This development is a prime example of ‘AI for Good’, demonstrating how AI can be utilized to enhance public services and contribute to societal well-being.
In a significant move towards harnessing the power of Artificial Intelligence (AI) for the greater good, the India Meteorological Department (IMD) has begun to experiment with AI in improving nowcast and short-range weather forecasts. These are crucial predictions that span a duration from three hours to seven days.
While it’s expected to take another one to two years for the creation of some AI-powered products, the IMD has already initiated the groundwork. A group of scientists dedicated to AI research has been established, and strategic collaborations have been forged with institutions such as the Indian Institutes of Technology (IITs), universities, and some private stakeholders.
Mrutyunjay Mohapatra, Director General of Meteorology, IMD, shared insights about the potential of AI in weather forecasting. “There is scope to improve the utilization of AI in the forecasts generated for a few hours to seven days. AI can help improve the accuracy and quality of resolution to give village or panchayat-wise forecasts. Through higher resolution, one can view smaller areas,” he stated.
Presently, the IMD provides forecasts based on numerical weather prediction modelling. But as observational data and model data have expanded significantly over the years, there is a wealth of information that AI can tap into.
Observational data in India, available in digital form since 1901, includes parameters like temperature, wind, rain, and more. Satellite data is accessible from 1983 onwards, and radar data, which has seen an uptick in recent years, is available from 2002 onwards. This extensive data, which portrays the current atmospheric status and also offers future atmospheric characteristics produced by models, can be significantly utilized by AI to extract valuable insights.
According to the Met department, AI can be exceptionally useful in certain areas for weather forecasting. By studying previous climate patterns, AI can forecast events for nowcasting, where not much scientific intervention is required. This can be particularly effective when dealing with massive volumes of data where AI can help streamline the decision-making process.
However, Mohapatra clarified that AI would not replace numerical modelling, which remains the backbone of the forecast. Instead, AI will complement the modelling system, optimizing data utilization and aiding in more precise predictions. “The objective is to convert the data into value, just like converting petroleum into petrol,” he concluded.
