimation from systematic samples in Minnesota timber stands
A 402 permanent-plot inventory of the University of Minnesota Cloquet Forestry Center was used to assess the adequacy of systematic samples for estimating population variance. The inventory plots were arranged in the form of four approximately equal-size systematic samples or clusters. The design was systematic sampling with multiple random starts. Population variances were estimated for number of trees, basal area, and volume per hectare for four different measurements spanning 17 years. Results indicate that the individual random starts and the aggregate of 402 plots treated as a simple random sample provide estimates of variance comparable to those obtained by treating the inventory as a cluster sample design. This report plus reports in the literature suggest that plots in the Lake States that are at least 80 to 362 m apart are likely to provide useful estimates of population variances and sampling errors for common forest-survey variables.