PREDICTING URBAN GROWTH WITH REMOTE SENSING 
AND DYNAMIC SPATIAL MODELLING 
Xiaojun Yang 
Department of Geography, Florida State University, Tallahassee, FL 32306, U.S.A. 
xyang@fsu.edu 
Commission VII, WG [1/4 
KEY WORDS: Remote Sensing, Spatial Information Sciences, Urban, Modelling, Monitoring, and Visualization 
ABSTRACT: 
This paper presents a research that integrates remote sensing, GIS, and dynamic spatial modeling for predicting urban spatial growth 
with different development conditions considered. The study area has been a fast growing American metropolis. The prediction is based 
on a cellular automate urban growth model governed by a set of complex transition rules combining both socio-economic and 
biophysical conditions. Historical urban extent data derived with remotely sensed imagery are used to calibrate the model. Two possible 
future growth scenarios are assessed. The first scenario assumes that the current development conditions do not change and therefore, 
can be termed as 'continuation'. The second is a hybrid growth strategy in which both conventional urban development and alternative 
growth efforts are addressed. It is found that many small-size urban patches would emerge and smaller ones would merge to form larger 
urban clusters. If current conditions do not alter, the process of urbanization would deplete vegetation and open space. A restrictive 
growth plan should be adopted in order to promote the livability and sustainable development in the study area. Overall, this study has 
demonstrated the usefulness of remote sensing, GIS, and dynamic modeling in urban and landscape planning and management. The 
methodology developed in this research can be casily adopted to other urban areas with similar growth patterns. 
  
1. INTRODUCTION 
Restless urban growth throughout the world has called for 
improved methods and techniques towards a better understanding 
of the dynamics of complex urban systems. Over the past 
decades, a great deal of research efforts has been directed to 
develop dynamic models in connection with urban and landscape 
applications (e.g. Meaille and Ward, 1990; Batty and Xie, 1994; 
Veldkamp and Fresco, 1996; White and Engelen, 1997; Clarke 
and Gaydos, 1998; Wu and Webster, 2000; and Wang and Zhang, 
2001). Among the documented models, those based on cellular Le | 
automata (CA) are probably the most impressive because they - f 
have grown out of an earlier game-like simulator and evolved qe 
into a promising tool for urban growth prediction and forecasting. 
  
  
  
This research has been focused on the exploration of cellular 
automata-based dynamic modeling approach for applied urban 
studies with Atlanta as a case study area (Figure 1). For the past Paulding 
three decades, Atlanta has been one of the American’s fastest | / 
growing metropolises as it emerged to become the premier SE e 
Cobb tip Gwinnett 
commercial, industrial, and transportation urban center of the a T a 
southeastern United States. Starting from 1996, the author has qo pesi us. 
been involved in various research projects focusing on the m |Clayton . 
understanding of the dynamics of change in Atlanta through the x eet [ 
use of geographic information technologies. This paper reports : { 
part of the result of urban growth simulation carried out with a 
cellular automate model, a key element of the above research 
effort. | 5 
This research was built upon the SLEUTH Urban Growth Model as 2 % 
(Clarke, 2000). SLEUTH derives its name from the six types of — — € 
data inputs: Slope, Land cover, Exclusion, Urban extent, 
Transportation, and Hillshade. This is a cellular automaton urban Figure 1 Location of the study area. 
growth model and its behavior is controlled by the coefficients of 
diffusion, breed, spread, slope resistance, and road gravity. The 
model considers four types of growth behavior: spontaneous 
neighborhood growth, diffusive growth and creation of new 
Coweta 
  
  
  
  
  
  
spreading centers, organic growth, and road influenced growth. 
Detailed description of the model can be found elsewhere (e.g. 
Clarke, 2000; Silva and Clarke, 2002). Using Atlanta as the study 
area, this project investigates the effectiveness of the SLEUTH 
1227