建立在固定样地上的林分生长线性联立方程组模型研究

Linear simultaneous equations models of stand growth based on the permanent sample plots.

为描述森林资源连续清查中两期固定样地因子之间的关系,该文分析了两期林分生长变量间的关系,以固定样地的两期(1990和1994年)观测变量为资料,在Borders模型基础上,引入标准地位指数和变量的线性变换,分优势树种构造出间隔期为5年的林分生长线性联立方程组模型簇YB+xΓ=ε,并获得该模型待估参数矩阵B、Γ的无偏估计.该模型是可识别且误差结构矩阵不是对角矩阵的联立方程组模型.它表述了样地期末断面积和蓄积是林分标准地位指数、年龄、期初断面积的隐函数,较深入地描述了两期固定样地林分因子相互依存、从属的关系.通过模型的外检验,蓄积平均精度可达90%以上,林分纯生长量的估计精度可达80%以上.该模型具有较好的适应性,并可在今后若干个离散的时间点上预估同类林分的收获量,为研究在固定样地上两期林分自然生长规律提供较科学的理论和方法.

To describe the factor relations of the two periods permanent sample plots （PSP） in continuous forestry survey, the paper analyzed the stand growth variables based on the observation materials in the years of 1990 and 1994. And then it introduced standard site index and linear transform according to the Borders model. Stand growth simultaneous equations model cluster（ YB ＋ xГ = ε ） was constructed based on the dominant tree species after an interval of five years. The model parameter matrices （ B and Г） were un-bias estimation. The simultaneous models were recognizable and their error structure matrices were not diagonal. The equations of stand basal area and volume in 1994 was an implicated function of the standard site index, stand age and initial basal area. The paper described the subordination and dependent relationship between the two periods PSP stand factors. Through the model external-test, the average precision of volume model was over 90 % and the estimation precision of stand net growth was over 80 %. The simultaneous models were well adaptive to estimate discrete data of stand yield and could describe the natural stand growth theoretically.