ABSTRACT

This study proposed an alternative approach that could be used to model irregular diameter distribution with fixed measurement threshold. Sixteen plots of 0.0625 ha size from a degraded teak (Tectona grandis L. f.) stands were used as case study, in which only trees with diameter ≥ 10 cm were considered. Four methodologies were evaluated: method 1 used the Weibull distribution neglecting the truncation point and irregularities; method 2 utilized truncated Weibull distribution; method 3 used two components 2-parameter Weibull in a mixture model, and in method 4, two components truncated Weibull in a mixture model was used. The models were assessed with different indices including Kolmogorov-Smirnov and Anderson-Darling statistics. Relative rank sum of the models was analyzed with one-way analysis of variance. The result showed that model developed with method 4 had the smallest fit indices and relative rank which was significantly different from models obtained with other methods. It also had the smallest root mean square error (5.146 m³ ha⁻¹) and critical error (1.362 m³ ha⁻¹) for estimating stand volume in diameter classes. Thus, incorporating truncated Weibull in a finite mixture is the best strategy for modeling diameter distribution of degraded teak stands with fixed measurement threshold.

摘要

本研究提出了一种替代方法，可用于模拟具有固定测量阈值的不规则直径分布。16 块 0.0625 公顷的退化柚木 ( Tectona grandis)L. f.) 林分用作案例研究，其中仅考虑直径 ≥ 10 cm 的树木。评估了四种方法：方法 1 使用 Weibull 分布，忽略截断点和不规则性；方法 2 使用截断的 Weibull 分布；方法 3 在混合模型中使用了两个分量 2 参数 Weibull，在方法 4 中，在混合模型中使用了两个分量截断 Weibull。使用不同的指标评估模型，包括 Kolmogorov-Smirnov 和 Anderson-Darling 统计。模型的相对秩和采用单因素方差分析。结果表明，用方法4建立的模型拟合指数最小，相对秩次最小，与用其他方法得到的模型有显着差异。它还具有最小的均方根误差（5.146 m³ ha ^-1 ) 和临界误差 (1.362 m ³ ha ^-1 ) 用于估算直径等级的林分体积。因此，在有限混合物中加入截断 Weibull 是模拟具有固定测量阈值的退化柚木林的直径分布的最佳策略。