11 March 2026·PIB

PARLIAMENT QUESTION: AI IN WEATHER FORECASTING

In this note

1. At a Glance

India Meteorological Department (IMD) under Ministry of Earth Sciences (MoES) is operationalising Artificial Intelligence / Machine Learning models for cyclone intensity, nowcasting, hyper-local forecasts and monsoon prediction [1].
Significant for UPSC as it links Science & Tech (GS-III), Disaster Management (GS-III) and Governance/Mission Mausam themes; recurring Prelims fodder on indigenous models (BFS) and global AI models (Pangu, GraphCast, FourCastNet).

Lok Sabha Parliament Question answered on 11 March 2026 by Ministry of Earth Sciences detailing AI/ML applications in IMD weather forecasting [1].
Follows the rollout of Mission Mausam (approved 11 Sep 2024, ₹2,000 crore, 2-year window) and operationalisation of the Bharat Forecast System (launched 26 May 2025) [2][3].

IMD founded 1875; nodal Indian agency for meteorology under MoES.
NCMRWF (Noida) set up for medium-range numerical weather prediction; IITM (Pune) for tropical meteorology research [3].
2024: Union Cabinet approved Mission Mausam to make India "Weather-Ready & Climate-Smart" [2].
2025: Indigenous Bharat Forecast System (BFS) launched — world's highest-resolution (6 km) operational global NWP [3].
2026: Parliamentary disclosure of AI/ML integration across cyclones, monsoon, nowcasting [1].

Nodal Ministry: Ministry of Earth Sciences (MoES); implementing agency IMD with NCMRWF & IITM-Pune [1][3].
AI/ML tools used by IMD [1]:
AiDT (AI-based Advanced Dvorak Technique) — cyclone intensity estimation.
Pangu (Huawei), GraphCast (DeepMind), FourCastNet (NVIDIA) — data-driven experimental forecasts.
Hybrid AI-physics ensembles — sub-seasonal to seasonal (S2S) outlooks, ENSO-monsoon links.
Extended Indian Monsoon Data Assimilation integration.
Mission Mausam: ₹2,000 crore, approved 11 Sept 2024, 2-year horizon; lead institutes IMD + IITM + NCMRWF [2].
Bharat Forecast System (BFS): launched 26 May 2025; 6 km × 6 km grid; up to 10-day rainfall forecast; panchayat-level outputs; 30% improvement on extreme rainfall, 64% better in high-risk zones [2][3].
Supercomputers [3]:
Arka — IITM Pune, 11.77 PetaFLOPS, 33 PB storage.
Arunika — NCMRWF Noida, 8.24 PetaFLOPS, 24 PB storage.
Use-cases of AI [1]: nowcasting, bias-correction, hyper-local prediction, cyclone tracks, monsoon rainfall.

Scientific / Technological
Shift from purely physics-based NWP to hybrid AI-physics ensembles cuts runtime from ~12 h to 3–6 h [3].
Adoption of foreign AI models (Pangu, GraphCast, FourCastNet) experimentally; indigenous BFS for operational runs [1][3].
Economic
Better monsoon & extreme-event forecasts critical for agriculture (kharif sowing), insurance (PMFBY), power demand and aviation [2].
₹2,000 crore Mission Mausam funds R&D, observation network and HPC [2].
Environmental / Disaster Management
Sharper cyclone tracking (AiDT) and nowcasting of cloudbursts critical for NDMA/SDMA early warnings [1].
Administrative / Governance
Virtual centre spanning IMD-NCMRWF-IITM integrates AI/ML into operational chain (bias correction, downscaling, multi-source fusion) [2].
Panchayat-level forecasts align with cooperative federalism for last-mile delivery [2].
Strategic / Self-Reliance
BFS makes India only country running global NWP at 6 km resolution in real-time — Atmanirbhar Bharat in Earth-system sciences [3].

Sept 2024 — Cabinet approval of Mission Mausam (₹2,000 cr) [2].
26 May 2025 — Launch of Bharat Forecast System by Dr. Jitendra Singh [3].
2025 — Commissioning of supercomputers Arka (Pune) & Arunika (Noida) [3].
11 March 2026 — Parliament reply detailing AI/ML models (AiDT, Pangu, GraphCast, FourCastNet) in IMD operations [1].

AI/ML cyclone intensity tool used by IMD: AiDT (AI-based Advanced Dvorak Technique) [1].
Foreign AI forecast models adopted experimentally by IMD: Pangu, GraphCast, FourCastNet [1].
Implementing ministry of Mission Mausam: Ministry of Earth Sciences (not MoEFCC, not MeitY) [2].
Mission Mausam outlay: ₹2,000 crore, duration 2 years, approved 11 Sep 2024 [2].
Bharat Forecast System resolution: 6 km × 6 km — highest globally for an operational model [3].
BFS launch date: 26 May 2025 [3].
Supercomputer Arka is housed at IITM-Pune (11.77 PF); Arunika at NCMRWF-Noida (8.24 PF) [3].
BFS delivers forecasts at panchayat level up to 10 days [2].
Hybrid AI-physics ensembles incorporate ENSO–monsoon links for S2S forecasts [1].
IMD established in 1875; MoES is parent ministry [1].

GS-III — Science & Technology / Disaster Management: indigenisation of weather tech; AI for early warning.
GS-III — Agriculture: monsoon prediction impact on cropping & food security.
Plausible stems: 1. "Discuss how AI/ML integration in IMD operations under Mission Mausam can transform India's disaster preparedness." (GS-III) 2. "Evaluate the significance of the Bharat Forecast System for India's agrarian economy." (GS-III) 3. "Examine the trade-offs between adopting global AI weather models and building indigenous numerical prediction capacity." (GS-III)

Mission Mausam — parent umbrella programme [2].
National Monsoon Mission — predecessor S2S programme.
IndiaAI Mission (MeitY, 2024) — horizontal AI policy linkage.
NDMA & Cyclone preparedness — downstream user of AiDT.
PMFBY (crop insurance) — consumer of monsoon forecasts.
ISRO INSAT-3DS (2024) — satellite feeding IMD data assimilation.
ENSO / IOD phenomena — physical basis of monsoon forecasting.
PARAM Siddhi / National Supercomputing Mission — HPC backbone overlap with Arka/Arunika.

Ministry confusion: IMD/Mission Mausam are under MoES, NOT MoEFCC or MeitY.
GraphCast (Google DeepMind), Pangu (Huawei), FourCastNet (NVIDIA) — easy to swap originators in MCQs [1].
BFS vs Mission Mausam: BFS is a model (NWP), Mission Mausam is the programme.
Arka ≠ Arunika: Arka at IITM-Pune; Arunika at NCMRWF-Noida [3].
IMD established 1875, not 1947 or 1950.