Constraining Individual Tree Diameter Increment and Survival Models for Loblolly Pine Plantations
Constraining Individual Tree Diameter Increment and Survival Models for Loblolly Pine Plantations
复制标题
DOI：
10.1093/forestscience/43.3.414
复制
发表时间：
1997-08
期刊：
影响因子：
1.4
通讯作者：
Shaoang Zhang;R. Amateis;H E Burkhart
中科院分区：
文献类型：
--
作者：
Shaoang Zhang;R. Amateis;H E Burkhart
在现有的森林生长和产量模型中，单木生长模型提供了最详细的信息，这对集约化森林经营是有用的。然而，由于单木预测的累积误差，单木生长模型对于林分水平的预测可能不准确。本文的目的是开发一种方法，以确保单木生长模型不仅能提供准确的单木预测，而且能给出准确的林分水平估计。基于各径级断面积与同一径级内树木直径生长量之间的概念关系，提出了一个约束回归方程体系。同样的方法也被应用于个体树木存活概率的建模。提出了一种估计多响应回归的算法。结果用火炬松(Pinus taeda L.)人工林试验表明，联立方程系统提供的单株直径生长量与无约束模型几乎相同，但对每个径级的断面积和林分总断面积的估计值更好。在数据中几乎没有死亡率的情况下，使用受限的单株树木生存模型来改善林分死亡率估计的潜力是有限的。为。SCI。43(3)：414-423。
Among extant forest growth and yield models, individual tree growth models provide the most detailed information, which is useful for intensive forest management. However, individual tree growth models may be inaccurate for stand-level predictions because of the cumulative error resulting from individual tree predictions. This paper is aimed at developing a method to ensure that individual tree growth models not only provide precise individual tree predictions but also give precise stand-level estimation. Based on the conceptual relationships between basal area of each diameter class and the diameter increment of trees in the same diameter class, a constrained regression equation system is proposed. The same approach is also applied to modeling individual tree survival probability. An algorithm for estimating multiresponse regressions was developed. Results using data from unthinned loblolly pine (Pinus taeda L.) plantations show that the simultaneous equation system provides nearly the same individual tree diameter increment estimates as the unconstrained model but better estimates for basal area of each diameter class and for total stand basal area. The potential for improving stand mortality estimation with a constrained individual tree survival model where there is little mortality in the data is limited. For. Sci. 43(3):414-423.