Objective In recent decades, China's rapid urbanization has led to large amounts of energy consumption and carbon dioxide emissions, triggering significant urban heat island effects and air pollution, which seriously threaten the physical and mental health of residents. How to improve the urban habitat has become the focus of urban planning. Urban ventilation, as an important way to deliver fresh and cold air to built-up areas, can effectively improve the comfort level. Although many international scholars have researched the optimization of urban ventilation, the existing research in China mainly focuses on the measurement of the current ventilation environment in cities. It neglects the long-term dynamic changes of the urban ventilation environment in response to different planning objectives in the process of urban development.

Methods This study proposed a research framework with general applicability to simulating future changes in ventilation environment within city-region systems for the first time. Firstly, this study defined five indicators (land surface temperature gradient, roughness length, forest canopy density, elevation variation coefficient, and slope) in three dimensions (surface air pressure gradient, surface roughness and surface relief) to evaluate the ventilation environment from three dimensions: surface temperature gradient, surface roughness, and surface undulation. Secondly, based on previous studies and considering the natural environment and socio-economic development of the study area, DEM, slope, distance to primary roads, distance to secondary roads, distance to tertiary roads, population density, GDP, and building density were selected as driving factors in this study to investigate the transformation pattern of land use types. The driving factor data and the land use type data of the study area in 2000 and 2010 were imported into the PLUS model. Three scenarios of natural development, farmland protection and ecological priority were set to simulate the land use types. Then, a prediction model was constructed based on the random forest algorithm. The land use types and ventilation environment of multiple scenarios in 2010 and 2020 were input into the validated prediction model to simulate changes in the future ventilation environment. Finally, depending on the historical trends in ventilation potential and differences across multiple scenarios within the study area, the applications were proposed for spatial planning and management of metropolitan areas.

Results Influenced by the urban development from 2000 to 2020, the ventilation environment of the Taiyuan metropolitan area varies with time, space, and planning objectives. It showed a decreasing trend and an increasing trend in local areas. Patches of better ventilation environments occurred at Jinyang Lake, Fen River, and Park. Several potential ventilation corridors were formed on the east and west sides of the built-up area. Under the scenario of natural development, farmland protection, and ecological priority, the ventilation environment of the study area shows a gradual improvement trend from 2030 to 2050. Specifically, from 2030 to 2050, the grade of the ventilation environment is gradually improved; from 2040 to 2050, the area of the ventilation environment of the grade of extremely good and good is significantly increased. Meanwhile, the ventilation environments in different regions show differences in spatial distribution. Under the natural development scenario, there is a significant increase in the area of ventilation environments classified as extremely good and good in 2030. From 2030 to 2040, the area of poorly rated ventilation environments increases between the central city and the Dongshan and Xishan regions. From 2040 to 2050, the ventilation environment in the above areas gradually improves, and a significant wind corridor with high ventilation potential appears between the eastern mountains and the central city. The continuous protection of basic farmland effectively limits the encroachment of construction land on farmland. By the end of 2050, a large amount of basic farmland is retained in the north and south plains. From 2030 to 2050, the villages and towns located in the southern plains transform from early isolated islands to a grid-like structure, and the ventilation environment level of the surrounding roads, rivers and other linear spaces gradually increases. Under the ecological priority scenario, water bodies, forests and sparse grasslands are preserved to the maximum extent possible. The area of ventilation environments with a rating of extremely good increases significantly from 2030 to 2050. These areas are mainly distributed around large water bodies such as Fen River and Jinyang Lake.

Conclusion In this study, a dynamic simulation model of the urban ventilation environment was constructed and the spatial distribution of the ventilation environment under different scenarios was mapped. To cope with a series of urban problems caused by urban sprawl and densification, the trends of the ventilation environment in the Taiyuan metropolitan area are predicted and identified, and in this way, a series of targeted optimization strategies are proposed. However, this study still has limitations. Since this study used historical data with an interval of 10 years, various factors, such as climate change and human activities, may have interfered with the simulation results during this period. Therefore, to improve the simulation accuracy, future studies can incorporate more environmental factors into the modeling process to explore the objective laws of the ventilation environment in the process of long time-series changes, which could provide guidance for the construction of local climate-adapted cities.