海岸带土地利用模拟结果对元胞邻域的响应研究

通过元胞自动机（Cellular Automata，CA）动态模拟能够重建和预测复杂系统的全局结构及其演变。海岸带土地利用变化模拟是CA应用研究的重要领域之一。元胞邻域及其组合作为CA的必要组成部分，对海岸带模拟结果有显著影响。以半径为1，2和3的Moore邻域为例，测试并分析了模拟结果对元胞邻域的响应。研究表明:这种影响表现在3个方面:分别是元胞数量、空间形态和不确定性。小范围邻域组合模拟结果形态真实、不确定性较低，但是模拟变化容易过早终止；大范围邻域组合模拟结果形态真实性较差、不确定性较高，但是模拟结果能够保持延续。针对具体的模拟案例，元胞邻域的选择需要结合影响系统演变的全局因素而确定。模拟结果的响应特征不仅适合海岸带土地利用模拟，也适合其他类型的空间系统演变模拟。

Response to Neighbor Configurations on Coastal Land Use Simulation Results

By dynamic simulation of cellular automata (CA) models,global structures and their evolution of complex systems could be automatically reconstructed and projected.A neighbor configuration off cells is an essential part of a CA model which remarkably affects simulated results.Based on Moore neighbors with radius 1,2 and 3 respectively,the responses of simulations are detected and analyzed.This research demonstrates that the impact of cell neighbors includes three aspects: quantity of cells,spatial morphology and uncertainty.Small range of neighbors performs well in spatial morphology and uncertainty,but could prematurely terminate the simulating process; while larger range of neighbors performs poor in spatial morphology and uncertainty,but could continue the process over time.For an actual case study area,the selection of neighbor configurations needs global factors impacting system evolution fully considered.Moreover,the response characteristics on simulation of coastal land use changes presented in this research are also suitable for modelling the evolution of other spatial systems.