Gasp: The Hunt for the Dirtiest Air in India

Delhi isn’t the most polluted Indian city. There is worse.

It was a laid back weekday while I was on vacation in New Delhi. I was casually surfing the web when I came across a video from the Wall Street Journal that claimed that Delhi’s air was 20 times worse than the safe limit.

This Is How Bad New Delhi’s Air Pollution Is

This video really upset me. Delhi is where I grew up. My entire family lives in the city. I was aware of the city’s air pollution problem, but I wasn’t expecting it to be this horrific. I thought…

How bad really is the air quality in India?

And how bad is the air that I am breathing right now?

I needed a quantitative answer, so I began researching. First, I did a quick Google search. I found several news articles with air quality values for specific days of the year, but the values were highly volatile, changing each day. Next, I came across some websites giving live air quality values for specific locations in India, but I couldn’t derive much long-term meaning from those. After that, I came across experiments that people had conducted in different locations using handheld air quality monitors, much like the one in WSJ’s video, which were interesting, but couldn’t be generalized. I realized that any such experiment wasn’t a good representation of the true situation for 2 main reasons:

• There was no way to verify the quality of the sensor used, and as a result, the reliability of the readings was undermined.
• The experiment was only applicable to the day, time and place where it was conducted, and was too microscopic.

I immediately knew that I had to find a data source that was both reliable and long-running. And I finally landed on the Central Pollution Control Board, India’s website. CPCB had published data online that was gathered using monitors that adhered to internationally recognized measurement standards and were scattered in various cities across India. This, however, was a raw data source that returned time series readings of a single sensor at a time after filling out a website form.

If I could just perform a simple aggregation on the data for all sensors in Delhi for a year, I would have a better understanding of the air quality that I was breathing.

I decided to scrape CPCB’s website.

I created a short iPython notebook that automated the form filling process to collect PM2.5 (small particulate matter) data for all sensors nationwide. The notebook and the derived data set for all sensors across India can be found here. You can use it to make your own observations.

Having gathered the data, I was able to gain some meaningful insights about the situation through some visualizations using Google Maps’ API.

PM2.5 Annual Average in India for 2016

It was easy to make out from this air quality heat map that there are multiple regions of bad air quality in and around cities such as New Delhi, Patna, Lucknow, etc. The full heat map for annual average values for all cities with CPCB sensors that I created is here.

Air Quality India Plot 2016

Looking at some quick numbers, it turned out that Delhi wasn’t the most polluted Indian city in 2016.

Gaya and Ahmedabad had worse PM2.5 annual averages than Delhi in 2016.

The 5 Most Polluted Cities in India:
Gaya: 196.22 ug/m3
Ahmedabad: 183.35 ug/m3
Delhi: 182.68 ug/m3
Kanpur: 166.36 ug/m3
Lucknow: 164.43 ug/m3
The 5 Least Polluted Cities in India:
Tirupati: 32.37 ug/m3
Solapur: 38.83 ug/m3
Mumbai: 40.38 ug/m3
Visakhapatnam: 45.88 ug/m3
Panchkula: 53.98 ug/m3

It was evident from the heat map that the least polluted cities were in the Southern region and were close to the coast. This would be useful for future studies.

Next, I wanted to inspect Delhi more granularly. So, I created this plot.

PM2.5 Annual Average in parts of Delhi for 2016

Air Quality Delhi Plot 2016

One thing I found very interesting about this one was that, although PM2.5 values were quite high in general, sensors specifically in DTU, ITO and Sirifort had exceptionally high values. By looking at just those 3 sensors more closely, one geographic commonality that came out was their proximity to the NH44.

Highest PM2.5 in areas along the NH 44

NH44 is India’s longest-running national highway that cuts through the heart of New Delhi. The continuous flow of heavy-vehicle traffic through it could be a potential source of poor air quality in the nearby areas.

At this point I knew, with some accuracy, the quality of air that I was breathing while living in New Delhi. The closest sensor to me was the one in Sirifort that had a dreadful PM2.5 average of 278.77.

What about the air quality in other major Indian cities?

There was no answer to this one because India just doesn’t have air quality sensors in all major cities, but I could come up with estimates on the basis of data from nearby cities with sensors.

I took the top 100 biggest cities in India and decided to predict the PM2.5 in 2016 for those. I interpolated PM2.5 values for the 100 most populated Indian cities using the data from the surrounding sensors and created another plot.

Air Quality India Plot with 100 most populated cities 2016

These numbers are highly speculative though, because by simply interpolating values from nearby sensors, I didn’t account for several other factors that could influence air quality, eg. coastal proximity, traffic patterns, terrain, elevation, wind patterns, population density, etc. I think a lot of these factors will have a high correlation with air quality. I will leave this detailed analysis for another time.

Closing Remarks:
Yes, the air quality in New Delhi and most of north India in general is really bad. We are way past the Indian PM2.5 (small particulate matter) safety limit of 60 ug/m3. I am personally breathing a PM2.5 value of 278.77 ug/m3 every day — more than 4 times worse than the safe limit.

What are the authorities doing about this? Well, they are trying to implement policies like the odd-even traffic schemes, with very limited real benefit.

I think there is a genuine need for each of us to address this problem more proactively. We need to be a lot more considerate of the environment as we perform our day-to-day actions. Whenever we are making a simple decision of using the car over the public transportation or throwing trash out on the street instead of waiting until later to find a dustbin, we need to think hard about how, through these small things, we are destroying our beloved Sone Ki Chidiya (The Golden Bird).

My only hope is that these findings will help us make better informed daily-choices.

All the scripts that I created in this process are available here for you to play around with.

Further Readings:
1. Hindustan Times’ Five Pollution Charts for 2016 http://www.hindustantimes.com/static/pollution-india-five-charts/
2. Times of India’s Blog Post — Air Pollution Study for a part of 2016
http://urbanemissions.blogspot.com/2016/02/cpcb-report-air-pollution-in-indian.html

Credits:
Johnny Chang, for helping me edit this post and for coming up with an apt title.