A modified spatial entropy for urban sprawl assessment

Abstract

Urban sprawl is recognized as a challenge in urbanization. It is an unplanned spread of cities into its surrounding area that is caused by rising urban population. This phenomenon, if not properly controlled, can lead to several issues, e.g. traffic congestion, water and air pollution and excessive fuel consumption. A recent extension of the Shannon entropy such as the spatial entropy can be used to model urban sprawl. We introduce a new measure of spatial entropy from a previous measure of spatial entropy. The proposed spatial entropy adds the compositional information of land use maps to the spatial entropy. In order to consider affecting factors in urban sprawl calculations, a conditional spatial entropy is suggested. The urban growth process is affected by various environmental and socio-economic parameters. Some of these parameters have such a low significance in urban growth process, and then feature selection have some advantageous like to reduce overall training times, to deal with overfitting and to increase generalizability. Fuzzy rough set theory is utilized as a feature selection method. The results show that the proposed model provides some valuable evaluation of the urban sprawl as compared to previous methods. Moreover, the feature selection has a minute impact on the entropy values, especially in the new modified entropy, this implying to remove unimportant features from the features.