Comparison of three sampling schemes for estimating frequency and D2H by diameter class - a simulation study

Abstract

Fixed-area plot sampling, point sampling and line sampling were compared in estimating the number of trees ( N), average D 2 H ( D 2 H), and sum of D 2 H values ( D 2 H T) by diameter class. In terms of efficiency, point sampling was easily the most efficient for all but the two smallest diameter classes for all three parameters, both considering costs and the same sampling intensity. Line sampling was second in efficiency for all three parameters for all but the four smallest diameter classes. Clearly, the efficiency gains of point sampling and line sampling over fixed-area plot sampling were due to the larger sampling intensity of the larger, much less frequently occurring diameter classes. There was little difference between the three sample selection methods in estimation bias. Classical variance estimates were slightly but consistently more reliable for estimating the precision of estimating N and D 2 H T by diameter class than the bootstrap variance estimator. The reliability of estimating the precision for both the classical and bootstrap variance estimates decreased considerably from the smaller to the larger diameter classes. This is the result of the small number of trees in these classes. Only a bootstrap variance estimator was available for estimating D 2 H by diameter class. This was of somewhat lower reliability than those for N and D 2 H T and spectacularly unreliable in some diameter classes with few trees.