The containment and closure policies adopted in attempts to contain the spread of the 2019 coronavirus disease (COVID-19) have impacted nearly every aspect of our lives including the environment we live in. These influences may be observed when evaluating changes in pollutants such as nitrogen dioxide (NO₂), which is an important indicator for economic, industrial, and other anthropogenic activities. We utilized a data-driven approach to analyze the relationship between tropospheric NO₂ and COVID-19 mitigation measures by clustering regions based on pollution levels rather than constraining the study units by predetermined administrative boundaries as pollution knows no borders. Specifically, three clusters were discovered signifying mild, moderate, and poor pollution levels. The most severely polluted cluster saw significant reductions in tropospheric NO₂, coinciding with lockdown periods. Based on the clustering results, qualitative and quantitative analyses were conducted at global and regional levels to investigate the spatiotemporal changes. In addition, panel regression analysis was utilized to quantify the impact of policy measures on the NO₂ reduction. This study found that a 23.58 score increase in the stringency index (ranging from 0 to 100) can significantly reduce the NO₂ TVCD by 3.2% (p < 0.05) in the poor cluster in 2020, which corresponds to a 13.1% maximum reduction with the most stringent containment and closure policies implemented. In addition, the policy measures of workplace closures and close public transport can significantly decrease the tropospheric NO₂ in the poor cluster by 6.7% (p < 0.1) and 4.5% (p < 0.1), respectively. An additional heterogeneity analysis found that areas with higher incomes, CO₂ emissions, and fossil fuel consumption have larger NO₂ TVCD reductions regarding workplace closures and public transport closures.

为遏制 2019 年冠状病毒病 (COVID-19) 的传播而采取的遏制和关闭政策几乎影响了我们生活的方方面面，包括我们生活的环境。在评估氮等污染物的变化时可能会观察到这些影响二氧化氮（NO ₂），这是经济、工业和其他人为活动的重要指标。我们利用数据驱动的方法来分析对流层 NO ₂和 COVID-19 缓解措施，通过根据污染水平对区域进行聚类，而不是通过预先确定的行政边界来限制研究单位，因为污染无国界。具体来说，发现了三个代表轻度、中度和差污染水平的集群。污染最严重的集群锯对流层NO显著减少₂，具有锁定期重合。基于聚类结果，在全球和区域层面进行定性和定量分析，以调查时空变化。此外，面板回归分析用来定量对NO的政策措施的影响₂减少。本研究发现，2020 年，严格指数（范围从 0 到 100）提高 23.58 分可以显着降低贫困集群中的 NO ₂ TVCD 3.2%（p < 0.05），这对应于最大降低 13.1%实施了最严格的遏制和关闭政策。此外，工作场所关闭和关闭的公共交通的政策措施可以显著降低对流层NO ₂ 6.7％（P <0.1）和4.5％（P <0.1），则差簇中分别。另一项异质性分析发现，收入、CO ₂排放量和化石燃料消耗量较高的地区在工作场所关闭和公共交通关闭方面的NO ₂ TVCD 减少量更大。