摘要
在景观格局优化方法的发展过程中,以景观模拟演化为核心的景观格局优化模式已显示出优越性,这种模式在大量景观变化模拟预测结果形成的状态空间中搜索景观格局的优化方案,优化的客观性和自动化程度较高。但由于目前在景观尺度上的格局和过程相互作用规律尚缺乏定量的描述,限制了这种优化模式的发展。而基于元胞自动机的空间直观模型,由于元胞自动机固有的特点,在解决一系列技术难点的前提下,有望解决这一模式在目前发展中受到的限制。本文介绍了元胞自动机的原理和在景观生态学中的已有应用,认为元胞自动机获得广泛应用的原因在于,它有且仅有以下3个最本质的特征:状态的表达在空间上的离散性;状态变化的表达在时间上的离散性;状态转变的空间相关性。在此基础上,阐释了元胞自动机应用于景观格局演化的一般优势和相对优势,以及它在景观格局优化应用中存在的困难。针对这些困难,提出了部分的解决方案。最后对于以基于广义元胞自动机的空间直观模型为核心的景观格局优化模式,提出了一个可能的体系结构和流程图。
General treatment of landscape pattern spatial optimizations is in its infancy. Among various modes of landscape pattern optimization, the most advanced and sophisticated one is a mode that searches optimal solutions in state space of simulation results of landscape changes. It has exhibited many advantages such as relatively high objectivity and automaticity of optimization. However, at present, there are no enough quantitative theories of interactions between pattern and process in landscape for extensive use of this optimization mode. Fortunately, due to inherent advantages of cellular automata （ CA）, the spatial explicit model based on CA is likely to bring the optimization mode into operation, on condition that a series of technical problems are conquered.
In in paper, principles of cellular automata and its applications in landscape ecology were introduced. A CA has three essential characteristics： spatial discreteness of expressions of states, temporal discreteness of changes of states, spatial correlation of rules of state transformation. Therefore, it has not only general advantages for analysis and simulation of spatial-temporal development, but also relative advantages for spatial simulation of landscape changes. The reasons for both aspects of advantages in the applicability of CA into landscape pattern optimization were elucidated in the paper. The former is that a CA model is constructed ＂from bottom to top＂, based on interactions of microscopic individual units, so that it can simulate landscape changes directly at a relatively small scale, regardless of quantitative laws at landscape scale. The latter is that its definition of neighborhood and associated rules of transformation can naturally meet the requirement of landscape ecology that attach much importance to horizontal processes.
The challenges of application of CA into landscape pattern optimization were analyzed, among which the two most distinguished are as fellows ： one is contradictions between its simplicities of construction and complexities of landscape change. The other is how to define rules of transformation that can reflect natural and human factors during landscape change. Other problems in application of CA into landscape pattern optimization include definition of scales, calibration of temporal paces, and computational complexity.
Some partial and tentative solutions to these problems were presented. Firstly, the conception of CA model was expanded with most generalized expressions so as to enable CA to simulate complex landscape changes. Secondly, models of dominant ecological process were integrated into rules of state transformation. Last but not least, in order to reduce computational complexity and ensure the mode to be practically operated, some computer algorithms and techniques of software engineering, such as search strategies and calculation multiplexing, were utilized. An architecture and flow chart for the landscape pattern optimization mode, centering on generalized-CA based spatial explicit model, was proposed.
出处
《资源科学》
CSSCI
CSCD
北大核心
2007年第4期85-91,共7页
Resources Science
基金
国家"973"项目(编号:2002CB111506)
关键词
元胞自动机
景观格局优化
空间直观模型
景观模拟演化
Cellular automata （CA）
Landscape pattern optimization
Spatial explicit model： Landscape change simulation
