Resumen de Machine learning-assisted evaluation of land use policies and plans in a rapidly urbanizing district in Chongqing, China
Tingting Xu, Jay Gao, Yuhua Li
Analysis of annual land use change is crucial to timely assessment of the impacts of land use policies, especially in rapidly urbanizing areas such as the Liangjiang New District of Chongqing, China. This research aims to assess the impacts of multi-level land use policies and plans and their effectiveness in protecting farmland, and to recommend policies to remedy defective ones. Markov Chain–Cellular Automata modelling was integrated with machine learning to predict future urbanization. The results show that prior to the establishment of the District in late 2010, the urban area grew at 13% annually because local policies promoted profit-motivated development. This rate declined drastically to 4% during 2011–2012 in direct response to the changed local land use policies and strict enforcement of national policies. These policies stabilized the urbanization rate at 7% over the subsequent years. The safeguard of previous farmland was possible only when the national policies were aligned closely with the local policies in their aim, provided both were rigorously enforced through a land management agency. Urban land in 2020 is predicted to growth by at least 464 km2 with some differences between two scenarios. In the baseline scenario (no change in policies), almost 70% of new growth will take place inside three industry zones and the duty-free ports. However, in the national overall plan scenario, a significant amount of marginal farmland will be changed into greenfields within the planned urban zones. It is recommended that a hierarchical farmland protection policy be implemented. It should incorporate the national policies that must be adapted to the local settings to achieve sustainable urbanization and to minimize farmland loss.
