基于LSMM与MSPA的深圳市绿色景观连通性研究

基于线性光谱混合模型(LSMM,Linear Spectral Mixture Model),引入形态学空间格局分析(MSPA,Morphological Spatial Pattern Analysis)进行城市地域绿色景观连通性评价。根据城市绿色景观特点和MSPA方法中的7种连通性类型的涵义,定义了城市绿色景观连通性功能类型。以深圳市1986年、1995年、2000年、2005年及2010年五期Landsat TM影像为数据源,应用线性光谱混合模型提取植被覆盖率,得到深圳市植被覆盖图。在此基础上,提取出高、全植被覆盖作为目标像元进行MSPA处理,分析植被覆盖状况与绿色景观功能类型的时序总体特征及空间梯度动态。结果表明:深圳市绿色景观破碎程度较高,表现为对结构连通性贡献最小的斑块类型总数最大。城市内部东西部连通性呈现出不同变化的趋势;右侧外圈层的大鹏半岛结构连通性最佳;在同一城市化发展梯度上,东部的样带连通性水平比西部要好。在城市化过程中,深圳市高、全覆被植被像元连通性大小受以下因素的影响:城市化程度,地形因素及区域定位。在同一城市化程度上,地形因素对景观连通性的影响较大。从整体的时间变化和空间梯度动态分析可知,在快速城市化过程中植被覆盖率和连通性功能均下降,而到稳定城市化阶段植被覆盖率和连通性均得到改善。研究表明线性光谱混合模型与形态学空间格局分析相结合可以较好的表征城市绿色景观连通性类型时空分布特征,进而明晰城市化过程与区域内绿色景观数量及连通性动态变化关系。

Landscape connectivity dynamics of urban green landscape based on morphological spatial pattern analysis (MSPA) and linear spectral mixture model (LSMM) in Shenzhen

Urban area is the main environment where human are living. The stability of internal ecosystem of a city is highly relevant with its sustainable development. Besides, the connectivity of urban green landscapes is a symbol of the integrity and stability of regional ecological functions. Urbanization has brought a dramatic transformation to urban landscape connectivity. The research on the dynamic changes of landscape connectivity is not only significant to the stability of an urban ecosystem, but also provides a basis for regional biodiversity conservation, urban planning, and land use planning and management. However, current connectivity indicators have obvious limitations, for example, indices over different landscape patterns may show familiar values; graph theory requires a human interpretation because of redundancy data, and research on large-scale landscape may cause data extinction during processing. In this paper, Linear Spectral Mixture Model (LSMM) was integrated into Morphological Spatial Pattern Analysis (MSPA) to evaluate the spatial and temporal dynamics of green landscape connectivity in Shenzhen. According to the urban landscape characteristics and MSPA theory, 7 types of connectivity were defined for urban green landscapes, and then the change features among different urbanization gradients were analyzed. We defined urbanization gradients as three different circle layers based on urbanization density, with a decrease of dense values from the first circle to the third. The main steps followed: 1) LSMM was applied to extract the vegetation coverage information from multi-temporal Landsat TM images. On that, the high and full covered vegetation pixels were defined as the foreground pixels (green landscape) in MSPA approach. 2) 7 types of connectivity were utilized to reveal the temporal and spatial variations of green landscapes in the process of urbanization. The results demonstrated that: 1) over 27% of green landscapes in Shenzhen did not contribute to connectivity during 24a. 2) The transition matrix of connectivity-pattern categories from 1986 to 2010 indicated that the connectivity areas were sharply fluctuated during 24a, and the majority of classes changed into non-green landscapes. Except for the core category, the areas of other connectivity categories showed an upward trend. The connectivity of internal urban landscapes showed different trends between eastern part and western part, and, the Dapeng Peninsula in the third circle showed the best connectivity among the whole city. The peak interval of connected categories showed that the eastern part of Shenzhen had more connectivity providers than the western part. 3) The overall connectivity of Shenzhen''s green landscapes followed a change of "decrease-increase" in sequence. Comparing with Shenzhen''s urbanization process, it is proved that the quantity and connectivity of green landscapes were affected by the following factors: urbanization level, topographic factor and regional policy. Additionally, it is found that the topographic factor had the greatest influence within the same urbanization level. The results from temporal variations and spatial gradients demonstrated that both vegetation coverage and connectivity showed a downward trend in rapid urbanization process, while the two have been improved in steady urbanization stage. The experimental results prove that the jointly analytical framework is efficiently applied to reveal the spatial and temporal dynamics of connectivity characteristics for urban green landscapes during the process of urbanization. Furthermore it enables us to know the relationship between urbanization and urban green landscape connectivity. This research can be applied in practice and provide benefits for monitoring urban green landscapes.