Parliament Question: Accuracy of weather forecasting systems

GOVERNMENT OF INDIA MINISTRY OF EARTH SCIENCES RAJYA SABHA UNSTARRED QUESTION NO. 1480 ANSWERED ON 12/02/2026 ACCURACY OF WEATHER FORECASTING SYSTEMS 1480. SHRI S. SELVAGANABATHY: Will the Minister of EARTH SCIENCES be pleased to state: (a) the latest assessment of accuracy of India Meteorological Department's (IMD) operational forecasts for southwest monsoon during the current year, along with year- on-year improvements achieved since introduction of multi-model ensemble forecasting; (b) the details of tools, models and observation systems presently in use for seasonal and short range weather prediction including high-performance computing facilities and AI/ML enabled forecasting systems; and (c) the status of expansion of Doppler Weather Radar (DWR) network across the country and the extent to which this has strengthened forecasting for severe events such as cloudbursts, thunderstorms, lightning and cyclones? ANSWER THE MINISTER OF STATE (INDEPENDENT CHARGE) FOR MINISTRY OF SCIENCE AND TECHNOLOGY AND EARTH SCIENCES (DR. JITENDRA SINGH) (a) The India Meteorological Department (IMD) has been following a seamless forecasting strategy for monsoonal rainfall. As per this strategy, it issues forecasts and warnings on different time scales and for different spatial scales. Nowcasting- up to six hours for all types of severe weather at all districts and around 1200 stations. Short to medium range (up to 7 days) forecasts for rainfall over cities, blocks, districts, and meteorological subdivisions. Extended range (up to 4 weeks) forecasts for 36 meteorological sub- divisions. Monthly and seasonal long-range forecasts for rainfall for the whole country and for homogenous region. The latest assessment of the accuracy of its seasonal long range for Southwest Monsoon in the current year 2025 shows it was highly accurate and the forecast, issued in April 2025, for the southwest monsoon (June-September) rainfall over the country as a whole was 105% of long period average (LPA) while the actual season rainfall for the country as a whole was 108 % of LPA and it was within errors range of the forecast issued. The spatial probability forecasts were also largely accurate across most regions of the country. Similarly, the monthly rainfall forecasts closely matched the observed values and remained within the forecast limits. The latest assessment of Heavy rainfall Forecast Performance shows in 2025, the heavy rainfall forecast demonstrated high skill, with Probability of Detection of 0.85, indicating it was in overall accuracy.IMD has adopted a new strategy for monthly and seasonal forecasting since 2021 based on the Multi-Model Ensemble (MME) approach. The strategy utilizes coupled global climate models (CGCMs from various global climate prediction and research centers, including IMD's Monsoon Mission Climate Forecasting System (MMCFS). The performance of IMD's seasonal forecasting system has shown improvement following the adoption of the MME-based approach. The verification details of IMD's seasonal forecasts for All India Summer Monsoon Rainfall for the period 2021 to 2025 are given below: ALL India Monsoon Rainfall (LPA) Year Actual Forecast Remark (%) (%) 2021 99 101 Accurate 2022 106.5 103 Accurate 2023 95 96 Accurate 2024 108 106 Accurate 2025 108 106 Accurate ***Model error ± 4% of LPA (b) For seasonal and short-range weather prediction, IMD uses a range of advanced tools, models, and observation systems as part of its operational forecasting framework. Under the Mission Mausam project, already Bharat Forecast System (BharatFS), an advanced computer simulation model has been developed, and it has been operational at a very high spatial resolution of 6 km. It has also a capability to provide predictions of rainfall events up to 10 days, covering the short and medium range. Further, to support such high-resolution models running regularly, the Computing facilities have also been substantially increased to integrate voluminous data and to run meso-scale, regional, and global models at higher resolution. Recently, with the implementation of the High Power Computing Systems "Arunika" and "Arka", the Ministry of Earth Sciences has enhanced its total computing power to 28 Peta FLOPS in 2025, a substantial increase from the previous capacity of 6.8 Peta FLOPS in 2014. IMD is gradually integrating artificial intelligence (AI) and machine learning (ML)– based methods to enhance model performance, post-process model outputs, pattern recognition, bias correction, and probabilistic forecast interpretation. The weather observation system presently consists of 48 Doppler Weather Radars (DWRs) covering nearly 92% of the country, along with high-resolution satellite-based monitoring and around 6,300 rain gauge stations. (c) There are a total of 48 DWRs installed and operational in India. The locations where the DWR network has been established across the country are given in Annexure-1. This has helped IMD improve monitoring and forecasting of severe events such as cloudbursts, thunderstorms, lightning, and cyclones.Annexure-1 Locations of the current Doppler Weather Radar (DWR) network in the country: S. No. State Location 1. Andhra Pradesh Machilipatnam 2. Andhra Pradesh Visakhapatnam 3. Andhra Pradesh Sriharikota, ISRO 4. Assam Mohanbari 5. Assam Jorhat 6. Bihar Patna 7. Chhattisgarh Raipur 8. Goa Goa 9. Gujarat Bhuj 10. Himachal Pradesh Jot 11. Himachal Pradesh Murari Devi 12. Himachal Pradesh Kufri 13. Karnataka Mangalore 14. Kerala Kochi 15. VSSC, Thiruvananthpuram Kerala (ISRO) 16. Madhya Pradesh Bhopal 17. Madhya Pradesh Silkheda (IITM) 18. Maharashtra Mumbai 19. Maharashtra Nagpur 20. Maharashtra IITM Solapur 21. Maharashtra Veravali 22. Maharashtra Mumbai, Juhu (IITM) 23. Maharashtra Mumbai, Panvel (IITM) 24. Mumbai, Kalyan, Dombivali Maharashtra (IITM) 25. Maharashtra Mumbai, Vasai, Virar (IITM) 26. Maharashtra Mahabaleshwar (IITM) 27. Meghalaya Cherrapunji (ISRO) 28. Odisha Gopalpur29. Odisha Paradip 30. Punjab Patiala 31. Rajasthan Jaipur 32. Tamil Nadu Chennai 33. Tamil Nadu Karaikal 34. Tamil Nadu NIOT Chennai 35. Telangana Hyderabad 36. Tripura Agartala 37. Uttarakhand Lansdowne 38. Uttarakhand Mukteshwar 39. Uttarakhand Surkanda Devi 40. Uttar Pradesh Lucknow 41. West Bengal Kolkata 42. Jammu & Kashmir Banihal Top 43. Jammu & Kashmir Jammu 44. Jammu & Kashmir Srinagar 45. Delhi Ayanagar 46. Delhi Palam 47. Delhi HQ Mausam Bhawan 48. Ladakh Leh ********