论文部分内容阅读
传统的林分生长与收获模型不能为林分和生态系统管理提供准确的空间信息。采用3种配对势函数的Gibbs点过程模型对美国东北部50块云冷杉样地里树木的空间分布模式进行模拟。该Gibbs模型能够较好地模拟这50块样地中的82%～84%,但其对完全随机分布和规则分布的样地模拟比对聚集分布的样地模拟效果要好。使用常用的林分变量如林分密度、公顷断面积、林分平均胸径、平均树高、平均冠幅和冠长建立经验回归模型对Gibbs模型的2个参数进行预测。结果表明这些回归模型对81%的样地可以得到满意的模拟效果,其中,100%的完全随机分布样地、71%的规则分布样地和56%的聚集分布样地模拟效果较好。选择3块样地对树木的模拟空间位置和实际观测位置的相似性进行对比和说明。
Traditional stand growth and harvest models do not provide accurate spatial information for stand and ecosystem management. The Gibbs point process model with three kinds of pairing potential functions was used to simulate the spatial distribution patterns of the trees in 50 plots of fir-tree in the northeastern United States. The Gibbs model can well simulate 82% ~ 84% of the 50 plots, but it is better for plots of plots with completely random distribution and regular distribution than those with aggregated distribution. Two parameters of Gibbs model were predicted using commonly used stand variables such as stand density, hectare area, mean diameter at stand, mean tree height, average crown length and crown length. The results show that these regression models can be satisfactorily simulated in 81% of the plots, of which 100% of the completely random plots, 71% of the regular plots and 56% of the aggregated plots have good simulation results. Three plots were selected to compare and illustrate the similarities between simulated spatial locations and actual observed locations.