Стратификовани узорак у инвентури шума / Stratified sampling in forest inventory

The paper analyzes the possibility and technique of using stratified sampling in forest inventory. The research was carried out in a management class of beech high forests. It was first stratified into stands (eleven in total), and then into homogenous groups of stands (three in total). The focus was on estimating the average and total management class volume. The anticipated effects of stratification were preliminarily checked by using the method of simple analysis of variance. It was concluded that there was a statistically significant difference in the values of the average volume per hectare between the stands. In other words, we can expect a significant effect of the stratification performed in esti- mating the management class volume. The result was confirmed with the homogeneous groups of stands as strata. Statistical data processing referred to the volume that was determined by establishing 500 m2 sample plots, systematically arranged in a 100 x 100 grid across the stands. At stand level, the estimate of the average and total volume was done by the simple random sampling formulas, while the simple and stratified random sampling formulas were used at the level of the management class. In this particular case, it was found that the stratified sample of the same size pro- vides a much more precise estimate of average and total management class volume than the simple sample. In the first case, the relative error of volume, with a 95% probability, is +/- 5.37% and in the second +/- 4.59%. To achieve the same level of precision of manage- ment class volume estimates obtained using a simple sample, we can use a stratified sample that is smaller by 27.2%. Instead of 242 sample plots in the simple sample, 176 sample plots would be enough in the stratified sample or 66 sample plots fewer than in the simple sample. Compared to the proportionate stratified sample, the optimal stratified sample provides an even greater reduction in the size of the applied simple sample (by 77 sample plots). However, this paper gives the preference to the stratified sample with proportionate allocation of sample plots (systematic arrangement) for the purposes of forest inventory of units larger than a stand. Simple systematic sample should be used for stands. However, we should not insist on achieving a high level of precision of volume estimates at stand level, but at the level of management classes and inventory units larger than stands, because if we use an economically justified size of a sample, we can achieve the necessary level of preci- sion (accuracy) and develop reliable management plans.