Simulation and analysis of urban growth scenarios for the Greater Shanghai Area, China

This research investigates the potential of an integrated Markov chain analysis and cellular automata model to better understand the dynamics of Shanghai’s urban growth. The model utilizes detailed land cover categories to simulate and assess landscape changes under three different scenarios, i.e., baseline, Service Oriented Center, and Manufacturing Dominant Center scenarios. In the study, multi-temporal land use datasets, derived from remotely-sensed images from 1995, 2000, and 2005, were used for simulation and validation. Urban growth patterns and processes were then analyzed and compared with the aid of landscape metrics. This research represents the first scenario-based simulations of the future growth of Shanghai, and is one of the few studies to use landscape metrics to analyze urban scenario-based simulation results with detailed land use categories. The results indicate that the future expansion of both high-density and low-density residential/commercial zones is always located around existing built-up urban areas or along existing transportation lines. In contrast to the baseline and Service Oriented Center scenarios, industrial land under the Manufacturing Dominant Center scenario in 2015 and 2025 will form industrial parks or industrial belts along the transportation channels from Shanghai to Nanjing and Hangzhou. The study’s approach, which combines scenario-based urban simulation modeling and landscape metrics, is shown to be effective in representing, understanding, and predicting the spatial–temporal dynamics and patterns of urban evolution, including urban expansion trends.

Research highlights
► We simulated urban growth scenarios of Shanghai based on Markov chain analysis and Cellular automata model.
► We analyzed and compared urban growth patterns and processes among different scenarios with the aid of landscape metrics.
► Multi-criteria evaluation and analytical hierarchy process methods with expert interviews proved to be an efficient way to make transition rules in the CA simulation.
► Scenario-based urban modeling and landscape metrics, is shown to be effective in representing, understanding, and predicting the spatial–temporal dynamics and patterns of urban evolution.