Detection of urban sprawl using a genetic algorithm-evolved artificial neural network classification in remote sensing: a case study in Jiading and Putuo districts of Shanghai, China

Urban sprawl results in the most complex process of land use and land cover change, which in turn has a compound impact on the structure and function of ecosystems in urban areas. The detection of urban sprawl based on remote sensing was studied in two districts of Shanghai, China. The study area includes Jiading district which is one of the fastest developing urban fringe areas, and Putuo district which is one of the downtown areas in Shanghai. The structure of the artificial backpropagation neural network (BPN) classifier was evolved by genetic algorithm (GA), including the connection values between neurons, hidden layer numbers and their neurons, and neuron correction values in all layers. A comparison of the proposed method was made with conventional classification methods such as the minimum distance (MD) classifier, maximum likelihood (ML) classifier and improved backpropagation neural network classifier. The result shows that the proposed approach has higher accuracy and reliability for the classification of remotely sensed data. Therefore, three epochs of Landsat Thematic Mapper (TM) imageries of the study area were selected in 1990, 2000 and 2006, and the changes of urban lands for different time intervals were detected. A comparison of the two districts and their towns was also made, which characterizes urban sprawl in the typical urban fringe and downtown areas of Shanghai.