Long-term Association between NO2and Human Mobility: A Two-year Spatiotemporal Study during the COVID-19 Pandemic in Southeast Asia

Since the COVID-19 pandemic, governments have implemented lockdowns and movement restrictions to contain the disease outbreak. Previous studies have reported a significant positive correlation between NO 2 and mobility level during the lockdowns in early 2020. Though NO 2 level and mobility exhibited similar spatial distribution, our initial exploration indicated that the decreased mobility level did not always result in concurrent decreasing NO 2 level during a two-year time period in Southeast Asia with human movement data at a very high spatial resolution (i.e., Facebook origin-destination data). It indicated that factors other than mobility level contributed to NO 2 level decline. Our subsequent analysis used a trained Multi-Layer Perceptron model to assess mobility and other contributing factors (e.g., travel modes, temperature, wind speed) and predicted future NO 2 levels in Southeast Asia. The model results suggest that, while as expected mobility has a strong impact on NO 2 level, a more accurate prediction requires considering different travel modes (i.e., driving and walking). Mobility shows two-sided impacts on NO 2 level: mobility above the average level has a high impact on NO 2 , whereas mobility at a relatively low level shows negligible impact. The results also suggest that spatio-temporal heterogeneity and temperature also have impacts on NO 2 and they should be incorporated to facilitate a more comprehensive understanding of the association between NO 2 and mobility in the future study.