Tree density and stand structure estimation by stratifiedcluster and systematic random sampling methods

We assessed the accuracy of two sampling methods by comparing their estimations to Full Calipering (FC). The two methods included 1000-m2 systematic random (SR) sampling plots designed within a 150 ×200 m grid, and acluster sampling (CS) method with 7500-m2 clusters each embracing 4 plots of 1875 m2 each. Full calipering was performed in 4 parcels of 116 ha each. The inventory resulted in total 40239 trees measured in 40 SR plots (sampling intensity=9.7 %) as well as 56 plots within 14 clusters of in CS approach (sampling intensity=3.44 %). Results from CS method were associated with optimum accuracy for tree density and forest structure. The CS method returned results closer to the FC when estimating the beech- and hornbeam-specific no. of trees per ha and their distribution in diameter classes and in parcels. The Maximum trees diameter of 165, 110 and 125 cm were recorded for FC, SR and CS approaches, respectively. Conclusively, this study suggests a forest type and species-specific stratification to be performed prior to each parcel-based inventory. In addition, sample size should be determined on a forest type-specific basis, ideally distributed within a CS design with 1900 to 2000-m2 plot sizes.