On the progress in urban ecosystem dynamic modeling
On the progress in urban ecosystem dynamic modeling
复制标题
作者：
Yulu Ya
本文对动态城市生态系统建模（DUEM）的研究进展进行了综述，总结了与城市生态系统建模领域相关的历史、方法、程序、软件开发和可用性等方面的研究进展。DUEM是一个多学科交叉的研究领域，涵盖景观生态学、城市人口学、社会学、城市规划、环境规划、环境经济学、防灾科学、城市卫生学、公共卫生等相关学科。DUEM地区的历史可以追溯到19世纪，当时研究者开始研究城市扩张和演变的过程。从那时起，世界各地的研究人员开发了各种各样的建模方法来帮助理解城市生态系统的动态和进化。在20世纪60年代以前，在城市系统建模研究中很少考虑城市发展的生态和环境方面。这一趋势在20世纪60年代开始发生变化，表现为发展综合城市生态系统模型，不仅模拟城市系统中的社会经济成分，而且模拟城市系统中的生态和环境成分。基于数学表述和系统表征的不同，现有的城市生态系统模型大致可分为四大类：(1)数学机制模型；(2)基于生态控制论的敏感性模型；(3)系统动力学模型；(4)多目标模型。除了这四种方法之外，其他类型的建模方法如生态足迹法、情景分析法和基于熵的方法也受到了研究者的广泛关注。尽管这些建模方法在外观上有所不同，但它们都遵循相同的一般程序，即包括模型定义、模型制定、计算机实现、校准/验证、模型性能评估和模型应用六个步骤。自从研究人员开始研究使用数学模型来研究城市系统的可能性以来，已经开发了许多建模系统。其中一些模型是通过使用通用计算机语言直接进行计算机编程开发的，而其他模型则是使用VENSIM、STELLA、DYNAMO和Matlab的Simulink工具箱等流行的仿真和优化软件包开发的。本文的表1列出了一些应用最广泛的城市生态系统模型，以及它们的开发商和适用性的信息。本文进一步解决了城市系统建模中的不确定性问题，指出城市系统建模总是受到整个模型开发和应用过程中产生的不确定性的影响。由于不确定性的重要含义，提出需要特别努力提高在DUEM研究中处理模型不确定性的能力。最后，本文总结了该领域潜在的研究方向，提出需要将技术和方法扩展到宏观和微观尺度，以实现该领域的进一步发展。此外，通过整合先进的人工智能技术和地理信息系统开发混合方法可能为模型改进提供另一种有希望的方法。
This paper presents a comprehensive literature review on the progress in Dynamic Urban Ecosystem Modeling (DUEM), summarizing various perspectives such as the history, method, procedure, and software development and availability that are pertinent to the urban ecosystem modeling area. DUEM represents a multi-disciplinary research area, which covers many related scientific disciplines including landscape ecology, urban demography, sociology, urban planning, environmental planning, environmental economics, disaster prevention science, urban hygiene, and public health. The history of the DUEM area can be traced back to the 19th century when researchers started to study the process of urban expansion and evolution. Since then, researchers throughout the world have developed a wide range of modeling approaches to help understand the dynamics and evolution of urban ecosystems. Before 1960s, the ecological and environmental aspects of urban development have rarely been considered in an urban system modeling study. This trend started to change in 1960s, which was manifested by the fashion of developing comprehensive urban ecosystem models that simulate not only the social-economic but also the ecological and environmental components in urban systems. Based on the difference in mathematical formulation and system representation, all the existing urban ecosystem models can be roughly classified into four broad categories: (1) mathematical mechanistic models; (2) eco-cybernetics based sensitivity models; (3) system dynamics model; and (4) multi-objective models. In addition to these four branches of methods, other types of modeling approaches such as the ecological footprint method, the scenario analysis method, and the entropy-based approach have also attracted wide attentions from researchers. Despite of their difference in appearance, all these modeling methods follow a same general procedure, which involves six steps including model definition, model formulation, computer realization, calibration/validation, model performance evaluation, and model application. Since the day researchers began to research the possibility of using mathematical models to study urban systems, numerous modeling systems have been developed. Some of these models are developed through direct computer programming with general computer languages, and the others are developed using popular simulation and optimization software packages such as the VENSIM, STELLA, DYNAMO, and Matlab's Simulink toolbox. Table 1 in the present paper lists a number of the most widely applied urban ecosystem models along with information regarding their developers and applicability. This paper further addresses the issue of uncertainty in urban system modeling, indicating that an urban system modeling is always subjected to uncertainty originated from the entire model development and application process. Due to the significant implication of uncertainty, it is proposed that special efforts need to be dedicated to improve the capability of handling model uncertainty in a DUEM study. Finally, this paper summarizes potential research directions for the DUEM area, suggesting that technologies and methods need to be extended to both macro and micro scales to achieve further advancement in this area. In addition, to develop hybrid approach through integrating advanced artificial intelligence technologies and geographical information systems might offer another promising way for model improvement.