土地利用分类对景观格局指数的影响

Research on the Influence of Land Use Classification on Landscape Metrics

基于景观格局指数的空间格局分析是当前景观生态学研究的重要基础内容,不仅数据源准确度、尺度效应显著影响景观格局指数,土地利用类型划分也对景观格局指数具有显著影响,但我们对这种影响的总体理解尚很缺乏。本研究选取24种常用景观格局指数,以深圳市宝安区为试验区,探讨景观格局指数随土地利用分类系统变化的基本规律。研究结果表明,土地利用分类对景观格局指数的确具有显著影响。而根据景观格局指数对土地利用类型数目变化响应的可预测性,可将其分为三类:①第一类指数随土地利用分类系统变化的可预测性强,能用简单函数关系(对数函数关系、S形曲线、反比曲线关系)来表达,包括斑块数目、斑块密度、边界密度、平均斑块面积、景观形状指数、平均斑块形状指数、周长面积比分维数、平均斑块分维数、聚合度、Shannon多样性指数、Simpson多样性指数和修改Simpson多样性指数;②第二类指数随土地利用分类系统变化可预测性较差,表现为典型的分段(或阶梯形)变化,存在多种可能(S形曲线、直线、反比曲线与复合曲线关系),包括斑块面积标准差、斑块面积变异系数、最大斑块指数、面积加权平均斑块形状指数、面积加权平均斑块分维数、分离度和斑块结合度;③第三类指数由于在度量相关空间格局特征时考虑了土地利用类型数多少的影响,随土地利用分类系统变化呈无规律变化,难以用一种简单函数或分段函数来预测其变化行为,包括蔓延度、Shannon均匀度指数、Simpson均匀度指数、修改Simpson均匀度指数和优势度指数。这些指数随土地利用分类系统变化的变化规律,使试验区不同时期、不同土地利用分类系统下的空间格局比较成为可能。

Landscape pattern analysis based on landscape metrics is a basic content of the research on landscape ecology. More and more researches proved that not only scale effects and the precision of remote sensed data had significant influence on landscape metrics, but also the difference of land use classification would make the change of landscape metrics. However, we still have not found out how land use classification affects landscape metrics and associated influence mechanism. In this paper, we chose Bao＇an of Shenzhen city as an experimental area, to analyze the characteristics of the change of 24 landscape metrics associated with the change of land use classification. The results showed that land use classification indeed influenced landscape metrics. And based on the shape of the land use classification effect curves and the predictability of these relations, the 24 landscape metrics can be divided into three groups. The first group included 12 indices, i.e., number of patches （NP）, patch density （PD）, edge density （ED）, mean patch size （MPS）, landscape shape index （LSI）, mean patch shape index （MSI）, perimeter-area fractal dimension （PAFRAC）, mean patch fractal dimension （MPFD）, aggregation index （AI）, Shannon＇s diversity index （SHDI）, Simpson＇s diversity index （SIDI）, and modified Simpson＇s diversity index （MSIDI）. The behavior of this group of indices with the change of the number of land use types was very predictable with simple function relations in regression analysis, which were mainly logarithm function, S function, and inverse function. The second group included seven indices, i.e., patch size standard deviation （PSSD）, patch size coefficient of variation （PSCV）, largest patch index （LPI）, area-weighted mean patch shape index （AWMSI）, area-weighted mean patch fractal dimension （AWMPFD）, landscape division index （DIVISION）, and patch cohesion index （COHESION）. The behavior of this group was not easy to predict with significant subsection. And function relations used in regression analysis mainly included S function, linear function, inverse function and compound function. The third group included five indices, i.e., contagion index （CONT）, landscape dominance index （DI）, Shannon＇s evenness index （SHEI）, Simpson＇s evenness index （SIEI）, and modified Simpson＇s evenness index （MSIEI）. The behavior of this group could not be predicted. Significant influence of the changing land use classification on landscape metrics indicated that only landscape with the same land use classification could be used for comparing landscape pattern characteristics.