线性抽样及分形理论在景观异质性研究中的应用

Application of linear sampling and fractal theory in landscape heterogeneity

景观异质性是景观生态学研究的核心,对景观格局、动态和功能具有决定作用。将线性抽样与分形理论相结合的方法应用于景观空间异质性的表达中,相比于传统方法,可以定量表达景观类型在空间序列中的组合规律,而该组合规律可以作为指标来测度景观空间异质性的高低。景观类型在空间序列中的组合越复杂,则景观的空间异质性越高。该方法以景观类型空间序列的线形图像为基础,利用分形理论计算线形图像的分维,以分维数值的大小来表达景观空间异质性的高低。但样线的布设方法、分维计算方法以及景观分类系统的不同都会对分维数值产生影响。以妫水河流域为试验区,对该方法的适用性进行验证,结果表明,该方法可以有效表达景观空间异质性的高低。

Landscape heterogeneity plays a decisive role in landscape structure, function, and dynamics. It is one of universal characteristics of the natural world, and is in a core position in landscape ecology studies. So it plays a crucial role in understanding landscape dynamics, as well as complex process of ecology and its interactive mechanism. The common measuring methods for landscape heterogeneity include landscape index, geostatistical analysis, fractal theory, and linear sampling. Although there are many ways to quantitatively express landscape heterogeneity, the focuses of these methods are different, and these methods are poor universality. In general, the current studies mainly focus on understanding the expression of quantitative relationship between landscape types, as well as general statistical characteristics description of landscape structure difference. However, this paper thinks that landscape heterogeneity studies should focus on excavation of landscape types＇ spatial combination law, namely spatial sequence feature of landscape types. In this paper, the method of combining linear sampling and fractal theory is applied to express landscape spatial heterogeneity. Compared with traditional methods, this method can be used to express landscape types＇ spatial combination law in spatial sequence, and this combination law can be used as an index to measure the level of landscape spatial heterogeneity. The more complex the spatial combination rule of landscape types in spatial sequence is, the higher landscape spatial heterogeneity is. The Guishui fiver basin is taken as a test area and a Landsat TM image taken in August, 2009 is used as basic data to verify the applicability of this method. A landscape distribution map of the study area is gained by interpreting the TM image through visual interpretation and decision-tree classification method. Linear sampling method is used to extract surface landscape pattern information, and fractal theory is applied to calculate fractal dimension of landscape types＇ spatial sequence with linear sampling. In order to test this method＇s capacity to express discrepancy of landscape heterogeneity among different regions, the study area is divided into three sub-areas according to the differences of natural environmental characteristics. The results show that this method can effectively express the discrepancy of landscape spatial heterogeneity within the study area.