Bharat Forecast System (BFS): India’s Panchayat-Level Weather Model
Bharat Forecast System (BFS) is India's 6-km resolution weather model launched 26 May 2025, developed by IITM Pune on the Arka supercomputer.
The Bharat Forecast System (BFS) is India’s highest-resolution operational weather model, launched on 26 May 2025 by the Ministry of Earth Sciences. The system was developed by the Indian Institute of Tropical Meteorology (IITM), Pune, and runs on the Arka supercomputer. With a 6-kilometre grid spacing across the country, the Bharat Forecast System BFS produces forecasts at finer spatial detail than any other national meteorological agency runs operationally.
The headline capability is panchayat-cluster forecasts. Earlier-generation models from the India Meteorological Department operated at 12-km resolution, which meant that forecasts effectively spoke to administrative blocks. BFS pushes the grid down to a level where individual panchayat clusters can be served, which has direct relevance for agriculture, disaster early warning, and energy planning.
What Makes BFS Different
Three engineering choices set BFS apart from its predecessors.
- A 6-km grid — the finest resolution operating at national scale globally
- The Arka supercomputer — 11.77 petaflops of compute and 33 petabytes of storage
- Integration with a national Doppler radar network — 40 radars currently, expanding toward 100
The combination matters. A finer grid alone is useless without compute that can run the model in a tight enough cycle to be operational. Arka was installed at IITM Pune specifically to make the BFS resolution viable in production.
The Arka Supercomputer
Arka is a high-performance computing system commissioned for atmospheric and ocean modelling. With a peak performance of 11.77 petaflops and 33 petabytes of storage, it is purpose-built for the BFS workload. Forecasts are produced through repeated model runs across the day, with shorter-range “nowcasts” updated at fine temporal intervals.
Why IITM Pune
The Indian Institute of Tropical Meteorology in Pune is the home of India’s tropical-weather modelling effort. IITM developed BFS in-house, with model physics, dynamical-core tuning, and assimilation systems designed specifically for the Indian monsoon system. The work is not a foreign import. It is also worth clarifying that BFS was developed at IITM Pune and not at IIT Delhi — the institutional confusion comes from the similar acronyms.
Capabilities of BFS
The Bharat Forecast System covers the full spectrum of weather forecasting timescales relevant to India.
Panchayat-Cluster Forecasts
The 6-km resolution means BFS can issue forecasts at the level of clusters of panchayats. For agriculture extension and disaster preparedness, this is a structural shift. A panchayat can know whether a rain band is expected to track over its fields or miss it by 20 kilometres.
Cyclone and Heavy-Rain Prediction
BFS has demonstrated a 64% improvement in accuracy in high-risk zones compared to the previous-generation models. For extreme-rainfall events, the improvement is around 30%. Both metrics matter because cyclones, depressions, and convective storms create the majority of India’s weather-related losses.
Nowcasting Through Radar Integration
BFS ingests data from 40 Doppler weather radars across the country, with the network expanding to 100. The radar data feeds short-term forecasts of 2 hours and shorter — the so-called nowcasts — that are critical for urban flood warnings, aviation, and event management.
How BFS Fits Into India’s Weather Architecture
The Ministry of Earth Sciences runs a layered observation and forecasting system. At one end are the satellites — INSAT-3D, INSAT-3DR, INSAT-3DS — operated with ISRO. At the surface are automatic weather stations, raingauges, and Doppler radars under the India Meteorological Department. In the middle are the numerical models — BFS at IITM Pune and the global model run by the National Centre for Medium Range Weather Forecasting (NCMRWF).
BFS sits as the high-resolution regional model in this stack. Its outputs flow into the IMD’s forecasting workflow, the Krishi Vigyan Kendra advisory system for farmers, and the NDMA’s early-warning pipeline. For related coverage on AI-driven monsoon forecasting and crop planning, see AI-based local monsoon forecasting supports kharif planning.
Use Cases on the Ground
The fine-grained forecasts have several practical applications:
- Farmer advisories tailored to a panchayat cluster rather than a district average
- Urban flood warnings triggered hours earlier through nowcasting
- Hydropower and wind-energy scheduling that follows local cloud and wind patterns
- Aviation and airport operations during convective seasons
- Disaster-response staging during cyclones and depressions
Why BFS Matters for Policy
A weather model is not just a science deliverable. It is infrastructure that other systems depend on. Agriculture, irrigation, disaster management, energy, and aviation are all downstream consumers of forecast data. The shift from 12-km to 6-km resolution is therefore a multiplier for every policy that depends on accurate, localised weather information.
There are also implications for climate-change adaptation. As extreme-weather events become more frequent, a finer model that can resolve localised heavy-rainfall events gives state governments better data to plan flood-resilient infrastructure, urban drainage, and crop-insurance products. BFS will also feed sub-seasonal and seasonal forecasting research that informs cropping-pattern advisories.
Scaling and Maintenance
Running a 6-km grid model operationally is computationally demanding. The Arka supercomputer has the headroom for the current configuration, but future improvements — assimilating additional satellite channels, running ensembles, embedding machine-learning emulators — will need continuous upgrades. The success of BFS therefore depends on sustained investment in the underlying compute and observation network, not just on the model code itself.
FAQs
When was the Bharat Forecast System launched?
The Bharat Forecast System was launched on 26 May 2025 by the Ministry of Earth Sciences.
Who developed the Bharat Forecast System?
BFS was developed by the Indian Institute of Tropical Meteorology (IITM), Pune. It was developed at IITM Pune and not at IIT Delhi.
What resolution does BFS operate at?
BFS operates at 6-kilometre horizontal grid resolution, the highest operational resolution at national scale globally.
What supercomputer powers BFS?
BFS runs on the Arka supercomputer, with 11.77 petaflops of compute and 33 petabytes of storage, installed at IITM Pune.
How is BFS different from the earlier IMD model?
The previous-generation operational model ran at around 12-km resolution and spoke effectively to administrative blocks. BFS halves the grid spacing and reaches the panchayat-cluster level, with 64% better accuracy in high-risk zones and 30% better extreme-rainfall forecasts.
What is a nowcast?
A nowcast is an ultra-short-range forecast for the next 0 to 2 hours, typically driven by Doppler radar data, used for urban flooding, aviation, and event safety.
How many radars feed BFS data?
BFS currently integrates data from 40 Doppler weather radars across India, with the network expanding toward 100 radars.
Who benefits from BFS forecasts?
Farmers, disaster managers, urban planners, hydropower and wind operators, aviation, and emergency services all benefit from the finer-grain forecasts. Krishi Vigyan Kendras use BFS outputs in advisories at the panchayat-cluster level.