Himalaya Harbinger, Uttarakhand Bureau.
To take precautionary measures and ease daily commute, the Indian Institute of Technology Bombay (IIT-B) and India Meteorological Department (IMD) have collaborated to provide hyperlocal rain alerts, including details about possible waterlogging. The processed data will be accessible to citizens on mumbaiflood.in, which was launched on Friday.
IIT-Bombay’s artificial intelligence (AI)-based weather model is set up at the Regional Meteorological Centre in Colaba. It is connected to two doppler radars located at Colaba and Veravali in Goregaon East to provide access to the weather data. “The model will process the data and send out alerts every 15 minutes and update it every eight minutes,” said Professor Subimal Ghosh, convenor of the Centre for Climate Studies at IIT Bombay.
The system will forecast for 90 minutes in each area, compared to the current system of Nowcast by IMD, which predicts a lead time of three to four hours. “For instance, if particular areas in central Mumbai receive heavy rainfall, then an alert will be put out. The system will also flag the waterlogging areas,” added Ghosh.
The model issues alerts by analysing past rainfall movements. The current forecast the IMD gives is for 24 hours, and the nowcast is for 3-4 hours. This new system will predict every 90 minutes.
The ‘report flood in your area’ option allows users to select a location on the map and report and submit the water level as per the height – ankle, knee, or waist level. “The IMD and BMC can broadcast it and give alerts on time,” Ghosh said. “For now, we have focused on qualitative analysis of the rainfall, but soon we will also focus on quantitative analysis, which will give the exact millimetres of rainfall as expected,” he added.
Bengaluru
Mumbai is a vast city, and the weather phenomenon is different in each area, said Akshay Deora, research scientist at the Department of Meteorology, University of Reading, UK. “While the initiative is interesting, it should bring more accuracy to the rainfall predictions,” he said. “The hyperlocal alerts will also have to be verified. If it gives false alerts, then, in times of actual disasters, people will not take it seriously. This brings up the challenge of information dissemination. The data collected should reach the people in time.
”
