摘要
以上海市陆域范围为研究区域,采用2004年Landsat TM图像解译的景观图为主要信息源,从空间粒度方面探讨了20种景观生态指标对粒度的敏感性及其随粒度变化的规律;利用Moran's I指数对不同粒度下景观生态格局的空间自相关性进行分析。在此基础上,利用半变异函数探讨景观生态指标的空间异质性与幅度的效应关系。研究表明:①在斑块类型层次,除相对丰度(PLAND)、最大斑块指数(LPI)和景观分离度指数(DIVISION)粒度效应不明显外,其他均表现出较明显的粒度效应;部分指数或某些斑块类型的指数出现明显的尺度转折点,研究区域内农田和建筑用地等优势斑块,其平均斑块面积(MPS)和面积加权平均形状指数(AWMSI)大致分别在180m和120m附近出现尺度转折点;②在景观层次,各景观生态指数都表现出明显的粒度效应,部分指标在150m,120m至180m出现尺度拐点;③上海市陆域范围的景观格局存在显著空间正自相关性,150m的粒度水平为景观空间自相关性对尺度响应的一个敏感点;④随着幅度的增加,Shannon多样性指数(SHDI)和面积加权平均分维数(AWMFD)的块金效应逐渐增强,空间自相关性引起的空间差异对总体变异的贡献逐渐减小;对于景观聚集度指数(CONT)和分离度指数(SPILT),其块金效应的尺度变化较为复杂,分别在1km和2km幅度下,块金效应最强;⑤随着幅度的增加,对于SHDI和CONT,空间自相关的存在范围增大,而对于AWMFD,则略有减小;对于SPILT,空间依赖性的存在范围在幅度为2km时最大;⑥半变异函数模型的拟合结果表明,对于SHDI,CONT和SPILT,2km对于研究上海市陆域范围景观格局的空间变化是一个比较合适的尺度。
Taking Shanghai as a research region, and based on the landscape map of interpreted from I.andsat TM images of 2004, this article discusses 20 ecological metrics sensitivity and change rules from the aspect of spatial grains, and analyzes the spatial auto--correlation of landscape pattern at different grains by using Moran's I. And then it measures the relationship between the spatial heterogeneity of ecological metric and spatial extent by means of Semivariogram. It can be concluded that: @At clot types level, excluding PLAND, LPI and DIVISION, the spatial grain effect of landscape metrics is significant; partial index or some clot typeg index have significant scaling turning point, the preponderant patches such as cropland and built--up area, the scaling turning points of MPS and AWMSI are about 180m and 120m respectively; @At landscape level, all the selected index show the significant grain effect, and some index have the turning points to the grain at the level of 150m or 120m to 180m; @The urban landscape structure of Shanghai land area show the spatial positive autocorrelation, the grain level of 150m is the sensitive point to scaling of spatial autocorrelation; @With the extent increasing, the nugget effects of SHDI and AWMFD increase. The spatial heterogeneity, which is caused by spatial autocorrelation, contributes less to the total spatial heterogeneity of research area; to the nugget effects of CONT and SPILT to scaling are complicated, the values of Co/ Co+ C reach their maximums at the extent level of lkm and 2km respectively, and the nugget effects are largest; @With the extent increasing, the ranges of spatial autocorrelation for SHDI and C ONT increase, while minish a little for AWMFD; for SPLIT, with the extent increasing, the range increases first and minish a little later, and the range of spatial dependence reach their maximum at the extent of 2km; @The fitting result of Semivariogram indicates that for SHDI, CONT and SPILT, the appropriate extent level to study the landscape pattern variance of Shanghai land area is 2km.
出处
《绿色科技》
2013年第8期1-9,共9页
Journal of Green Science and Technology