India to validate its own AI model for weather forecasting after monsoon

The evaluation will compare AI-based predictions with actual rainfall and climate data, India Meteorological Department (IMD) director general Mrutyunjay Mohapatra told Mint. If successful, the model could improve early warnings, disaster preparedness, and agricultural planning nationwide, he said, adding that AI systems would analyze large historical weather datasets and real-time observations to detect patterns.

The initiative will complement India’s Bharat Forecasting System (BharatFS)—a high-resolution weather prediction system launched in May 2025, operating on a 6-km grid, designed to improve forecasts for monsoon, cyclones and extreme weather events. A high-resolution model can provide better guidance for weather predictions and extremes at the district and block levels, which are typically 12 km or more.

Predicting the behaviour of the southwest monsoon is a major challenge for weather scientists due to the complex interaction of global and regional climatic factors. Climate patterns such as the El Niño–Southern Oscillation, the Indian Ocean Dipole and the Madden–Julian Oscillation play a crucial role in shaping rainfall trends over India. Changes in sea surface temperatures in the Indian Ocean and the Pacific Ocean can significantly alter wind patterns and moisture flow towards the subcontinent.

“In addition, local weather systems and growing climate variability have made it increasingly difficult to accurately predict the distribution and intensity of rainfall well in advance. We have been using AI-based models in weather forecasting for some time now,” Mohapatra said. “However, compared to heatwave forecasts, predicting the monsoon is far more complex. Therefore, we plan to validate the indigenous artificial intelligence weather forecasting model after the monsoon season to assess its accuracy and reliability.”

While forecasting models have improved over the years, experts say predicting the exact behaviour of the monsoon across different regions of India remains a complex scientific task.

BharatFS uses multiple numerical weather prediction models to improve the accuracy of forecasts. These include global models such as the Global Forecast System, Global Ensemble Forecast System (GEFS), Ensemble Weather Forecast, Weather Research and Forecasting (WRF) model and regional models like Regional Unified Ensemble Systems, WRF (Weather Research and Forecasting) Forecast System to forecast weather conditions.

“It’s a fantastic move by the IMD. What we now need is to install more sensors and monitoring stations across the country. The systems must also be better calibrated and standardised to ensure greater accuracy of data,” said Deepak Maheswari, a digital policy expert and senior policy advisor, Centre for Social and Economic Progress (CSEP)-an independent, public policy think tank. “At the same time, strong connectivity is essential so that the data collected from these stations can be seamlessly fed into a centralised system for real-time analysis and improved forecasting.”

According to environmentalists, better forecasts also help farmers plan agricultural activities, improve water management, and support disaster preparedness. " This would help provide early warnings for extreme events such as floods, cyclones and heatwaves, reducing loss of life and property," said Paras Tyagi, president, Centre for Youth, Culture, Law & Environment, an NGO.