欢迎访问林业科学,今天是

林业科学 ›› 2021, Vol. 57 ›› Issue (8): 1-12.doi: 10.11707/j.1001-7488.20210801

• 论文与研究报告 • 上一篇    下一篇

基于敏感度排序的风景林色彩格局指数筛选

曹瑜娟,徐程扬*,任雅雪,李夏榕   

  1. 北京林业大学城市林业研究中心 干旱半干旱地区森林培育及生态系统研究国家林业和草原局重点实验室 森林培育与保护教育部重点实验室 北京 100083
  • 收稿日期:2020-03-05 出版日期:2021-08-25 发布日期:2021-09-30
  • 通讯作者: 徐程扬
  • 基金资助:
    北京市朝阳区平原生态林定向抚育关键技术集成与示范(CYSF-1904)

Selection of Color Pattern Indices of Scenic Forest Based on Sensitivity Ranks

Yujuan Cao,Chengyang Xu*,Yaxue Ren,Xiarong Li   

  1. Research Center for Urban Forestry of Beijing Forestry University Key Laboratory for Silviculture and Forest Ecosystem Research in Arid and Semi-Arid Region of National Forestry and Grassland Administration Key Laboratory for Silviculture and Conservation of Ministry of Education Beijing 100083
  • Received:2020-03-05 Online:2021-08-25 Published:2021-09-30
  • Contact: Chengyang Xu

摘要:

目的: 研究景观格局指数对风景林色彩格局变化的敏感度,筛选出能够客观定量描述景观色彩格局特征的指数,以期为科学有效地分析彩色风景林视觉质量提供试验支撑与理论依据。方法: 以黄栌风景林秋景照片为研究数据,利用Python编程进行色彩量化和色彩分类,实现照片色彩批量预处理。通过Fragstats软件计算初选指数值并用SPSS软件完成层次聚类分析。计算各指数对由色彩分类、观景距离引起的景观色彩格局变化的敏感度。对敏感度综合排序靠前的指数进行独立性检验和含义重复性排查,在兼顾指数含义、尽量涵盖各大聚类、减少信息冗余的原则下完成指数筛选。结果: 结合指数含义和类型,从24个初选指数中筛选出对色彩分类和观景距离敏感度较高、线性不相关、含义无重叠的7个指数,即斑块数量、最大斑块指数、平均斑块面积、斑块丰富度、平均最邻近距离、平均周长面积比和修正Simpson’s多样性指数,用于定量表征风景林色彩格局特征。结论: 人工干预下的自动化色彩分类程序,简化了从色彩分类到指数计算的操作步骤,缩短了数据预处理用时,提高了照片处理效率,增强了结果稳定性。基于敏感度排序的色彩格局指数筛选方法,结合独立性检验和含义重复性排查,减小了主观选定指数可能引起的误差,该方法选取的指数能够有效区分因色彩分类和观景距离不同导致的景观色彩格局的变化,为客观定量分析风景林景观视觉质量提供了可能。

关键词: 景观色彩格局, 敏感度, 风景林, 黄栌, 视觉质量

Abstract:

Objective: The sensitivity of landscape pattern indices to color patterns of scenic forests was studied. And indices capable of characterizing landscape color patterns were selected, aiming to provide experimental supports and theoretical basis for studies of visual quality of colorful landscape forests. Method: Photos of scenic forest dominated by Cotinus coggygria var. cinerea taken in autumn were used in the research. An application for color quantification and classification was implemented using Python program, effective for data preparation in batch. Pre-selected indices were calculated in Fragstats software, and then hierarchical cluster analysis was performed in SPSS software. Sensitivity of indices to landscape color pattern changes related to color classification and viewing distance was calculated and used for indices ranking. Under the principles of covering most clusters and reducing information redundancy, indices were selected after performing independence tests and information overlaps check. Result: Seven indices, namely, number of patches, largest patch index, mean patch area, patch richness, mean nearest neighbor distance, mean perimeter/area ratio and modified Simpson's diversity index, which are sensitive to color classification and viewing distance, were selected from 24 pre-selected ones. They could be used to quantitatively characterize color patterns of scenic forests. Conclusion: The semi-automated procedures for color classification, which requires operator intervention during analytic process, simplify operations from color classification to index calculation, reduces time for data preprocessing and enhances the stability of the result. The proposed method for selecting color pattern indices based on sensitivity ranks, combined with independence test and information overlaps check, reduces errors caused by subjective selectior indices. And indices selected are effective to distinguish changes of color pattern caused by color classification and viewing distance, and possible to objectively and quantitatively analyze visual quality of scenic forest.

Key words: landscape color pattern, sensitivity, scenic forest, Cotinus coggygria var. cinerea, visual quality

中图分类号: