基于林分状态特征的森林经营试验样地设计新方法

传统田间试验设计方法如顺序设计、拉丁方设计以及随机区组设计方法在很大程度上考虑了林分立地、土壤或肥力等方面的差异,但没有关注立地之上的"试验对象即林分"的状态差异,加上其有限的随机过程和缺失量化的目标函数等缺陷,从而谈不上是优化的试验设计,无法保证试验开始前处理间林分状态特征的一致性。本文提出了一种全新的森林经营田间试验优化设计方法——基于林分状态特征的完全随机优化设计,并将其应用于甘肃小陇山锐齿栎天然林的森林经营田间试验设计中。基于林分状态特征的完全随机优化设计得到小区排列方案,其处理间林分状态特征参数的差异均小于5%,优于传统田间试验设计方法,能确保试验林分状态的一致性。利用R语言实现该优化设计并使其可以在计算机上自动得到最佳的小区排列方案,提高了森林经营田间试验的精度,为科学试验的开展奠定了坚实的理论基础。

A New Design of Field Trial for Forest Management Based on Stand Characteristics

The traditional designs of field trials for forest management, such as sequential design, Latin square design and RBD, consider mainly on the differences in soil and other site conditions, but seldom the conditions of the stand itself. This paper proposes a new design-completely random and optimization design, which removes away the limit on randomization and provides optimization using a quantitative objective function. The design has been applied to a Quercus aliena var. acuteserrata natural forest in Xiaolongshan, Gansu Province. The difference of stand characteristics among treatments in the new design was less than 5%. This indicates a better consistency of stand state characteristics than the traditional designs. R language was used to achieve the optimal design and automatically generated the best arrangement scheme by computer. The new design can improve the accuracy of field trial for forest management and provides some references for forest management trials.