Pixel based urban growth model for predicting future pattern

Abstract

Urban growth refers to an increase in the absolute of people living in urban areas and the spatial expansion of those urban areas. The study of the past and present situations of an area gives the influential factors on the pattern of urban growth. In fact, it is possible to predict the future pattern if the growth can be properly modeled. The urban environmental problems are usually valid to a specific area (locally) because most of the land area is depended on its geography, social, economic and policy factors. The factors may be different from one place to another. In this paper, we propose a generic solution to detect urban growth pattern for any region of interest by using pixel-based approach. Digital images are captured using satellites. The images are processed using Geographical Information System (GIS) to classify the urban and rural areas. A Cellular Automata (CA) theory is applied to determine the states of all pixels before filtration and selection on any region of interest are performed. Later, a back propagation Artificial Neural Network (ANN) algorithm is used to detect the pattern of urban growth and provides the prediction for future growth. As a case study, we choose an area called Klang Valley (the place of most rapid urban growth) in Malaysia (a country in South East of Asia). With six imagery data set of a same location (for 15 years duration), we model the urban growth by focusing on the pixel based approach. In this way, future urban growth can be predicted based on the past history. In addition, it is possible to focus on certain region of interest in the digital map for any particular urban influential factor. Thus, urban monitoring can be carried out to help the ecological system as well as producing systematic planning for future living.