Exploring urban morphology using multi-temporal urban growth data: a case study of Wuhan, China

Abstract

Understanding temporal urban growth process is crucial to the interpretation of urban morphology and a key challenge for the study of rapid urbanization in contemporary China. As an evolutionary process of urban form or landscape, urban morphology helps us track the trend of urban form development, which is characterized by multi-temporality in terms of data. The increasingly improved multi-temporal data availability, as pushed by the massive advances in geospatial technology, particularly remotely sensed imagery, offers great opportunity for measuring and describing urban morphology. Previous studies have reported the measurement of urban morphology using spatial methods including fractals, landscape matrix and density gradient. However these methods are subject to poor interpretation of urban morphology and low-level understandability to local planners. This paper demonstrates an innovative application of a machine learning method – Maxent for analyzing the urban morphology of a fast growing city – Wuhan, China. Multi-temporal data sets for 1955, 1965, 1993 and 2000 were processed from remotely sensed imagery. The model results facilitate local planners better to track and interpret the urban form development process of Wuhan city in the past half-century. The paper also illustrates that Maxent is an effective exploratory method and tool for analyzing urban morphology.