Analysis Framework of Urban Expansion in Taiwan and its Implication for Long-Term Developments using Satellite-Image Time Series data

The rapid growth of the urban population across the globe over the last decades has been increasingly contributing to urban expansion and with that, increasing pressure on the surrounding natural and rural environments has been exerted. This study aims at developing an analysis framework focusing on land use and land cover (LULC) mapping along with an integration of urban expansion indicators to provide representative metrics. By applying machine learning classification methods on multitemporal remote sensing data, including Support Vector Machine (SVM), Random Forest (RF), and Convolutional Neural Network (CNN), high accuracy LULC maps can be generated and employed as fundamental data for investigating urban sprawl patterns. Results of this investigation show that indicators such as the Abstract Achieved Population Density in Expansion Areas as well as the Shannon Entropy can be considered as a tool for cross-comparison and validation, and allow in particular to highlight changes between various land cover types derived by human development. This method has the potential to be transferable to other locations for quantitatively assessing urban expansion comprehensively.