For operational forest management, a forest inventory is commonly conducted to determine the timber stocking and the value of standing trees in a stand. With time and costs constraints, appropriate sampling designs and models are required to perform the inventory efficiently, as well as to obtain reliable estimates for the variables needed to make projections. In this dissertation research, a simulation study was conducted to extensively explore four important topics in forest inventories: selection of measurement trees in point samples, projection from plot- and stand-level aggregations, subsampling height for volume estimation, and updating stand projections using periodic inventories. A series of simulated loblolly pine plantations with varying degrees of spatial heterogeneity were generated at different stages in stand development. Repeated sampling was used to examine various sampling schemes and growth projection methods. Highlights for the four topics follow:

1. Stand total volume can be reliably estimated using measurement trees tallied by Big BAF, point-double sampling, or random selection of a specified number of trees. However, number of trees per unit area in small-size classes were overestimated across the three tree-selection methods when sample data were aggregated into diameter classes.

2. Plot-level and stand-level projections produced similar estimates for dominant height, basal area, and stems per unit area. As spatial heterogeneity increased, stand-level projections indicated a significant bias of predicted total volume compared with the plot-level projections.

3. Sampling intensity, stand age and spatial heterogeneity have greater influence on the reliability for total volume estimation compared to subsampling intensity and measurement error for height measurements.

4.The variability of total volume estimates increases with increasing projection length (i.e., longer time intervals between inventory entry points). However, the estimates of stand total volume can be greatly improved by updating the models with information obtained in periodic forest inventories, especially when the original models are not well calibrated.

The results of this study provide useful guidance and insights for forest practitioners to design forest inventories and improve growth projection systems in operational forest management.