景观格局指数间相关关系对植被覆盖度等级分类数的响应

运用景观指数进行景观格局分析已成为景观生态学的主要研究方法。但是,由于景观本身的复杂性以及景观格局指数之间的相关性,其应用与发展受到制约。选用面积/周长/密度、形状和蔓延度等3类共28个指数,以广州市中心城区为研究区域,以TM遥感影像为数据源形成植被覆盖度等级图,在景观类型水平上探讨各类型内指数之间的相关性及其对景观分类数的敏感程度。相关性分析表明:景观格局指数间都存在一定的相关性,但是相关程度差异较大,其中面积/周长/密度类指数间的相关性最强,蔓延度类指数其次,形状类指数间独立性相对较强。敏感性分析结果显示:117个指数对的相关关系都随景观分类数发生变化,根据其变化类型和程度,可以分为简单型、分段型、复杂型等3种指数关系响应的类型,分别包括12、31和74对指数;敏感曲线还表明4或5是指数对之间相关关系最敏感的分类数;不仅如此,分类数对景观格局指数之间相关关系的影响程度因景观格局指数所属的类别而异,面积/周长/密度类型指数不敏感,其次是蔓延度类指数,最敏感的是形状类指数;最后,研究结果表明景观格局指数间的相关性对分类数的敏感性存在较大的空间变异。

Response to classification numbers of vegetation types on correlative coefficients among landscape metrics

It is important to analyze characteristics of landscape pattern indices for landscape ecological research, but its development and application in the field has been limited because of complicated relationships among the indices. There are more than 100 indices used to describe characteristics of landscape system. In this research 28 indices were analyzed, which were defined in three types of landscape indices (area/perimeter/density, shape, and contagion). The basic characteristics of a regional landscape were based on maps of vegetation fraction grading, which were translated from TM remote sensing images of Guangzhou on Oct. 22, 2005. This research was performed in the city center (8 districts) of Guangzhou. All landscape indices were calculated with the software Fragstats3.3. At the level of class metrics, correlative relationships among indices, which were in the same type of classes, were explored firstly. Then, the response of the correlative relationships was studied under different classification systems of vegetation fraction for the same landscape, and seven kinds of classification systems of vegetation fraction were designed, namely 2, 3, 4, 5, 6, 9 and 17. The results of correlative relationships were as following: relationship between every index pair was obvious for almost all indices, but there were great difference among different index pairs. For the type of area/perimeter/density metrics the relationships were most significant, and for the type of shape metrics indices were strong independent from each other. Some indices, such as PD,have their different ecological meaning at the landscape level and at the class metrics level. This research got a conclusion, which was different from literatures, by calculating correlative relationship between PD and AREA_MN. It revealed that it should be important to choose suitable indices for the goal of research at the class level and the landscape level. The results on sensitivity of those indices to number of types were analyzed as following: most of indices were very sensitive to classification number of landscape types, and the response degree and tendency were different among different metrics. The response tendencies of 117 index pairs were divided into three groups according to their action under different numbers of landscape types, such as non-sensitiveness, monotonous increasing and complex, and the number of index pairs of each group was 12, 31, and 74, respectively. The response to the number of landscape types indicated that classification numbers 4 and 5 were most sensitive. Another result showed that response difference stood among different types of class metrics, and indices about Area/Perimeter/Density metrics was non-sensitive with changing of classification numbers of types. The metrics about shape of landscape was the most sensitive. Additionally, the results revealed that the correlative relationship of indices were related to spatial characteristics of position.