基于地块对象的城市空间结构演化精细模拟

The mismatch of urban functions at the parcel level has caused the decreased efficiency and traffic congestions which further lead to serious economic loss and environmental problems. In this sense, effective use and optimization of urban parcels are important to fulfill the transition toward ‘the new kind of urbanization’. This demands the support from well-designed spatiotemporal simulation models for assisting urban planning and resources managements. However, existing models are applied to simulations at macro scales and inappropriate to be used in intra-urban scales with much finer details. Moreover, most of contemporary models adopt an ad hoc representation of unify pixels for geographical entities. This cannot accurately delineate the parcel entities in real-world and their complex interactions either. Therefore, in this project we used parcel objects as our basic units for analysis. We will obtain the natural and socio-economic features of the parcel objects by fusing the information from high-resolution remote sensing images and social sensing data, respectively. We will also establish a land parcel-based ABM-CA model (LPB-ABM-CA) to reflect the dynamics of human activity and their impacts on the evolution of urban structure at micro scales. The contents of our project include: (1) identifying the land-use types for parcel objects based on the integration of high-resolution images and social sensing data; (2) acquiring the functional characteristics of parcel objects by using multi-sourced social sensing data; (3) mining association rules between residents’ activity and built-environment; (4) collectively simulating spatial evolution of urban structures and residents’ activities through the LPB-ABM-CA model.

我国城市地块功能错配已导致严重的效率低下和交通拥堵问题，亟需建立基于地块对象(Land Parcel-based; LPB)的城市模拟模型来揭示地块层面的“人-地”作用规律，优化城市规划与管理决策。然而，现有基于像元的模型难以准确表达真实地块的空间关系和相互作用。本项目拟建立基于地块对象的ABM-CA模型(LPB-ABM-CA)解决这一不足。结合高分辨率遥感影像和多源社会感知数据获取地块对象属性和城市空间结构精细特征，并揭示居民活动与城市空间结构的相互作用，从而驱动LPB-ABM-CA模拟城市空间结构演化过程。研究内容包括：(1)融合高分影像与社会感知数据识别地块对象的土地利用类型；(2)借助多源社会感知数据和主题模型获取地块对象的属性和城市空间结构的精细特征；(3)利用关联规则分析挖掘居民活动与城市空间结构的作用关系；(4)建立LPB-ABM-CA，实现城市空间结构与居民演化模拟。

我国城市地块功能错配已导致严重的效率低下和交通拥堵问题，亟需建立基于地块对象的城市模拟模型来揭示地块层面的“人-地”作用规律，优化城市规划与管理决策。然而，现有基于像元的模型难以准确表达真实地块的空间关系和相互作用。本项目拟建立基于地块对象的模拟模型解决这一不足。结合高分辨率遥感影像和多源社会感知数据获取地块对象属性和城市空间结构精细特征，并揭示居民活动与城市空间结构的相互作用，从而驱动城市地块模型模拟城市空间结构演化过程。在项目资助下，完成了既定的研究内容，包括：(1)融合高分影像与社会感知数据识别地块对象的土地利用类型；(2)借助多源社会感知数据和主题模型获取地块对象的属性和城市空间结构的精细特征；(3)利用机器学习挖掘居民活动与城市空间结构的作用关系；(4)建立城市地块模拟模型，实现城市空间结构与居民演化模拟。在此基础上，项目发展了耦合模拟与优化的城市增长边界模拟模型，并开展了全球1990-2010年30米分辨率的城市建设用地制图。项目共发表SCI/SSCI论文13篇，共有4篇入选ESI高被引论文，1篇入选ESI热点论文。