Every active fire on Earth is visible from orbit. NASA and NOAA satellites scan the planet's surface every few hours, detecting thermal anomalies down to a hectare. Learn how scientists track, classify, and forecast wildfires — and what the data reveals about a planet under fire stress.
🔥 OPEN LIVE WILDFIRE MAPOn any given day, tens of thousands of fires burn across the planet simultaneously. The vast majority are agricultural burns in sub-Saharan Africa, seasonal grassland fires in Australia, or controlled burns in managed forests. But embedded within that constant background are the catastrophic events — megafires that consume millions of hectares, generate their own weather, and inject smoke into the stratosphere. Distinguishing signal from noise, and tracking hazard in real time, is the mission of global wildfire monitoring systems.
Fire detection from space relies on thermal infrared remote sensing. Burning vegetation emits radiation across a broad electromagnetic spectrum; at mid-infrared wavelengths (~3.9 µm), actively burning fires stand out dramatically against cooler background surfaces. NASA's MODIS (Moderate Resolution Imaging Spectroradiometer) and the more capable VIIRS (Visible Infrared Imaging Radiometer Suite) on board the Suomi-NPP and NOAA-20 satellites scan the entire Earth's surface every 1–2 days, flagging pixels with anomalous thermal signatures as "active fire detections".
The Fire Information for Resource Management System (FIRMS), operated by NASA, aggregates MODIS and VIIRS detections and makes them publicly available within 3 hours of satellite overpass. Each detection record contains geographic coordinates, brightness temperature, fire radiative power (FRP), detection confidence, and acquisition time. FIRMS data feeds government emergency management systems, insurance risk models, air quality forecasts, and — via platforms like Pandita Data — real-time public monitoring dashboards. The system processes over 3 million fire detections per year globally.
Fire Radiative Power (FRP) is the single most operationally useful metric in the satellite record. It measures the rate at which a fire releases radiative energy — a direct proxy for combustion intensity, fuel consumption rate, and smoke emission. A grass fire in the Sahel may produce 20–50 MW FRP; a crown fire in a dense boreal forest can exceed 5,000 MW. The 2019–2020 Australian Black Summer fires produced sustained FRP values above 10,000 MW in multiple pixels simultaneously — values that had never been recorded in the MODIS era and were used to explain the unprecedented pyroconvective columns that injected smoke into the stratosphere.
Satellite detection tells you where a fire is burning. Fire weather indices tell you why — and where conditions are primed to explode. The Fire Weather Index (FWI), developed by the Canadian Forest Service and adopted globally by the Copernicus Emergency Management Service, combines temperature, relative humidity, wind speed, and precipitation into a composite danger rating. FWI values above 30 indicate extreme danger; the 2019 Australian fires saw sustained FWI values above 50 across millions of hectares — a historically unprecedented combination of heat, drought, and wind.
| FIRE WEATHER INDEX | DANGER CLASS | BEHAVIOUR |
|---|---|---|
| 0–5 | Low | Fire unlikely to spread; controllable if started |
| 6–11 | Moderate | Slow spread; control possible with standard resources |
| 12–19 | High | Rapid spread possible; air tankers may be required |
| 20–29 | Very High | Intense surface fire; significant spotting, difficult control |
| 30–37 | Extreme | Crown fire potential; firefighter safety threatened |
| > 38 | Catastrophic | Fire generates own weather; uncontrollable by any means |
The most extreme wildfires cross a threshold where they generate their own atmospheric dynamics. Pyrocumulonimbus (pyroCb) clouds form when intense heat from a fire creates an updraft powerful enough to reach the upper troposphere — and sometimes the stratosphere. A pyroCb behaves identically to a thunderstorm, complete with lightning (which ignites new fires), violent downdrafts (which accelerate existing fire spread), and precipitation that rarely reaches the ground. The 2021 BC wildfires produced a pyroCb that injected smoke to 23 km altitude — detectable by the CALIPSO lidar satellite for months afterward as it circled the hemisphere.
The Black Summer fire season burned approximately 18.6 million hectares across southeastern Australia — an area larger than England and Wales combined. At peak intensity, over 100 fires burned simultaneously in New South Wales alone. VIIRS detected FRP values above 10,000 MW in multiple pixels, a record for the sensor. The fires generated at least 23 discrete pyroCb events, injecting aerosols to 35 km altitude and measurably cooling southern hemisphere surface temperatures for months.
NASA FIRMS data enabled near-real-time perimeter mapping that fed evacuation decisions for over a million residents. The event fundamentally changed how Australian and global fire agencies use satellite data operationally — shifting from post-event analysis to active incident support.
The global fire map has a clear structure driven by climate, land use, and seasonality. Sub-Saharan Africa accounts for roughly 70% of all global burned area — dominated by savanna burning for land management, which is ecologically normal but a major source of atmospheric carbon. South America sees intense fire seasons in the Amazon and Cerrado, driven by deforestation-linked ignitions that peak in August–October. Boreal forests in Siberia and Canada burn large areas during warm summers, with individual fires exceeding one million hectares. Mediterranean Europe — Greece, Portugal, Spain — sees catastrophic fire years during summer heatwaves combined with drought. The common thread across all regions: fire risk is amplifying as climate warming extends dry seasons and intensifies heat extremes.
| REGION | PEAK SEASON | PRIMARY DRIVER | TREND |
|---|---|---|---|
| Sub-Saharan Africa | Nov–Mar | Agricultural burning, savanna | Stable / declining (mgmt) |
| Amazon / Cerrado | Aug–Oct | Deforestation ignitions, drought | ↑ Increasing |
| Siberian Boreal | Jun–Aug | Lightning, permafrost thaw drying | ↑↑ Strongly increasing |
| Western North America | Jul–Oct | Drought, heat, fuel accumulation | ↑↑ Strongly increasing |
| Mediterranean Europe | Jun–Sep | Heatwaves, rural abandonment | ↑ Increasing |
| Southeast Australia | Dec–Mar | Drought, heat, wind | ↑ Increasing |
Satellite fire detection requires a clear line of sight between sensor and fire. Thick smoke plumes — often present over the most intense fires — can partially or fully obscure MODIS and VIIRS detections. Similarly, cloud cover during active fire seasons (Amazon wet season boundary, monsoon Africa) creates detection gaps. Agencies compensate by fusing multiple satellite overpasses and applying atmospheric correction, but near-real-time data in these conditions should always be treated as a lower bound — actual fire extent is likely larger than detected.
A wildfire's direct burn footprint is its least far-reaching impact. Smoke from large fires travels hemispheric distances, degrading air quality thousands of kilometres from the source. The 2023 Canadian wildfires turned New York City's sky orange and pushed PM2.5 concentrations to hazardous levels across the northeastern United States. NOAA's Hazard Mapping System (HMS) and the Copernicus Atmosphere Monitoring Service (CAMS) produce daily global smoke forecasts by ingesting satellite FRP data into atmospheric transport models — generating 5-day aerosol optical depth forecasts that drive public health advisories across continents.
The 3D Wildfire simulation above pulls active fire detections from NASA FIRMS in near-real time, rendering each detection as a point scaled by Fire Radiative Power. Bright, large points indicate intense, actively burning fires; small points represent smouldering or low-intensity detections. The global view reveals the fire distribution structure instantly — the African savanna belt, the Amazon fire season, boreal hotspots — in a way no flat map can convey.
Rotate the globe, zoom to a region of interest, and watch the live feed update. Each point is a real thermal anomaly detected by a satellite within the last 24–48 hours.