Forecasting rain in India and around the world is a complex process that isn’t necessarily 100 per cent accurate. However, three engineering students and a professor from Chitkara University in Punjab are challenging this notion. Using their artificial intelligence-enabled technology, for example, they can tell you that it will rain exactly at 12.15 p.m locally.
Instead of depending on image data from satellites, they use real-time data to forecast rains at a local level.
Speaking to The Better India, Kartik Vij says, “We collect information from ground-based ‘BhuGoal’ nodes that we have developed instead of being entirely dependent on satellite image data. The data collected is in real-time. Therefore, the prediction is precise. The Meteorological Department uses outdated satellite image data that are 24 hours old. We use their data as well as our real-time data to provide accurate forecasts. The forecast that they (Met Dept) provide is not even close to precise. Due to global warming, the weather has become unpredictable over a 24-hour period. Also, the information they provide is for a whole city and does not cover most villages. We provide different forecasts for every five square kilometres.”
How does it work?
Inspiration for this AI-enabled technology came from Kartik’s mother. One fine day, his mother asked him to collect dry clothes from the terrace, fearing that it may rain. It was an epiphanic moment for him. Thinking about the bigger picture, he realised that weather monitoring and forecasting must be done in real-time to avoid such uncertainties.
“We use distortions in signals received by a TV set-top box and dish antenna to predict rain. We take the help of TV to crack it. We’d been noticing signal distortions whenever the weather turned bad. That’s when we realised dish TV signals could tell us about the weather. We installed a $5 device (‘BhuGoal’ nodes) between the set-top box and the dish antenna, and recorded variations in signals. The variations were fed to AI algorithms, which then have precise predictions,” says Vij.
Mentored by Dr Nitin K Saluja, the team comprises Kartik Vij, Rahul Kinra and Debarshi Ghosh, who are all final year engineering students at Chitkara University, Punjab.
According to Vij, their startup ‘Too Decimal Private Limited’ is being incubated at the Nadathur S Raghavan Centre for Entrepreneurial Learning (NSRCEL) at IIM Bangalore, although most of their work is still done out of Chitkara University.
How can this technology help?
The real-time information could be shared with farmers, who have lost a lot due to unpredictable rains. They could also share it with grain house owners.
“Every year, we see news of tonnes of grains being washed away by rain. In our first stage of deployment, we are aiming to solve this specific problem,” shares Vij.
As per media reports, their technology was among the eight that won awards at the Nasscom Foundation’s Tech For Good Summit earlier this week in Bengaluru.
Know more about the ATL Tinkering Innovation Marathon here.
(Edited by Shruti Singhal)