As the method in the Future Land Use Simulation (FLUS) model, System Dynamics (SD) and Cellular Automata (CA) are used to explore the spatiotemporal change features of land use in Shenzhen City, a typical rapidly urbanized region. Future land use demand was obtained by using SD model based on historical data from 2004 to 2018 for the use of spatial prediction. Then, CA model was used to predict the spatial distribution of eight land types (cropland, scrub, forest, grassland, urban land, traffic land, water area, and other land) in 2025 based on the 2014 land use map and the dynamic change of land use was analyzed. The simulation accuracy of land use pattern containing multiple land types in different time periods was relatively reliable, indicating that the SD-CA coupling model could be used to simulate the spatial patterns of land use in regions undergoing drastic changes of multiple land types. It was estimated that the percentage decline of the area of cropland, scrub, forest, grassland, and water in 2025, would reach to 39.26%, 32.72%, 10.06%, 55.3%, and 16.93%, respectively compared with that in 2014. The proportion of construction land (urban land and traffic land) area would account for more than half. The distribution of natural land types would become more fragmented and discrete while construction land would be continuously expanded and internally connected by encroaching on the ecological land. The results could provide support for land planning and evaluation of ecological and environmental effects from land use change in Shenzhen City.

利用未来用地模拟(FLUS)模型预测土地利用变化, 探讨深圳市在快速城市化进程中土地利用格局的时空演变特征。首先, 基于2004—2018年的历史数据, 用系统动力学(SD)模型预测未来用地需求。然后, 使用元胞自动机(CA)模型, 基于2014年土地利用图, 预测2025年的耕地、园地、林地、草地、城镇用地、交通用地、水域和其他用地等8个地类的空间分布, 并分析土地利用的动态变化。对不同时段多地类土地利用空间格局的模拟结果进行精度评价, 得到相对可靠的精度, 说明SD-CA耦合模型可用于模拟复杂地类剧烈变化地区的土地利用空间格局。预测结果表明, 与2014年相比, 2025年耕地、园地、林地、草地和水域面积减幅将分别达到 39.26%, 32.72%, 10.06%, 55.3%和16.93%, 建设用地(城镇用地和交通用地)面积占比将超过50%。自然地类分布将更加破碎和离散, 建设用地将通过侵占生态用地持续扩张和连通。研究结果可为深圳市土地规划和土地利用变化的生态环境效应评价提供基础资料。