Prediction Modeling and Driving Factor Analysis of Spatial Distribution of CO₂ Emissions from Urban Land in the Yangtze River Economic Belt, China

Abstract

In recent years, China’s urbanization has accelerated, significantly impacting ecosystems and the carbon balance due to changes in urban land use. The spatial patterns of CO2 emissions from urban land are essential for devising strategies to mitigate emissions, particularly in predicting future spatial distributions that guide urban development. Based on socioeconomic grid data, such as nighttime lights and the population, this study proposes a spatial prediction method for CO2 emissions from urban land using a Long Short-Term Memory (LSTM) model with added fully connected layers. Additionally, the geographical detector method was applied to identify the factors driving the increase in CO2 emissions due to urban land expansion. The results show that socioeconomic grid data can effectively predict the spatial distribution of CO2 emissions. In the Yangtze River Economic Belt (YREB), emissions from urban land are projected to rise by 116.23% from 2020 to 2030. The analysis of driving factors indicates that economic development and population density significantly influence the increase in CO2 emissions due to urban land expansion. In downstream cities, CO2 emissions are influenced by both population density and economic development, whereas in midstream and upstream city clusters, they are primarily driven by economic development. Furthermore, technology investment can mitigate CO2 emissions from upstream city clusters. In conclusion, this study provides a scientific basis for developing CO2 mitigation strategies for urban land within the YREB.

Keywords

• urban expansion
• spatial prediction
• LSTM
• geographical detector
• driving factors