基于地质统计学影像纹理的草地植被群落空间结构分析

Analysis of different spatial structural characteristics of vegetation using geostatistical image model parameter texture

【目的】基于地质统计学影像纹理,定量分析草地植被群落空间结构的变异规律,为草地生态系统的恢复与重建提供依据。【方法】从科尔沁草原地区奈曼县57个退牧还草区域中遴选出25个有代表性的样地,应用地质统计学方法,计算10种典型植被群落的试验变异函数并拟合到球状模型,用变程和基台值来表达植被群落的空间结构特征,然后将从影像中计算得到的变异函数模型参数与样地植被群落空间结构特征对应起来,反演植被群落的空间结构,分析样地植被群落的空间分布规律。【结果】在分析的25个样地植被群落中,大多数种群的空间相关性和变异强度均较大;根据变异函数值、变程推断得到的样地优势物种与实际调查结果吻合率达到95%。【结论】应用地质统计学影像纹理研究草地植被群落空间结构特征的方法可行,该方法能较准确地从影像上直接推断优势物种并分析种群空间结构的变异规律,极大地减少了人工调查的投入。

【Objective】 The study was to provide evidence for grassland ecosystem restoration and reconstruction,and to analyze the law of spatial structure variation of grassland vegetation communities quantitatively,based on geostatistical image model parameter texture.【Method】 Taking Naiman County located in Horqin steppe region as research object,25 representative plots from the 57 regions of returning cropping land to forage land were selected.Then based on the difference of vegetation communities type on the spatial constructions,geostatistical image texture was used to calculate 10 typical vegetation communities experimental variogram and fitting spherical model,applying the range and to express spatial structural character of vegetation communities.To analyze the spatial distribution patterns of vegetation communities in the grassland plots quantitatively,we retrieved spatial structure of vegetation communities by corresponding them to the model parameters of variogram which were computed from remote sensing image.【Result】 The majority of the populations had strong spatial correlation and variation in the 25 grassland plots.By comparing with field observation,we found that the speculation precision of dominant species which were deduced according to the value of semivariogram and variational distance came up to 95%.【Conclusion】 The results mean that geostatistical image texture was effective in the analysis of spatial structure of vegetation communities,we could infer dominant species and analyze the law of spatial structure variation of grassland vegetation communities from geostatistical image texture,which would reduce investment for grassland investigation greatly.