To understand the impact of air pollution on human activity we use 50 million geo-tagged Weibo (the Chinese Twitter) check-ins as a proxy. The air quality data consists of daily records of all the monitored pollutants for 251 cities. We define sensitivity as the impact of air pollution on urban activity: a decrease in Weibo check-ins indicates a greater sensitivity to air quality.
我们使用了2015-2016年的五千万个具有地理标记的微博签到数据来研究空气污染对城市活动的影响。经过必要的数据清洗，我们将研究范围确定在251个地级市。研究所使用的空气污染数据包括了这些城市每日的空气质量指数和六大污染物平均浓度。我们使用“敏感度”来定义空气污染对人类活动的影响，污染发生时签到活动减少的越多标志着对空气污染有着更大的敏感度。
As different cities react differently to air pollution, we observe 3 main trends. One of the factor that can explain this spatial heterogeneity is income.
不同城市的居民对空气污染有着不同的敏感度。我们主要总结了空间上的三大特征（具体描述见论文）。通过面板模型检验相关假设，我们发现收入水平是解释这种空间异质性的首要因素。
People in a city with an income of 20,300 Yuan above the average are 1.6x more sensitive.
收入超过平均收入20,300元的城市居民对空气污染的敏感程度是平均收入居民的1.6倍。
Air pollution alone can’t explain sensitivity. Here we compare two hypothetical cities with different air pollution and income levels. We observe that a less polluted city with a high average income has about the same sensitivity as a more polluted city with a lower income.
This suggests that richer people can protect themselves by avoiding urban activity and reducing their exposure to pollution. That advantage is an environmental injustice.
仅仅空气污染不能完全解释“敏感度”。通过本研究，我们比较了两个具有不同空气污染程度和收入水平的城市。我们发现，空气污染程度较低同时收入水平较高的城市具有和低收入以及高污染的城市相同的“敏感度”。
这个发现表明富人能够通过避免户外活动以及避免暴露于污染来保护他们。这种行为也侧面证实了环境不公正现象。
Locals are at least 4x more sensitive to air pollution than tourists. Travelling expenses and limited travel duration could explain tourists’ minor concerns of air pollution, whereas locals have more flexibility.
本地居民对空气污染的敏感程度比游客高出四倍。这种现象是因为游客相较于空气污染更为关心旅游的费用以及旅行的有限时间。
The leisure-related activities are 4-6x more influenced by air pollution than the work-related activities. The flexibility of leisure activities allows locals to avoid exposure risks.
休闲型的活动受空气污染的影响程度是工作型的活动的4到6倍。
Sensitivity to air pollution impacts people’s decisions. We found a greater effect on general locals’ activities compared to visitors, and on locals’ leisure activities compared to work activities.
To understand better that avoidance effect we visualize the four most common types of pollutants along with AQI.
An interesting finding is the role of national Holiday, in which urban activities experience a drop in sensitivity and causes more exposure risk than any other days. The necessity of recreation activities on holidays might be a cause.
对空气污染的敏感程度还随着不同人群和不同活动类型而变化。我们的研究表明，本地居民普遍具有更高的“敏感度”，而休闲型活动相比于工作活动也具有更高的“敏感度”。
为了更好的表现人群和活动的差异，以上我们通过可视化来展示其对空气质量指数和四种常见污染物的“敏感度”差异。
另外一个有意思的发现是节假日所扮演的角色。在国家法定节假日期间，城市活动对空气污染的“敏感度”显著降低，从而造成较其他时间更多的污染暴露风险。
download press release
Exploring the Effect of Air Pollution on Urban Activity in China Using Geotagged Social Media Check-in Data
