A review of Delhi’s long term air quality

Gautam Pradhan
Mar 8, 2023

Sometimes we need to take a pause to just understand the scope of gargantuan problems.

Delhi’s air quality is one such issue.

Earthmetry contributed to a paper driven by Sarath Guttikonda of Urban Emissions (along with Sai Krishna D, Bhargav Krishna, Hiren T Jethva and Puja Jawahar) that undertook a long term review of the situation.

You can read about the key insights in this post or on Twitter.

Delhi’s pollution levels have been in a tight range for 30 years now. Has nothing changed? Why is this so hard to fix?

To understand this better we need to borrow an idea from statistical process control.

Take a look at the below chart. Imagine that it describes the same type of system and that higher numbers are bad. Which case is easier to improve? Blue or Red?

Which case is easier to improve? Blue or Red?

The answer is counter-intuitive.

Blue describes a predictable system with natural variance. All the obvious issues are likely fixed. So you need to dramatically change how the system works to produce any improvement.

The numbers in Red show a large deviation starting at some point. There is a high chance of a singular cause.

Air pollution in Delhi is similar to the Blue problem. Yes there are some abnormal events like Diwali (a few days) or Crop burning (a few weeks) but even fully fixing those would only make a dent of ~5%.

The big causes are: Vehicle exhaust, road and construction dust, industries, cooking and heating and waste burning.

Some factors are uncontrollable, like the weather.

The weather in winter keeps polluted air within a tight volume both in terms of height (low mixing layer height) and spread (low wind). This makes pollution in winter 20–60% higher than average.

The wind also brings in dust during the dry summer prior to the monsoons.

The overall numbers (bars in the below chart) make one feel like a mouse running on a wheel!

The numbers in the 2020s are lower by ~30% from the pre-CNG era and ~20% from the mid-2010s. Some of this could be due to more monitors covering more locations now compared to the past.

Are there any success stories?

Delhi has 8% of the vehicles in the country concentrated in a small area.

Interventions like the metro, BS norms, mandated conversion to CNG and now EVs have helped us keep a check on pollution even though the number of vehicles have shot up by 5x in 30 years.

We have increased LPG usage for cooking. This also helps improve indoor air quality which disproportionately affects women and children.

About ~500,000 LPG customers were added in neighbouring states (Ujjwala program). But refill cost remains an issue.

Many power plants in and near the NCR have been closed (red) or lowered generation (yellow) over the years.

The remaining contribute 5–7% of the pollution in Delhi. It would be better to use more gas and less coal, and implement long-delayed flue gas desulphurization.

We need to continue the above initiatives but where is more work needed?

The dust problem, both road and construction: Even if we switch over completely to BEVs, pollution due to vehicles won’t go away. They leave a lot of dust on the road (tyre wear). They also kick up and resuspend road dust as they move.

Waste management: A lot of Delhi’s waste burns in both landfills and localties. This needs segregation at source and better waste processing while at the same time ensuring dignity and livelihood to sanitation workers.

Winter heating: Winter is cold and many resort to burning coal, wood etc. to stay warm. We need to find ways to get electric heating in peoples’ hands.

A lot has been tried in 20 years and some experiments are failures. The biggest one is the smog towers. These are completely unscientific and attack the wrong end of the problem.

What is the role of data?

  • Expand our ways of tracking human activities that lead to emissions
  • Have 100+ monitors just in Delhi, with many more low cost monitors and satellite-based continuous monitoring
  • Track health impacts (COPD, asthma, mental health etc.)

But there is no need to wait for data to act!

There is a lot more information and nuance in the paper so please head over and read it for yourself!

Transform your data journey

Join us to get easy access to a
wide portfolio of datasets.