We have known the true utility behind wearing a mask only once the pandemic set in, but Reshma Kosaraju has been familiar with it for some time now. That's because when she moved to the San Francisco Bay Area from Chicago a few years ago, it was around the same time when a major wildfire broke out in Northern California. The 15-year-old recalls how even though they were over 200 miles (321 km) away from the fires, they still had to wear masks. And this got Reshma thinking. "I started researching and found that the current methods focus on detection of forest fires after they have occurred and have several limitations. I wanted to investigate and see if I could figure out a way to predict fires before they occurred, which could save lives, money and the environment," she says. It followed this train of thought that led to her winning the Children’s Climate Prize (CCP) in November this year. Instituted by the Swedish company Telge Energi, this award is given to youngsters between the ages of 12 and 17 years for their efforts towards making the planet a little more habitable.
Forest fires are a bitter reality, whether it was Odisha's Simlipal forest that was burning this year or the blaze that set Reshma's mind to work — it's all-consuming, which is why it needs fervent attention. So, Reshma decided to harness the power of Artificial Intelligence (AI) to see if there was a solution that could be arrived at. But it was a challenge because AI wasn't Reshma's domain, all she knew was that it was useful for predictive tasks and that was enough for her to get started. This long journey of acquainting herself with the software began three years ago and she made the most of online resources. "Even throughout the project, I had to constantly reference external resources to help me build the model," she shares. Now, the Deep Learning model she has built uses factors like dryness of the soil, humidity in the air, speed of wind and temperature. With the help of all these parameters, her model is able to detect when forest fires are likely to occur, that too with 90 per cent accuracy.
But achieving that level of accuracy was another challenge that this teenager who resides in California's Saratoga area had to surmount. "I found that the accuracy was quite low due to the small amount of data that was in my initial dataset. Thus, I used a meteorological government database and manually expanded my dataset, which improved the accuracy greatly," shares the youngster. For now, most of her data comes from various locations in the United States and soon, she would like to add data from other regions around the world so that the model can be used globally as well. "I plan on operationalising it by making it into an app, so that it is accessible to the general population," says Reshma.
As far as winning the prize itself goes, for which youngsters from 32 different countries were nominated, she says, "Winning the Children's Climate Prize is truly an honor and tells me that others think my project has the potential to have a significant positive impact on the environment. This validation from the scientific community and the public at large has given me the confidence I need to continue doing what I am doing."
