欢迎访问林业科学,今天是

林业科学 ›› 2017, Vol. 53 ›› Issue (1): 28-37.doi: 10.11707/j.1001-7488.20170104

• 论文与研究报告 • 上一篇    下一篇

森林空间结构分析中基于Voronoi图的样地边缘校正

刘帅, 张江, 李建军, 周国雄, 吴舒辞   

  1. 中南林业科技大学 长沙 410004
  • 收稿日期:2015-10-08 修回日期:2016-01-21 出版日期:2017-01-25 发布日期:2017-03-03
  • 通讯作者: 张江
  • 基金资助:
    国家自然科学基金项目(31570627);湖南省科技计划项目(2015WK3017);湖南省教育厅科研项目(13C1143)。

Edge Correction of Voronoi Diagram in Forest Spatial Structure Analysis

Liu Shuai, Zhang Jiang, Li Jianjun, Zhou Guoxiong, Wu Shuci   

  1. Central South University of Forestry and Technology Changsha 410004
  • Received:2015-10-08 Revised:2016-01-21 Online:2017-01-25 Published:2017-03-03

摘要: [目的] 提出一种新的样地边缘校正方法——Voronoi图结点距离判定,为森林空间结构分析和调控提供技术支撑。[方法] 为评估Voronoi图结点距离判定的性能,引入量化评估指标RMSE,采用角尺度、聚集指数和大小比数等空间结构指数构造RMSE,并通过设置不同场景的模拟样地和调查样地,将Voronoi图结点距离判定与其他校正方法进行比较和分析。[结果] 缓冲区法、最近邻体校正、Voronoi图结点距离判定虽同属减少样本类校正方法,但各校正方法的性能及效果受到林木样本量、林木空间格局、空间结构指数、样地情况等因素影响,差异很大。缓冲区法虽简单易行,但样本浪费严重,且同时存在漏判和误判2种误差,其校正样地的RMSE曲线通常维持在较高水平。Voronoi图结点距离判定可视为缓冲距离随边界林木实际分布自适应变化的校正方法,具备更强的校正能力,能最大限度地排除样地外林木的边缘干扰,在角尺度、大小比数以及样本量充足的聚集指数计算中均优于其他校正方法,但也存在误判及样本利用率偏低等不足。最近邻体校正的性能介于缓冲区法和Voronoi图结点距离判定之间,其RMSE曲线的整体水平要高于Voronoi图结点距离判定。[结论] 林木样本量与样地边缘校正密切相关,对于样地大小和林分密度的任何改变最终都体现在林木样本量的增减上,维持充足的林木样本对于样地校正尤为重要。林木空间分布是影响样地边缘校正的另一重要因素,当林木呈随机分布或均匀分布时,样地边缘校正的重要作用能得到更好体现。此外,样地经过校正后,各类空间结构指数的计算精度均能获得显著提高,且聚集指数的计算需要样地提供充足的林木样本量。

关键词: 森林空间结构, Voronoi图, 边缘校正, 空间结构指数

Abstract: [Objective] In order to provide technical support for the analysis and control of forest spatial structure, this paper put forward a novel edge correction method, that is, distance determinant of Voronoi nodal.[Method] To assess the performance of distance determinant of Voronoi nodal, this study compares and analyzes this new method with other edge correction methods by firstly introducing the quantitative evaluation index RMSE, then applying uniform angle index, aggregation index and other spatial structural index to form RMSE, and finally setting up different simulation plots and survey plots. [Result] Although buffer zone, nearest-neighbor correction, distance determinant of Voronoi nodal are all called minus-sampling correction methods, their performances and effects are affected by several factors, such as sample size, tree spatial pattern, forest spatial structure indices, plot conditions, etc. buffer zone is simple and easy to operate, but its samples are seriously wasted and there also exist leakage and errors, which leads to a usual high level of RMSE curve. Distance determinant of Voronoi nodal can be seen as a correction method which adaptively changes with the actual border forest distribution. It has greater ability to correct and can maximize the elimination of edge effects in plots, thus which is better than the other edge correction methods when calculating uniform angle index, size ratio and aggregation index with sufficient samples. However, there exist some shortcomings such as error determinant, low sample utilization ratio, etc. The performance of nearest-neighbor correction stands fell the middle of the above-mentioned two methods, and its overall level of RMSE was higher than that of distance determinant of Voronoi nodal. [Conclusion] Sample size is closely related to edge correction methods, any changes in the plot size or forest density are ultimately reflected as the increase or decrease of sample size. Therefore, maintaining adequate samples for edge correction is particularly important. Spatial distribution is another important factor which would affect edge correction. When the spatial pattern is random or uniform, the important role of edge correction can be better reflected. In addition, after edge correction, the calculation accuracy of all kinds of spatial structure indices can be improved significantly and adequate forest sample sizes are needed to calculate the aggregation index.

Key words: forest spatial structure, Voronoi diagram, edge correction, spatial structure index

中图分类号: