India is significantly strengthening its weather forecasting capabilities through the integration of artificial intelligence technologies, high performance computing infrastructure and advanced modelling systems under Mission Mausam, according to information presented in the Lok Sabha by Jitendra Singh.
The India Meteorological Department, working in coordination with various institutions under the Ministry of Earth Sciences, has begun deploying artificial intelligence and machine learning technologies to enhance the accuracy, speed and resolution of weather forecasts across the country. These efforts are aimed at improving early warning systems for extreme weather events and supporting sectors such as agriculture, disaster management, aviation, fisheries and energy.
One of the key applications of artificial intelligence in forecasting is the use of an AI and machine learning based Advanced Dvorak Technique to estimate the intensity of tropical cyclones. This technology helps meteorologists assess cyclone strength more accurately using satellite data, enabling better monitoring and forecasting of storm systems over the oceans.
The meteorological system is also using advanced AI driven weather prediction models including Pangu Weather, GraphCast and FourCastNet. These data driven models use deep learning techniques to analyse vast atmospheric datasets and generate experimental forecasts that complement traditional numerical weather prediction models.
Artificial intelligence is also being used to enhance nowcasting capabilities, improve bias correction in forecast models and generate hyper local predictions. These technologies are helping improve the accuracy of cyclone track predictions, monsoon rainfall forecasts and short term weather outlooks.
Seasonal forecasting has also benefited from hybrid AI physics ensemble models that combine artificial intelligence with physical climate models. These systems help improve sub seasonal to seasonal predictions by incorporating complex climate relationships such as the link between the El Nino Southern Oscillation and the Indian monsoon. The models also integrate data from the extended Indian Monsoon Data Assimilation and Analysis reanalysis system to refine predictions.
At the National Centre for Medium Range Weather Forecasting, several global AI foundation models including Pangu Weather, GraphCast, FourCastNet and GenCast are being integrated into operational forecasting systems. These models run on the Arunika supercomputer and are initialized using outputs from the centre’s Mithuna coupled forecasting system.
The integration of these models allows meteorologists to produce rapid medium range forecasts, generate probabilistic predictions of extreme weather events such as heavy rainfall and heatwaves and downscale forecasts to block level resolution. This level of detail is expected to significantly improve decision making at the local administrative level.
Artificial intelligence is also playing an important role in weather information dissemination. The meteorological department has adopted an AI enabled language translation tool known as Bhashini to deliver weather related information to farmers in multiple regional languages. This initiative ensures that critical weather advisories reach farmers in a format they can easily understand.
Weather based crop advisory services supported by these technologies provide farmers with real time information on weather conditions, crop health and recommended agricultural practices. These advisories help farmers plan irrigation, fertilizer application, pest control and harvesting activities more efficiently, thereby improving yields and reducing losses.
To strengthen last mile communication, real time weather updates and early warnings are disseminated to farmers in climate vulnerable districts through multiple communication channels. These include print media, electronic media, television broadcasts on Doordarshan, internet platforms and SMS based alerts under public private partnership initiatives.
The meteorological department has also launched Panchayat level weather forecasts covering nearly all Gram Panchayats across India. These forecasts are accessible through digital platforms such as the e Gramswaraj portal, the Meri Panchayat mobile application, the e Manchitra platform and the Mausamgram portal developed by the meteorological department.
In addition, an artificial intelligence based rainfall forecasting tool called meteoGAN has been developed to generate area specific rainfall predictions with a spatial resolution of approximately 300 meters. This technology enables highly localized weather forecasts that are particularly useful for agriculture and disaster preparedness.
Farmers are also able to access location specific forecasts and agrometeorological advisories through mobile applications such as Meghdoot and Mausam. These services are further supported by social media dissemination through platforms including WhatsApp and Facebook, which help expand outreach among rural communities.
The meteorological department has integrated its forecasting services with the information technology platforms of 21 state governments. As a result, approximately 15.6 million farmers are currently accessing weather related information through these state platforms in both English and regional languages.
In parallel with artificial intelligence adoption, India has expanded its high performance computing capabilities under Mission Mausam. As part of the programme, advanced supercomputing systems named Arka and Arunika were inaugurated on 26 September 2024 at the Indian Institute of Tropical Meteorology in Pune and the National Centre for Medium Range Weather Forecasting in Noida.
The Arka system has a computing capacity of 11.77 petaflops, while the Arunika system has a capacity of 8.24 petaflops. In addition, a dedicated artificial intelligence and machine learning system with a computing capacity of 1.9 petaflops has been deployed. Together, these systems have increased the Ministry of Earth Sciences’ total computing capacity to 21.91 petaflops.
This expanded computational infrastructure enables the development of high resolution weather and climate models, large scale data processing and more advanced applications of artificial intelligence in forecasting systems.
These technological improvements have already led to measurable gains in forecasting accuracy. Between 2021 and 2025, tropical cyclone track forecast errors were reduced by 5 to 10 percent for lead times up to 48 hours and by 20 to 25 percent for longer lead times. Forecasting of cyclone intensity has also improved significantly, with accuracy increasing by about 33 to 35 percent for lead times up to 72 hours.
Landfall prediction accuracy has also improved considerably. The average 24 hour landfall forecast error declined from 31.9 kilometres during the period 2016 to 2020 to 19 kilometres during 2021 to 2025. Similarly, the 48 hour landfall forecast error declined from 61.5 kilometres to 34.4 kilometres over the same period.
Heatwave forecasting capabilities have also advanced, with warnings now issued four to five days in advance. This lead time allows state and district administrations to implement heat action plans and protect vulnerable populations.
The meteorological department has also strengthened forecasting for monsoon rainfall and heavy precipitation events. Forecasts are issued at multiple time scales, including nowcasts for up to six hours, short and medium range forecasts up to seven days and extended range forecasts up to four weeks.
Seasonal forecasts for the southwest monsoon have also demonstrated improved accuracy. In 2025, the forecast issued in April predicted monsoon rainfall at 105 percent of the long period average, while the actual rainfall recorded for the season was 108 percent of the long period average, well within the forecast range.
Heavy rainfall forecasting has also shown strong performance, with a probability of detection of 0.85 in 2025, indicating a high level of forecast skill. Thunderstorm forecasting has improved as well, with the probability of detection for three hourly nowcasts rising to 0.92 in 2025 compared with 0.83 in 2022.
These improvements in forecasting accuracy have delivered significant socio economic benefits. Early warnings for cyclones, heavy rainfall, heatwaves and thunderstorms have enabled timely evacuations and improved disaster preparedness, reducing casualties and economic losses.
The benefits are particularly evident in cyclone related disaster management. While the 1999 Odisha super cyclone resulted in the loss of around 7000 lives, recent tropical cyclones have caused fewer than 100 deaths across affected regions due to improved forecasting and early warning systems.
Accurate forecasting of a single cyclone is estimated to save approximately ₹1100 crore by reducing costs associated with evacuation operations, compensation payments and damage to infrastructure and economic sectors such as power, marine transport, aviation and railways.
Weather forecasting improvements have also supported agricultural planning. District and block level forecasts allow farmers to make informed decisions on sowing, irrigation, fertilizer use and harvesting, reducing crop losses and improving farm productivity.
Marine forecasts have improved safety for fishermen by providing timely warnings of adverse ocean conditions. These alerts help prevent accidents at sea and protect fishing vessels and livelihoods.
Impact based forecasting systems now provide colour coded warnings and risk assessments at district and sub city levels. These warnings allow local administrations to take preventive measures and manage disaster response more effectively.
Mission Mausam has been designed as a multi phase programme aimed at transforming India into a weather ready and climate smart nation. The initiative focuses on expanding the national weather observation network, improving high resolution modelling systems and strengthening artificial intelligence applications in forecasting.
Future phases of the mission will further enhance observation infrastructure through additional automated weather stations, radars and rainfall monitoring systems. The programme will also strengthen real time monitoring, forecasting and early warning systems supported by high performance computing and advanced data integration technologies.
Officials stated that continuous modernization of forecasting infrastructure, combined with artificial intelligence technologies and expanded dissemination systems, will further improve the accuracy and accessibility of weather services in India. These advances are expected to enhance disaster resilience, support agricultural planning and strengthen climate preparedness across the country.
