Putting Out the Fire

A 16-year-old from West Bengal is using satellite technology and machine learning to predict and prevent crop-residue fires
SK Reezaal Arafat with one of his devices
SK Reezaal Arafat with one of his devices
Updated on: 
3 min read

For two winters, 16-year-old climate activist Sk Reezaal Arafat was under the weather. He would watch ash settle on the windows of his home in Medinipur, West Bengal. The western skies were growing hazy, and the air was thick with smog. Initially, he assumed people were burning bushes or clearing nearby playgrounds. But an investigation revealed a more widespread culprit: farmers in neighbouring villages were burning the residue left after harvesting paddy. On some days, the AQI touched 300. “I thought we should be getting crystal-clear air here. But over the past two or three winters, we have been facing constant smog,” says Arafat, a Class 11 student.

His investigation soon uncovered the economics behind the practice. Unlike residue from manually harvested paddy, which can be collected and reused, the chopped stubble left behind by harvesters is harder and more expensive to transport. With few alternatives and little time between cropping cycles, burning becomes the easiest option. “Just telling farmers not to burn is not a solution. The transportation cost is too high, they don’t always have a use for the residue, and they are working within a tight timeframe,” he says.

Rather than remain just another climate activist, Arafat began exploring how technology could offer a solution. This led to the founding of SATARK, a student-led initiative that combines satellite imagery, machine learning and field-level monitoring to tackle agricultural fires. Working with 30 volunteers, most of them teenagers, Arafat developed a fire-detection model for thermal satellite imagery. Conventional polar-orbiting satellites pass over regions at fixed intervals, leaving fires that break out between passes undetected. The problem is particularly acute in West Bengal, where fragmented landholdings mean farmers often burn residue in several small fires. SATARK’s system scans for thermal anomalies every hour and sends alerts to a Telegram channel, along with a Google Maps link pinpointing the suspected fire.

But identifying fires after they start was not enough. When Arafat approached local authorities, he found that they were more interested in preventing them. “If you are only telling them where a fire is, you are making them less blind. But the fire has started, and so have its consequences,” he says.

SATARK subsequently developed Harvest Alert, a predictive system that identifies fields at risk of being burnt. It uses the Normalised Difference Vegetation Index (NDVI), derived from satellite imagery, to identify recently harvested fields. This information is combined with weather parameters such as humidity, temperature, rainfall, wind direction and atmospheric pressure to assess the likelihood of burning. The aim is to give authorities time to intervene before a fire breaks out. Yet Arafat recognises that alerts alone cannot solve the problem. Farmers need practical alternatives, including access to residue-collection machinery, affordable transportation and facilities that convert agricultural waste into biomass. “Just knowing where the fire will be or where it is happening is not a solution. You have to have a way out for the residue,” he says.

Stubble burning
Stubble burningRiya Aggarwal
Arafat recognises that alerts alone cannot solve the problem. Farmers need practical alternatives, including access to residue-collection machinery, affordable transportation and facilities that convert agricultural waste into biomass

This understanding also shapes SATARK’s engagement with farming communities. Volunteers are encouraged to learn about local cultivation practices, expenses and crop losses rather than simply ask farmers to stop burning. Building trust, Arafat believes, is essential to securing their cooperation. “Social capital should be built before a disaster and should be spent before it, not during it. In a disaster, you cannot make something new. You have to go with what you have already built,” he says.

SATARK has received a grant from Emergent Ventures to develop its infrastructure. For Arafat, however, the larger challenge remains translating satellite data into meaningful interventions. The ash on his windows first alerted him to the problem. His initiative is an attempt to ensure that the solution goes beyond detecting fires to addressing why farmers light them in the first place.