What VayuMet Provides

Our Data Sources

VayuMet does not generate its own forecasts or observations. We download, process and visualise output from publicly available, authoritative sources: numerical weather prediction models operated by NOAA (USA) and ECMWF (Europe), long-term reanalysis and climatology archives from the NOAA Physical Sciences Laboratory (PSL), observed gridded rainfall and temperature from the India Meteorological Department (IMD), Pune, and live satellite imagery from the Japan Meteorological Agency (JMA).

Forecast models

Dataset	Operator	Use in VayuMet	Resolution	Update Cycle
GFS (Global Forecast System)	NOAA / NCEP	Short-range weather: rain, wind, temperature, clouds, humidity, CAPE/CIN, soil	0.25°	4× daily (00Z, 06Z, 12Z, 18Z)
CFSv2 (Climate Forecast System v2)	NOAA / NCEP	Seasonal forecasts: precipitation & temperature anomalies	1.0°	4× daily, monthly outlooks
IFS-HRES (Integrated Forecasting System)	ECMWF open data	Physics-based rainfall input to the Rain Futures index projection (VM-IFS)	0.25°	Daily
AIFS (AI Forecasting System)	ECMWF open data	Machine-learning rainfall input to the Rain Futures index projection (VM-AIFS)	0.25°	Daily
ECMWF-CAMS Aerosols	ECMWF	Air quality: PM2.5, PM10, AOD, dust AOD; input to experimental visibility	0.25°	Daily

Observations & live feeds

Dataset	Operator	Use in VayuMet	Resolution	Update Cycle
AWC (Aviation Weather Center)	NOAA	METARs, TAFs, SIGMETs and significant-weather charts	Station / airspace	Real-time
Himawari-9 (TRM/B13 IR, Sandwich RGB)	JMA — Meteorological Satellite Center	Live satellite imagery over South & Central Asia	Satellite native	Continuous (~10 min)
IMD real-time gridded rainfall	IMD Pune	Observed rainfall for the Monsoon Monitor's forecast-vs-actual comparison	0.25°	Daily, subject to IMD publishing lag
NCDEX public disclosures (CDR spot, LPA, futures prints)	NCDEX	Official RAINMUMBAI and RAINCHNNAI index values, against which the Rain Futures desk verifies every day	Contract / city	Daily, 09:15 IST

Climatology & reanalysis archives

The District Climatology and Climate Change pages are built from long-term observed and reanalysis archives rather than forecast model output.

Dataset	Operator	Use in VayuMet	Resolution	Period
IMD daily gridded archives (rain, tmax, tmin)	IMD Pune, via imdlib	Rainfall and temperature day-count distributions, yearly and monthly anomaly series, warming-trend analysis, and the reference climatology used to bias-correct seasonal forecasts	0.25° rain · 1° temperature	1951–2025
CPC gridded daily climatology	NOAA PSL	Average temperature and precipitation charts; daily climatology curves	~0.5°	1991–2020 normals
NCEP-DOE Reanalysis 2 (daily 10 m u/v wind)	NOAA PSL	Wind speed distributions, wind roses and monthly wind normals	~1.9°	1996–2025
NCEP monthly cloud climatology	NOAA PSL	Cloud-cover percentages, sunny/overcast day estimates, yearly cloud series	~2°	1979–2025
CPC Global Precipitation	NOAA PSL	Weekly rainfall analysis products	0.5°	Ongoing

How Data is Processed

Short-Range Forecasts (10 days)

Raw GFS GRIB2 files are downloaded from NOAA's NOMADS server, regridded to a uniform India bounding box (30°E–130°E, 5°S–50°N), and exported as GeoTIFF files. The GeoTIFFs are served directly to the browser, where a custom WebGL renderer applies colour ramps and overlays the data on an interactive Leaflet map.

Seasonal Forecasts

Monthly Climate Forecast System ensemble output is bias-corrected against IMD Pune reference climatology and then spatially joined to India's district and state boundaries. The result is a district-level anomaly map (precipitation in %, temperature in °C departure) updated monthly for the June–October monsoon season window.

Monsoon Monitor — Forecast vs Observed

The Monsoon Monitor places the seasonal forecast beside what has actually fallen. Observed rainfall comes from IMD's real-time 0.25° daily gridded rainfall, accumulated from 1 June to date and aggregated to district and state polygons, then expressed in the same IMD percentage-departure categories used in official monsoon reporting. Because IMD publishes with a short lag, the most recent day or two may be absent until the archive catches up.

District Climatology & Warming Trends

Long-period statistics are computed from real daily records rather than model output. IMD Pune daily rainfall (0.25°) and tmax/tmin (1°) grids covering 1951–2025 are downloaded via imdlib, scanned year by year, and reduced to district-level distributions — rainfall-amount bins, wet-day counts, hot-day and cold-night counts — and to monthly and yearly anomaly series for the warming-trend analysis. Wind statistics come from NCEP-DOE Reanalysis 2 daily 10 m winds (1996–2025) held at NOAA PSL, and cloud statistics from PSL's NCEP monthly cloud series.

One caveat is handled explicitly: IMD's temperature grid is 1° (~111 km), which does not resolve Himalayan terrain — a naive read placed Shimla near 20 °C annual mean against a true value close to 13.6 °C. The Climate Change series is rebased onto a finer reference to correct this rather than presenting the raw grid value.

Wind Animation

U and V wind components at each pressure level are extracted from GFS output and converted to Leaflet-Velocity JSON format, enabling smooth particle-flow animation in the browser without server-side rendering.

Experimental Visibility Estimation

Two complementary methods derive surface visibility from ECMWF-CAMS aerosol fields, both corrected for humidity-driven aerosol swelling using a per-pixel f(RH) factor computed from GFS 2 m temperature and dew point via the Magnus formula:

• AOD-based IGP formula: V = 4 × Heff / (AOD × f(RH)) — integrates column aerosol optical depth over an effective mixing height (default 1,000 m; ERA5 boundary-layer height used when the --met-grib option is supplied for the full operational formula).
• PM₂.₅ Koschmeider formula: βext = (3×f(RH) + 0.3)×PM₂.₅ + 10 then V = 3 912 000 / βext — requires no mixing height, deriving visibility directly from surface fine-particle concentration. Particularly suited to urban and industrial pollution haze.

These are experimental products intended for fog/haze situational awareness, not safety-critical navigation. Results should be cross-checked against METAR visibility reports.

Impact-Based Warnings and Advisories

Most weather services stop at the event: heavy rain, orange warning, 15–17 July. That tells you what the atmosphere will do, but not what it will do to you — and the answer differs completely depending on whether you are a district administrator, a farmer with a standing crop, a flight dispatcher or a wind-farm operator. Following the direction set by the WMO's Impact-Based Forecast and Warning Services framework, VayuMet carries the forecast one step further and states the consequence and the recommended action.

Every district's 10-day forecast and meteogram — both the standalone page and the advisory tray on the main map — is passed through a rule engine that converts the raw meteogram into four parallel sector bundles:

Sector	Read from	Example of what it says
General Public & Tourism	Temperature and feels-like, rain, wind, weather type	Heavy rain (IMD orange) — waterlogging possible in low-lying areas, plan travel with buffer time. Heat wave — avoid outdoor exposure 12–16 IST. Thunderstorm — move indoors at first thunder; unsafe for open-area sightseeing, high-profile vehicles and boating.
Agriculture	Spray window, ETo, soil moisture, soil temperature	Spraying favourable on named days (calm, dry windows — prefer early morning). Sowing window suitable — adequate seed-zone moisture and soil temperature. Irrigation advised: crop-water demand ≈ X mm ETo against only ≈ Y mm expected rain. Prolonged leaf wetness — elevated fungal-disease pressure. Gusty squalls — lodging risk in standing crops; support orchards and trellises.
Aviation	Flight category, ceilings, icing, turbulence, CB	IFR/LIFR periods expected, mainly night and early morning — plan alternates and fuel accordingly. Embedded CB with tops to FL — expect convective SIGMET conditions, avoid penetration. Moderate icing between named flight levels, with the freezing level in metres AMSL.
Renewable Energy	Solar GHI and PV yield, 100 m wind, power density	Reduced PV yield on named days — schedule solar-plant maintenance in these windows. High cell-temperature derating (~8–15% below rated) on hot afternoons. Sustained generation ≥12 h/day in the full-power band. Storm cut-out risk at 100 m wind ≥ 90 km/h — expect turbine shutdowns and steep ramp events.

Three design decisions matter here. First, every advisory names the dates it applies to — collapsed into readable ranges such as 15–17 Jul, 20 Jul — so guidance is actionable rather than general. Second, the thresholds follow IMD warning criteria, ICAO meteorological practice and GKMS-style agromet guidance rather than being invented, and each bullet is dotted in its IMD tier colour (red, orange, yellow, green). Third, the rules live in a single source of truth shared by both the standalone meteogram page and the main-map advisory tray, so a threshold changed in one place changes both surfaces and the two can never drift apart.

The aviation bundle carries a standing caveat on every card: it is advisory only, and live flight operations must use current METAR, TAF and SIGMET from the aerodrome meteorological office under ICAO Annex 3.

Rainfall Index Projections — Rain Futures Desk

The Rain Futures Index Desk tracks the two exchange-traded rainfall indices listed by NCDEX — RAINMUMBAI (southwest monsoon, 1 June – 30 September, IMD Santacruz and Colaba) and RAINCHNNAI (northeast monsoon, 1 September – 31 December, IMD Meenambakkam and Nungambakkam). Each index is a Cumulative Deviation Rainfall (CDR) series: it opens the season at the seasonal Long Period Average anchor — 2,206.7 mm for Mumbai, 998.25 mm for Chennai — and each day adds the observed rainfall and subtracts that calendar day's LPA.

VayuMet takes the latest official CDR spot published by the exchange and runs it forward 14 days using two structurally independent rainfall forecasts drawn from ECMWF open data:

• VM-IFS — daily rainfall from ECMWF's physics-based IFS-HRES, extracted at the contract's designated observatory grid points.
• VM-AIFS — the same extraction applied to ECMWF's machine-learning AIFS, a genuinely different model formulation rather than a second member of the same one.

Both trajectories are published without bias correction. Out of fold, every correction scheme tested scored worse than leaving the model alone (VM-AIFS: 122 days, leave-one-month-out; VM-IFS: 242 days, leave-one-season-out), so none is applied. The gap between the two lines is published as the honest measure of uncertainty, and it is wider than either model's own error statistics imply — both under-forecast heavy rainfall, so any active spell in these trajectories should be read as a lower bound. Every projection is verified the following day against the official published spot, and that verification record is shown on the page.

VIMI — The VayuMet Indian Monsoon Index

Most monsoon diagnostics measure an ocean signal that influences the monsoon. VIMI measures the monsoon itself. It is a seven-component weighted composite index computed on the semi-permanent circulation features that constitute the Indian summer monsoon — the monsoon trough, the Somali low-level jet, the Tibetan anticyclone, the tropical easterly jet, the heat low, the Mascarene high and Bay of Bengal flow. Each component is derived from mass, wind and thickness fields, weighted by its contribution to monsoon strength, and combined into a single index value.

The index is computed on reanalysis for the historical record and on operational GFS output to 16 days for the forward outlook, giving a phase forecast of the active–break cycle. Verified over JJAS 2009, 2018, 2019 and 2023, VIMI carries 72% skill on active/break spell identification and 81% accuracy on monthly rainfall departure direction. The full methodology and verification are published in the VIMI scientific paper (PDF).

Alert Event Classification (Experimental)

The weather alert pipeline applies an experimental classification step that discriminates convective from non-convective (stratiform) precipitation events and derives an associated peak gust speed forecast for each district polygon (~25–50 km scale). Convective events are flagged for rapid-onset hazards (sudden gusts, hail, wind shear); stratiform events are flagged for persistent-rainfall and flooding risk. This event-type + gust pairing provides operationally specific guidance for aviation, district administration, agriculture, and renewable energy sectors that a single rainfall-amount alert cannot deliver, and it feeds directly into the impact-based advisories described above.

Accuracy & Limitations

• GFS is a global model — it does not resolve fine-scale terrain features like the Western Ghats or Himalayan foothills at sub-district level.
• Forecast skill degrades beyond 5 days for rainfall, and beyond 3 days for convective events.
• Seasonal forecasts are probabilistic — CFSv2 anomaly maps indicate the expected departure from normal, not a precise monthly total.
• Aviation data (METARs, TAFs) reflects observations from NOAA AWC feeds and may lag real-time by 5–15 minutes.
• Climatology resolution varies by source — IMD rainfall is 0.25° but IMD temperature is 1°, and the PSL reanalysis products used for wind and cloud are coarser still (~1.9° and ~2°). District values are grid-cell averages, so mountainous districts are represented less faithfully than plains.
• Observed rainfall lags — IMD publishes its real-time gridded rainfall with a short delay, so the Monsoon Monitor's observed side may trail the current date by a day or two.
• Impact advisories are rule-based, not site-specific — they translate a district-scale forecast through published IMD, ICAO and agromet thresholds. They cannot see your field's drainage, your aerodrome's local effects or your plant's equipment, and they inherit every limitation of the underlying forecast.
• Rainfall-index projections are not settlement values — the VM-AIFS and VM-IFS CDR trajectories are VayuMet's own analysis, carry no bias correction, and both models under-forecast heavy rain. They are not official NCDEX values, not a forecast of any contract price, and not investment or trading advice.
• Two gauges per city — RAINMUMBAI and RAINCHNNAI settle on rainfall at exactly two IMD observatories each, so an index value can diverge sharply from rainfall a short distance away. That basis risk is inherent to the contracts, not to our projection.
• VIMI is a circulation index, not a rainfall forecast — it describes the strength and phase of the monsoon circulation. Rainfall follows circulation imperfectly, particularly at district scale and in orographic terrain.

Expertise & Authority

VayuMet is founded and led by Air Commodore Neeraj Sharma, M.Phil., M.Sc., an Indian Air Force veteran whose career spans over three decades of hands-on operational aviation meteorology, numerical weather prediction, agrometeorology, climate data analysis, international United Nations peacekeeping service with MONUC, large-scale organisational leadership, policy formulation, technology adoption and professional education — extended in recent years into commodity weather-derivative analysis. This is not a hobbyist project — it is built on the kind of domain authority that comes only from sustained frontline service at the highest levels of India's defence meteorological establishment.

This operational depth — spanning frontline aviation weather services under ICAO procedure, United Nations peacekeeping deployment in the Congo basin, numerical weather prediction across the short-range to seasonal spectrum, agrometeorology, long-period climate data analysis, weather-derivative and commodity-futures work, strategic policy leadership, large-scale personnel management, sensor technology expertise, and meteorological education — directly shapes VayuMet's design philosophy: aviation-grade rigour, transparent data sourcing, and practical usability for Indian conditions.

Methodology References

Our processing pipeline references methodologies, datasets and standards published by NOAA, ECMWF, JMA, IMD, and the Indian Institute of Tropical Meteorology (IITM).

• NOAA GFS Model Description — nco.ncep.noaa.gov
• NOAA CFSv2 Seasonal Forecast System — ncei.noaa.gov
• NOAA Physical Sciences Laboratory gridded datasets (CPC daily climatology, NCEP-DOE Reanalysis 2, NCEP cloud series) — psl.noaa.gov/data/gridded
• IMD Pune gridded daily rainfall and temperature archives, accessed via the imdlib package — imdpune.gov.in
• IMD Southwest Monsoon Climatology and rainfall-departure categories — imd.gov.in
• JMA Meteorological Satellite Center — Himawari imagery — data.jma.go.jp/mscweb
• IITM Monsoon Research — tropmet.res.in
• ECMWF open data — IFS-HRES and AIFS forecast fields — ecmwf.int/en/forecasts/datasets/open-data
• NCDEX RAINMUMBAI and RAINCHNNAI product notes, contract specifications and daily index disclosures — ncdex.com
• ICAO Annex 3 — Meteorological Services for International Air Navigation (aviation met standards)

VayuMet is an independent platform. It is not affiliated with, endorsed by, or operated on behalf of NOAA, ECMWF, JMA, the India Meteorological Department or NCDEX; their data is used under the open-data and public-disclosure terms each publisher provides.

Contact & Feedback

We welcome feedback, data correction reports, and collaboration enquiries from researchers, meteorologists and developers. Correspondence reaches Air Cmde Neeraj Sharma directly.

📧 Email: contact@vayumetweather.com

🌐 Website: www.vayumetweather.com

Open VayuMet Live Maps →