基于GIS和PCA降维的森林景观格局等级特征分析

森林景观空间格局及其内在等级特征辨识是森林资源管理和生态环境规划的重要基础。以新疆天山北坡西段霍城林场为对象,以森林资源二类调查资料和DEM数据为基础,基于GIS和PCA降维方法,通过景观等级分类和景观指数分析,对林区森林景观空间格局及其等级特征进行了研究。结果表明:天山北坡森林景观格局存在明显的等级特征,在不同等级上森林景观破碎化严重,斑块大小空间变异强烈;随着森林景观等级细化,森林核心斑块大小、破碎化、景观丰富度、空间邻近程度等指标变异加强;在较高等级水平上,斑块密度(PD)、平均斑块面积(AREA_MN)、景观形状指数(LSI)、周长-面积分形维数(PAFRAC)等指数对概括森林景观空间格局具有决定性。森林景观资源保护和精细化管理应注重对景观破碎化、丰富度、斑块空间邻接度等指标的考虑及调控,进而为林区景观格局优化、森林景观资源健康经营与管理提供有力参考。

Hierarchical Characteristics Analysis of Forest Landscape Pattern Based on GIS and PCA Dimension Reduction Method

Forest landscape spatial pattern and identification differences in their internal hierarchical characteristics are important basis for forest resource management and ecological environment planning. Huocheng forest farm, locating in the western section of north slope of Tianshan Mountain in Xinjiang, a place with relatively stable landscape diversity, was selected as typical study area, based on forest resource inventory data for management and digital elevation model (DEM) date, and with the support of geo information system (GIS), dimension reduction method of principal component analysis (PCA) and landscape pattern indices (LPIs), the spatial patterns and grading characteristics of forest landscape in the forest region were studied through landscape classification and landscape index analysis. The following results were obtained. In the first place, the forest landscape pattern had obvious hierarchical characteristics in the north slope of Tianshan Mountain. At different levels of organizational hierarchy, landscape was seriously fragmentized, and the patch size showed considerable spatial variability, in addition, with the refinement of forest landscape organizational hierarchy, spatial variability of some landscape indexes core patch area, fragmentation, landscape richness, adjacency were increased. In the second place, at a higher grade level, patches density (PD), average patch area (AREA_MN), landscape shape index (LSI) and perimeter area fractal dimension (PAFRAC) were key indexes for summarizing the spatial pattern of forest landscape. The research result can provide a strong reference for optimizing forest landscape pattern and healthy management of forest landscape resources in landscape level by using the dimensionality reduction method of landscape pattern information. Most noteworthy, the protection and refinement management of forest landscape resources should be fragmentation, richness and spatial adjacency of landscape.