A groundbreaking machine learning-based forecasting system is set to transform short-term rainfall prediction in Mumbai. Developed by a research team from IIT Bombay, the model has been installed at the Regional Meteorological Centre in Colaba and is expected to be activated next week.
Once operational, the system will provide nowcast alerts every eight minutes, offering rainfall forecasts up to 90 minutes in advance. These updates will be published on the public portal Mumbaiflood.in, significantly enhancing the city’s preparedness for sudden downpours. The nowcasting tool has been designed using artificial intelligence, with radar data provided by the India Meteorological Department (IMD). Recent meteorological patterns—such as rainband movements, wind speeds, and cloud formation—are analysed through machine learning algorithms. Predictions are generated by examining these evolving weather systems to anticipate rainfall with greater accuracy.
The system’s development was spearheaded at IIT Bombay’s Centre for Climate Studies. The core model was created by Akshay Sunil, a postdoctoral researcher, while Professor Subimal Ghosh, the department’s head, guided the project as principal investigator. It was explained by the team that the model interprets short-term weather fluctuations using radar inputs and adapts to real-time changes, improving the precision of forecasts.
Funded by a banking institution, the initiative is being implemented in collaboration with the IMD and the Brihanmumbai Municipal Corporation (BMC). During preliminary trials, the forecasting system is reported to have shown promising accuracy in predicting rainfall patterns. Currently, the Mumbaiflood.in portal offers 24- to 48-hour hyperlocal forecasts, along with archived data on rainfall and water levels. It also features a citizen reporting mechanism for localised flooding, where users can share information on water levels and locations.
In parallel, IIT Bombay’s climate research centre is progressing with the development of a flood forecasting model based on hydrological simulations. This additional tool is expected to complement the nowcast system by offering predictive insights into potential flood events. With this AI-powered model, Mumbai is poised to strengthen its early warning systems and reduce the disruptive impact of unpredictable rainfall. The integration of real-time data analysis and machine learning in public weather services marks a first-of-its-kind advancement in India’s urban meteorological preparedness.