Tracking fires using AI

Last week, US space agency Nasa’s satellite images recorded scores of farm fires across Punjab and Haryana, indicating that the annual stubble burning had started. According to a report in the Down To Earth magazine, several red dots were seen, especially over the districts of Amritsar, Ludhiana and Patiala in Punjab, and Karnal, Kurukshetra and Ambala in Haryana.

According to a Down To Earth report, farmers in the two states burn over 12 million hectares of open fields to dispose of 20 million tonnes of crop waste in autumn after harvesting paddy to make way for the sowing of wheat. This results in high levels of particulate matter (PM), nitrogen dioxide, carbon monoxide and greenhouse gas emissions, resulting in a spike in pollution levels in north India and damaging the health of 300 million Indians.

A screenshot of the BreeZo app

What if one could monitor these fires better? A team of software developers, environmentalists and scientists has come together to create an AI-enabled platform to do exactly that. Gurugram-based collective Blue Sky Analytics has been using AI and satellite data to create a set of apps and dashboards that monitor air quality and emissions—while BreeZo, which is currently available for free on Google Play and has more than 4,500 users, looks at ambient air quality, Zuri is dedicated to farm and forest fires and Zorro tracks industrial emissions. When it comes to stubble burning, Zuri hopes to provide data analysis with detailed parameters and insights like fire counts, intensity, emissions, expected crop waste and calorific value, burning predictions, high-risk zones, and other information that can be used for monitoring and regulation.

The Blue Sky Analytics team

The idea for these apps came to the brother-sister duo of Abhilasha and Kshitij Purwar in 2017 when they started researching on air and water quality. “At that time, Abhilasha was working in a private equity firm in the US. But she came back, wanting to do something related to the quality of the environment," says Prachi Mishra, head of strategy, Blue Sky.

Several incidents contributed to the setting up of Blue Sky. For instance, in 2014, while working on air pollution projects for The Ministry of Environment, Forest and Climate Change, Abhilasha came across a river tributary that had turned completely orange in Chandrapur, a coal mining area in Maharashtra. The scene had a lasting impact on her and left a deep desire to do something about pollution. Finally in October 2018, after looking at satelite data from various sources, the Purwars decided to start a Geospatial Data Intelligence company to track air pollution emissions.

As it happens, they began to come across instances of AI being added to the “climate change toolbox" internationally. In May, for instance, Google awarded $1.7 million (around ₹11.9 crore) to a team from WattTime, a not-for-profit, as part of the Google AI Impact Challenge. The team offers technology solutions that make it easy for anyone to achieve emissions reductions without compromising on cost and function. Another organisation which has been doing pioneering work in this field is, Carbon Tracker, a London-based think tank , which “ focuses on the financial part of transitioning away from fossil fuels," mentions Earther, an environmental news website.

Last year, there was another article, in Forbes, on the use of Al and deep learning in climate research. The article, “The Amazing Ways We Can Use AI To Tackle Climate Change" by Bernard Marr, cited the example of the Green Horizon Project that analyses environmental data, predicts pollution and tests “what-if" scenarios that involve pollution-reducing tactics.

Inspired, the duo started Blue Sky Analytics to track air quality. “It got rejected by some investors as they thought the idea seemed possible only on paper. But there were others who thought this was a unique proposition and connected them with like-minded scientists and academicians," says Mishra. One such person was Sagnik Dey from the Centre of Atmospheric Sciences at the Indian Institute of Technology, Delhi (IIT, Delhi), who felt that such scalable data and web products could better track emissions and prioritize mitigation strategies.

Both Blue Sky and Dey made a presentation at a conference on “Stubble Burning—Findings, Ground Issues And Policy Perspectives", organized by the Centre of Excellence for Research on Clean Air (Cerca) at IIT, Delhi in September.

At present, Blue Sky is building automated data pipelines by sourcing data sets from Nasa and European Space Agency, public APIs (application programming interface). “We get nearly 3-5 TB of satellite data every month. Using AI, this data is then cleaned, organized, and analyzed," explains Saheel Ahmed, data scientist at Blue Sky. The algorithm processes and organizes this data, keeping what is required, backing up the rest. The insights and data analysis go to the apps and dashboards—this entire process is part of the Cloud Native AI-powered Data Refinery. Say, you zoom into the map of Punjab on the Zuri dashboard, it will give you details of the number of farms, village-wise and district-wise, and also on the intensity of these fires using FRP (fire radiative power) captured from NASA satellites.

This year, the focus is on collecting data. From next year, Blue Sky Analytics will start doing predictions well before the burning season, allowing policymakers to focus in advance on farms that have been known to burn crop waste historically. “For instance, the government has allocated subsidy for farms in 52 districts of Punjab and Haryana. But not all farms in the 52 districts produce the same amount of emissions," says Arun Duggal, an IIT, Delhi alumnus and founder, Cerca. “For that we need to study the air flow patterns as well." AI-generated data sets have revealed that during October and November, air flow brings in maximum emissions from six of these districts to Delhi. “Maybe it will be better to give more subsidies to these six districts," he adds.

Meanwhile, Blue Sky Analytics hopes to serve as a marketplace enabler as well. Through Zuri, it seeks to empower the farmer so that instead of burning stubble, she can sell it to enterprises working towards sustainability—government holdings or companies such as Ikea and Fabindia, which have shown interest. The stubble can be used for packaging. “We want to bring all the stakeholders on to one platform so that, say, an Ikea doesn’t have to chase farmers for stubble. And also so that farmers get a fair price. It should be a win-win situation for all—the farmers and the community at large," says Abhilasha.