EVALUATING DESIGN UNBIASEDNESS OF THE PRE-FELLING INVENTORY IN PENINSULAR MALAYSIA

The forests in Peninsular Malaysia are sustainably managed under the Selective Management System, which requires completing a pre-felling (Pre-F) inventory prior to harvest. The goal of this study was to investigate design unbiasedness of the Pre-F inventory using theory and simulation models. Simulations were carried out with an artificial and an actual forest, the former representing a cyclic forest of mature and young trees and the latter being the 50-ha Pasoh census plot. Theoretically, the Pre-F inventory is biased in mean estimates because it consistently undersamples portions of a population. Simulation with the cyclic forest showed that the Pre-F inventory was biased in estimating mean stand density, basal area and volume when the inventory was set up aligned with the cyclic pattern. Results from the Pasoh plot showed that the Pre-F inventory was unbiased in mean stand variables. The contrast was likely due to seemingly random tree distribution in Pasoh. However, using a design unbiased sampling method is more desirable because underlying tree distribution pattern in a forest is normally unknown. We recommend the Random Point Start method, which is unbiased in all simulation trials in addition to retaining all the advantages of the current Pre-F protocol.