The Bharat Forecast System (BharatFS) is India’s high-resolution global weather model, built by IITM Pune and launched on 26 May 2025. This lesson covers who built it, how its 6 km grid works, the supercomputers behind it, and where it stands in October 2026.
Contents
Must know: what BharatFS is
What is the Bharat Forecast System?
The Bharat Forecast System is a computer weather model. It takes today’s weather readings and calculates how the atmosphere will change. The result is a forecast for the next ten days.
When was it launched?
- The Ministry of Earth Sciences (MoES) launched BharatFS on 26 May 2025.
- The launch ceremony was held at Vigyan Bhawan, New Delhi.
- Dr Jitendra Singh dedicated it to the nation. He is Minister of State (Independent Charge) for Earth Sciences.
- The ministry called it one of the world’s first indigenously developed high-resolution forecast systems.
What does it predict?
- It gives 10-day forecasts of rainfall, temperature and low-pressure genesis.
- The output grid is about 6.5 km wide, per a January 2026 reply. The system is usually called 6 km.
- Its results appear on the NWP page of the India Meteorological Department (IMD).
Sources: PIB launch; PIB reply; MoES reply.
Who built the Bharat Forecast System?
The answer is IITM Pune, a research institute and not an IIT.

Which institutions were involved?
- The Indian Institute of Tropical Meteorology (IITM), Pune developed it.
- IITM is an institute of the Ministry of Earth Sciences, per the launch release.
- NCMRWF, Noida, and IMD supported the work.
- IITM says it was developed by IITM and operationalised by IMD.
What else did the launch release say?
- Four women scientists spearheaded the effort. The release does not name them.
- The release called the system an Atmanirbhar Bharat and Make in India milestone.
- It said other tropical regions could also benefit.
Sources: PIB launch; PIB reply; IITM newsletter.
Good to know: the grid and the supercomputers
How does the 6 km grid work?
A weather model cuts the atmosphere into small boxes. It calculates the weather inside each box. Smaller boxes mean finer detail.
What changed from the older model?
- The older global model (GFS T1534) had a grid of about 12 km.
- BharatFS runs at about 6 km.
- The ministry says other operational global centres use 9 to 14 km.
- The ministry also says India is the only country running a global model at this resolution for real-time use. This is a ministry claim.
What is the TCo grid?
- BharatFS uses a Triangular Cubic Octahedral (TCo) grid.
- This design lets the model reach about 6 km over the whole globe.
- The model also accounts for India’s geography, such as the Himalayas and the Western Ghats.
How good is it?
- The ministry says extreme rainfall forecasts improved by about 30% in research mode.
- At launch, the minister cited a 64% gain in “core zones”.
- Later official replies say “significant improvement” over the core monsoon region. An August 2025 reply also gives about 30% better accuracy for extreme rainfall. Treat 64% as a launch-day figure.
Sources: PIB reply; PIB Aug 2025; PIB launch.
Which supercomputers power it?
A 6 km model needs huge computing power. Two machines made it practical.

What are Arka and Arunika?
- Arka is at IITM Pune. It has 11.77 petaflops.
- Arunika is at NCMRWF Noida. It has 8.24 petaflops.
- The Prime Minister inaugurated both in September 2024. The cost was Rs 850 crore.
- A separate 1.9 petaflop AI and machine-learning system was added.
- Arka means the Sun. Arunika means the glow of dawn.
What difference did they make?
- Runtime fell from about 12 hours to 3 to 6 hours. IITM’s own newsletter says 13 hours to 4.
- That speed allows real-time forecasting.
- The ministry says its computing grew from 6.8 petaflops in 2014 to 22 in 2024. It reached 28 in 2025 (22 petaflops at the 2024 inauguration, 28 per a March 2026 reply).
Sources: PIB Sep 2024; PIB reply; MoES reply; PMO release.
Great to know: uses, radars and status
Why does finer detail help?
A coarse grid can miss a small storm. A finer grid can see it.

What does 6 km mean on the ground?
- The ministry says 6 km is about the size of a cluster of panchayats or villages.
- The aim is forecasts at panchayat-cluster level.
- The July 2026 reply says BharatFS serves block levels and then panchayat levels.
Who uses such forecasts?
- Farmers use them for crop planning, irrigation and harvesting.
- Water authorities can manage reservoirs better in the monsoon.
- Disaster teams use them to prepare for extreme rainfall.
Sources: PIB reply; PIB launch; MoES Jul 2026; PIB Aug 2025.
What do radars and Mission Mausam add?
BharatFS is one part of a bigger push. That push is Mission Mausam.
What is Mission Mausam?
- The Prime Minister launched it on 14 January 2025. A December 2024 reply gave September 2024.
- Its goal is a “Weather-ready and Climate-smart” nation.
- At the BharatFS launch, the minister gave its budget as Rs 2,000 crore.
How many radars does India have?
- India has 50 Doppler Weather Radars as of mid-2026.
- Older counts were 39 in December 2024 and 48 in February 2026.
- Another 50 are proposed, taking the total to nearly 100 within two years.
- An August 2025 reply says radar data mainly help nowcasts. A December 2024 reply says radar data are fed into weather models.
What are nowcasts?
- A nowcast is a very short-range alert, such as for thunderstorms.
- A 2025 reply says radar-based thunderstorm warnings cover up to 3 hours.
- IMD nowcasts for heavy rain cover up to 6 hours.
Sources: PIB Aug 2025; MoES Feb 2026; MoES Jun 2026; MoES Jul 2026; PIB launch; PIB Dec 2024.
Where does BharatFS stand in 2026?
Since May 2025 IMD has used the model for operational forecasts.
Is it operational?
- Yes. A March 2026 reply says it was made operational in May 2025.
- IMD uses it to issue operational forecasts.
- The July 2026 reply calls it operational at 6 km.
- Its public outputs appear on IMD’s NWP site. They are not sent directly to stakeholders.
What other tools sit beside it?
- The older GFS and NCUM models run at 12 km. They have been operational since 2018.
- Mausamgram gives village-level forecasts. They are hourly for 36 hours, three-hourly for 5 days and six-hourly up to 10 days.
- IMD developed the Panchayat Mausam Seva portal with the Panchayati Raj ministry.
- On 12 May 2026 the minister launched an AI-driven monsoon advance forecast. It also launched a 1 km rainfall pilot for Uttar Pradesh.
What is still unclear?
- The 64% figure is the minister’s launch-day claim.
Sources: MoES Mar 2026; MoES Jan 2026; MoES Jul 2026; PIB May 2026.
How do the dates fit together?
Use this order to keep the dates straight.
- Sep 2024: Arka and Arunika inaugurated.
- 14 Jan 2025: Mission Mausam launched.
- 26 May 2025: BharatFS launched at Vigyan Bhawan.
- 2025: MoES computing reaches 28 petaflops.
- 12 May 2026: AI monsoon advance forecast launched.
- Jun 2026: 50 Doppler radars, nearly 100 planned.
How can you remember it?
- Computers came first, then the mission, then the model.
- 6 km is half of the old 12 km.
- IITM Pune builds. IMD runs.
Sources: PIB Sep 2024; PIB launch; MoES Jun 2026.
Check your understanding
Review the feedback after each answer. The understanding-check score lasts for this page session.
Reset the checks when you want another attempt. This score is separate from the Native Practice score below.
This score resets when the page reloads.
Questions to think through
Why does a finer grid need a faster computer?
Smaller boxes mean many more boxes to calculate. Arka and Arunika cut the runtime to 3 to 6 hours.
Why should you attribute the 64% figure?
The minister cited it at launch for “core zones”. Later official replies dropped it and said “significant improvement”, with about 30% better extreme-rainfall accuracy.
Why is IITM Pune the answer, not an IIT?
IITM is the Indian Institute of Tropical Meteorology. It is a Ministry of Earth Sciences institute and is not an IIT.
Why does the radar count need a date?
The count grew from 39 to 50 in about 18 months. Always say “as of” a month.
Native Practice
Apply the developer, grid size, supercomputers, radars and current status. Read the connected explanation after each answer.
0 of 8 answered
Question 1 of 8
Question 2 of 8
Question 3 of 8
Question 4 of 8
Question 5 of 8
Question 6 of 8
Question 7 of 8
Question 8 of 8
Leave a Reply