Rethinking Urban Inefficient Land Identification and Optimization: A Case Study in Shenzhen, China

Identification and optimization of urban inefficient lands is an important policy of Chinese government to achieve high quality urban development. Currently, the mechanism for identifying inefficient lands is lack of scientific support, which affects the policy effectiveness. Previous studies focused on exploring optimization strategies for inefficient lands from the perspective of land use classification. Few studies directly considering the specific optimization sequence of those different parcels with the same land use function. To fill these gaps, this study develops a typology-based framework including five aspects (functional vitality, built environment, development suitability, spatial integration and economic intensity). We build an inefficient lands identification and optimization system using PCA, K-Means and Random Forest approaches. Taking Shenzhen as an example, a rapidly developing city in China, we validate our framework and methods. Three conspicuous aspects in this study were identified to facilitate decision-making: (1) The typology-based framework has high identification accuracy and it can reliably reflect the current state of urban inefficient lands in Shenzhen. (2) The encoded expression from the typology pedigree provides detailed information. It helps to develop customized optimization strategies for each parcel. (3) The typology pedigree combined with the random forest algorithm has broad application prospects in studying the specific optimization sequence. This case can serve as an effective benchmark for future urban redevelopment decisions, especially for other cities facing the challenges of rapid urbanization and optimization decisions.