What nitrogen dioxide emissions tell us about the fragile recovery in South Asia

South Asia is experiencing an unprecedented health crisis as a devastating new COVID-19 variant has destroyed thousands of lives causing more deaths in last two months than in the entire preceding period.

The full extent of the impact of the pandemic can not immediately be discerned by national official statistics. Moreover, COVID-related health and economic effects are not distributed evenly geographically: populous urban centers have been hit harder than outlying areas. However, we can increasingly rely on other sources of data to indirectly measure the impacts of COVID-19 on economic activity.

One example may be in changes to concentration of nitrogen dioxide (NO2) in the air. NO2, one of a group of gases called nitrogen oxides, is a pollutant monitored from space-based satellites. NO2 is a gas primarily released by fuel consumption from vehicles (particularly in traffic congestion), power plants and other industrial activities. Because so much economic activity in cities depends on the burning of fossil fuels, we can estimate how much economic activity was occurring by studying patterns of NO2 concentrations.

Data on NO2 concentrations are available in near real-time from the European Space Agency’s Sentinel5-P program. With cloud-based processing platforms like Google Earth Engine, we can quickly obtain estimates of NO2 emissions across the globe. A number of studies compare NO2 emissions from before and during any lockdowns that occurred in 2020. A 2020 study of cities in Sub-Saharan Africa finds that NO2 provides a useful, albeit “noisy”, real-time proxy measure of how COVID-19 has affected economic activity. Deb and colleagues also found a clear relationship with industrial production growth and NO2 in a sample of countries.

NO2 emissions were undoubtedly down in April and May 2020 during the lockdowns, but can we detect the subsequent economic recovery and more recent shutdowns? To find out, we averaged NO2 concentrations in a circle of 20 km around 10 cities in South Asia (Kabul, Dhaka, New Delhi, Kolkata, Mumbai, Kathmandu, Quetta, Karachi, Islamabad, Colombo). We then averaged the estimated emissions within that circle across each calendar month.

Mean NO2 emissions in Dhaka, Bangladesh in April 2019 (left) and April 2020 (right). 20 km circle drawn on map.

Earth Engine has NO2 data available only from August 2018, and the data shows that different cities have different levels of NO2 throughout the year. With a longer series of data, we could have better understood how normal or abnormal NO2 concentrations have been during the pandemic. As it is, because there’s not much data to work with, we compare each data point with the year before it for an ‘apples-to-apples’ comparison, but also against a ‘benchmark’ year, the period of August 2018-July 2019 before the crisis.

The figures below show the smoothed year-on-year change in average monthly NO2 concentrations in these 10 urban areas in South Asia. The dotted lines show the change against the ‘benchmark’ year, since April and May 2020 already saw full lockdowns in much of South Asia. The grey shaded sections correspond to time periods when there were national lockdowns or strict restrictions on mobility such as curfews. In some cases, restrictions were very gradually loosened or applied unevenly depending on affected localities (as represented in the faded shading). Beginning in the third quarter of 2020, most countries implemented geographically targeted restrictions. 

Growth in Nitrogen Dioxide (NO2) concentration in major South Asian cities (3-month smoothing)

_{Note: Calculated taking log difference compared to the year-earlier month, 3-month moving average. Last observation labelled is average growth of March-May 2021 compared to previous year. Dotted lines denote 3-month average of NO2 change for the last observations to May 2021 (circle marker) against the month two years earlier, before the pandemic. Shaded areas refer to lockdown or mobility restrictions dues to COVID-19.}

The figure confirms that NO2 concentrations in general fell during periods of restrictions, and more recently rose very quickly in cities prior to the most recent surge of COVID.  Five observations are evident.

1. In all cities, NO2 concentration rebounded to pre-COVID levels only after falling into negative territory in mid-2020.  Smaller cities such as Kabul, Islamabad, Kathmandu and Quetta did not see a very radical increase in NO2 as the economy opened, in part because they are less urbanized.  
2. Compared to 2020, Mumbai, Kolkata, New Delhi, Kathmandu and Colombo only recently started to see the positive growth in NO2, (since around March 2021). These are the cities ravaged by the surge of the new, more deadly COVID-19 variant since end-March 2021, which has overwhelmed the health systems.
3. Colombo experienced a surge in May following a few well-contained COVID-19 cases last year which has coincided with increase in NO2 levels as the economy opened up. Sri Lanka’s economy had started to languish since around April 2019 when terrorist attacks negatively impacted tourism. 
4. Dhaka, Karachi, Kolkata, Mumbai and New Delhi all saw a significantly large increase by early 2021; with NO2 emissions growing by over 48 percent by May 2021 compared to 2020. The first four cities also happen to be the most densely populated cities in the world.
5. The sharp increase in NO2 emissions seen in April-May 2021 in all cities but Quetta is mostly because the comparison is against April-May 2020 when all urban areas were placed under very strict lockdowns. If we compare the last nine months against the same period in 2019 before the pandemic (the dotted lines), growth plateaued or turned negative compared to the same months in 2019 (circle marker shows May growth level). In other words, the recent mobility restrictions and curfews do seem to be translating into reduced NO2 emissions when compared to a more ‘normal’ period. The sharp retreat of NO2 in New Delhi and Mumbai is noteworthy.

This resumption of economic activity since end-2020 is also corroborated with data on nighttime light intensity around the geographic locations of these cities as shown in the Spring 2021 South Asia Economic Focus. By January-February 2021 nighttime light intensity showed economic activity recovering to pre-COVID levels in the geographic areas around Colombo, Mumbai, Karachi and New Delhi, which experienced the largest surge in per-capita COVID-19 cases by April-May 2021--much more than most other parts of the region.

Separately, The sharp increase in NO2 levels may also cost lives. While all of the nitrogen oxides are harmful to human health and the environment, NO2 is of greater concern, according to the US Environmental Protection Agency, because it has been linked to respiratory issues like asthma and difficulty breathing.New evidence by Zivin and others in the United States established a clear direction of causality: a 10 percent improvement in the air quality index in the United States led to a 15 percent drop in hospitalization from the main respiratory illness, influenza. In parallel, studies conducted in advanced economies find that exposure to air pollution, especially NO2 and PM2.5, increase the susceptibility of infection and mortality specifically from COVID-19. Though there is no comparable study for South Asia, air quality in urban areas is typically much worse than in the United States and COVID-19 is of course a much deadlier respiratory disease than influenza. The implication is that higher pollution levels could contribute to a greater incidence of COVID-19 in South Asia as well.

More research is needed to see whether this high-frequency indicator is a reliable proxy for economic activity. But for the moment, this adds to the battery of high-frequency indicators showing the impact of COVID-19 and the disparities across South Asian cities.