Evolution of intercity population flow networks in China based on nighttime light remote sensing data

Intercity networks are interconnected systems of cities and towns that collaborate and interact through social, economic, and infrastructural connections. Intercity networks constantly evolve, and analyzing their evolution is crucial for urban planning and regional cooperation. Existing intercity network simulation methods that rely on flow data have short-time series and face privacy issues, while methods based on statistical data suffer from data gaps in certain regions and are updated slowly. Nighttime Light (NTL) data offer a valuable alternative due to their advantages in time-series accessibility, rapid updating, and broad coverage. However, current studies based on NTL data are limited to regional scales, which restricts their applicability for comprehensive, large-scale, and historical analyses of urban development. This study uses machine learning models to simulate China’s intercity population flow networks, using intercity population flow data from Baidu as the target variable and features extracted from NTL, land cover, and road data as input variables. Three machine learning models, including Extreme Gradient Boosting (XGBoost), Random Forest, and Light Gradient Boosting Machine, were trained and tested on data from 2020 to 2021. The XGBoost model achieved R² values of 0.82 on the validation set and 0.77 on the test set and was selected as the optimal model for constructing intercity population flow networks in China for 2012, 2017, and 2022. The evolution of the intercity population flow network was examined from both spatial structure and city centrality perspectives. The findings revealed a shift in China’s intercity population flows spatial structure from a multipolar pattern to a combination of multipolar and rhombus-shaped patterns. From 2012 to 2022, China’s largest cities first developed and then drove the development of neighboring cities. This study introduces a novel method for intercity population flow network simulation on a large scale, offering valuable insights for urban planning and strategic decision-making.