The India Meteorological Department (IMD) plans to employ artificial intelligence (AI) and machine learning (ML) to ensure more accurate weather forecasts.
On Sunday, IMD Director General Mrutyunjay Mohapatra said in an official statement that the usage of artificial intelligence can enhance weather forecasting, especially for issuing nowcasts, which can help improve 3-6 hours prediction of extreme weather events. He reiterates that while artificial intelligence and machine learning in weather forecasting are not as prevalent as in other fields, this step will ensure new beginnings in this area.
According to Mohapatra, the Ministry of Earth Sciences has asked research organizations to examine how artificial intelligence (AI) can be leveraged for fine-tuning weather forecasting processes. He also stated that the IMD intends to collaborate on this idea with the Indian Institutes of Information Technology (IIITs) at Prayagraj and Vadodara and IIT-Kharagpur for the technology upgrade. The IMD has also teamed up with Google to provide precise short-term and long-term weather forecasts.
The IMD issues forecasts for extreme weather, such as thunderstorms and dust storms. However, thunderstorms, which bring lightning and torrential rains, are more difficult to anticipate than cyclones because of their extreme weather phenomena that form and disperse in a relatively short amount of time.
At present, IMD uses radars, satellite imagery, and other tools to issue nowcasts, which help in offering information on extreme weather events occurring in the next 3-6 hours. The satellite imagery comes from the INSAT series of geosynchronous satellites, as well as the Real-Time Analysis of Products and Information Dissemination (RAPID), a weather data explorer application that serves as a gateway and provides quick interactive visualization as well as 4-Dimensional analysis capabilities.
It also relies on ISRO for ground-based observations from the Automatic Weather Stations (AWS), the Global Telecommunication System (GTS) that measure temperature, sunshine, wind direction, speed, and humidity. All these data include cloud motion, cloud top temperature, and water vapor content, all of which aid in rainfall estimation, weather forecasting, and cyclone genesis and direction.
Despite the vast dataset, the weather is a dynamic phenomenon i.e. temperature, wind speed, tides, etc., are never constant. Further, the rising global warming and greenhouse effect have introduced more unpredictability of changes in oceanic and wind activity that are key parameters in weather forecasting.
The pattern-recognition capabilities of artificial intelligence and machine learning make them suitable assets in weather prediction and forecasting. AI models are fed with enormous quantities of weather data and trained to identify a storm that can bring lightning or tornadoes. This enables the models to predict the possibility of the occurrence of thunderstorms or any other meteorological event using weather and climate datasets. Now meteorologists can make predictions with improved accuracy and thus save lives and money.
Also, these artificial intelligence models depend on the computational power of supercomputers for large-scale data processing and pattern recognition. As per IMD, an increase in computing power, from 1 teraflop to 8.6 teraflops, has helped the nodal department in better processing observational data.