Research

London's heavy traffic with road emissions significantly impacts air quality, which becomes being a major concern. This study predicts future NO₂ concentrations based on spatiotemporal data around citywide monitoring stations.

Key contributions

• Compiled a comprehensive spatiotemporal feature set with traffic characteristics (e.g., road conditions, network structure, traffic volume), providing valuable insights for transportation policy.
• Incorporated pollutant dispersion knowledge to build a spatiotemporal dynamic graph of monitoring stations, enhancing accuracy.
• Developed a hybrid model combining Graph Neural Networks and LSTM, capturing complex spatiotemporal dependencies of road emissions.

Experiments on hourly air pollutant data from London's roadside stations, sourced from the London Air Quality Network (LAQN), validate the model. Results show a 12.5% accuracy improvement with the graph attention mechanism and extended forecast horizon to 12 hours using the LSTM encoder-decoder, outperforming baselines, especially in long-term (9-12h) predictions.

Extensive literature supports the efficacy of zonal control policies in reducing emissions and promoting green transport. This study examines the impact of London’s ULEZ on electric vehicle (EV) adoption, using the 2021 Inner London Boundary to set treatment and control groups.

Given the rapid growth of EV adoption in Greater London, potentially due to spatial spillover effects from the ULEZ policy, additional variables accounting for distance decay, spatial interaction and spatial autocorrelation were included to capture spatial patterns. Moreover, wedge-shaped urban partitions were constructed to address radial and directional heterogeneity within similar geographic regions.

Regression results indicate that, after controlling for spatial and temporal heterogeneity, the ULEZ policy's effect on EV adoption is not consistently significant within Greater London. This suggests synchronized EV adoption growth inside and outside the boundary due to complex socio-economic factors and transport infrastructure, beyond simple spatial effects. Anticipatory effects may further explain this limited impact, as people likely adapted to future standards ahead of ULEZ enforcement.

Future research considers employing staggered DID for multi-period data and propensity score matching to balance samples and control for confounding variables.

To achieve carbon peaking and carbon neutrality goals, it is necessary to recognize the spatiotemporal evolution patterns of urban low-carbon competitiveness at a regional scale. This study focuses on the low-carbon competitiveness of 41 cities in the Yangtze River Delta from 2000 to 2020.

Key contributions

• Developed a comprehensive index system for evaluating urban low-carbon competitiveness, consisting of three dimensions: carbon emissions, socio-economic factors, and low-carbon technologies and policies.
• Analyzed the time-series evolution of low-carbon competitiveness using the TOPSIS grey relational analysis method and identified spatial patterns with LISA time path and space-time transition models.
• Classified the 41 cities into 7 spatiotemporal evolution types to reveal different spatial characteristics of low-carbon competitiveness.

The study shows significant changes in low-carbon competitiveness in the Yangtze River Delta over 20 years, with Shanghai ranking highest and Anhui Province lowest. The spatial correlation of carbon emissions varied greatly, while socio-economic factors were weaker. 81.1% of cities exhibited path dependence in their spatiotemporal evolution.