Urban growth modeling with artificial intelligence techniques
Abstract
The goal of this study is to simulate and predict urban dynamics changes and quantify their impact on different land uses. For this purpose, a multi-spatiotemporal urban growth model is designed based on artificial intelligence techniques including, cellular automata, genetic algorithms, fuzzy logic and neural networks. Cellular automata defines a set of transition rules as a function of spatial neighborhood structure and input data. It serves as a modeling engine to simulate changes in spatiotemporal urban dynamics. Calibration of such rules is performed spatially on a township basis and temporally as a function of time. Three quality measures are developed as part of the evaluation scheme for model calibration. To increase the calibration efficiency, genetic algorithms are introduced to the cellular automata modeling process. The genetic algorithms parameters, including string design, encoding, selection criteria, crossover, and mutation are discussed. The use of fuzzy logic provides good initials for the cellular automata model and allows for including semantic knowledge into urban growth modeling. Fuzzy logic preserves the continuity of urban dynamics spatially by choosing fuzzy membership functions, fuzzy rules, and the fuzzification-defuzzification process. Finally, neural networks are used to model urban boundary growth. The developed model and approach are tested for the historical development of Indianapolis, Indiana for a period of 30 years. Land cover maps derived from satellite imagery, road networks, population data, and digital elevation are used as input data for the study. The modeling results show satisfactory fitness with close urban match patterns between the real and simulated data. The developed methodology can be used to support urbanization related studies, such as regional planning, sustainable development, environmental monitoring, ecosystem protection, and global warming.
Degree
Ph.D.
Advisors
Shan, Purdue University.
Subject Area
Civil engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.