Wildfire Satellite Detection Race Heats Up as FireSat, NOAA, and OroraTech Deploy Orbital Networks
Three separate satellite programs are converging in 2026 to transform wildfire detection from space, while university researchers benchmark AI models that could process the flood of new orbital data.
For decades, wildfire detection from space relied on a handful of government weather satellites whose instruments were designed for broad atmospheric observation, not pinpointing ignitions. In 2026, that constraint is dissolving. Three distinct satellite programs — the Google-backed FireSat constellation, NOAA’s Next Generation Fire System, and Munich-based OroraTech’s dedicated CubeSat fleet — are reaching operational milestones within months of one another, creating overlapping layers of orbital fire surveillance that did not exist a year ago.
Meanwhile, university research groups are stress-testing the AI models that will ultimately process these satellite feeds, and NASA is building a Wildfire Digital Twin that aims to fuse all of these data streams into real-time forecasts. Together, the developments mark the most significant shift in wildfire monitoring infrastructure since NOAA’s GOES satellites began carrying dedicated fire-detection algorithms in the early 2000s.
FireSat: 5-Meter Resolution and 20-Minute Revisits
The most ambitious program is FireSat, a joint effort of the nonprofit Earth Fire Alliance, satellite manufacturer Muon Space, and Google Research. A protoflight satellite launched in March 2025 aboard SpaceX’s Transporter-13 mission from Vandenberg Space Force Base and has since returned its first wildfire images, capturing active fires across Oregon, Ontario, Alaska, and Australia.
FireSat’s six-band multispectral infrared payload — spanning mid-wave, long-wave, and short-wave infrared channels — can detect fires as small as five meters by five meters, according to Muon Space. That represents an improvement of roughly two orders of magnitude over existing geostationary fire-detection instruments, which typically resolve hotspots at the kilometer scale. The sensor’s 1,500-kilometer observation swath delivers a 50-meter ground sample distance from overhead, enabling it to map fire perimeters, track progression, and estimate intensity through smoke and cloud cover.
The Earth Fire Alliance plans to deploy the first three operational FireSat satellites in mid-2026, which will provide twice-daily global observation. The full constellation — more than 50 satellites — is targeted for completion by 2030 and would scan Earth’s most fire-prone regions every 20 minutes. The Gordon and Betty Moore Foundation and the Environmental Defense Fund are among the program’s backers.
The projected impact figures are striking. According to the Earth Fire Alliance, even a one-hour revisit rate in the United States alone could annually save more than one billion dollars in fire damage, protect 3,500 homes, reduce burned land by 1.3 million acres, and prevent 21.9 million metric tons of carbon emissions.
NOAA Opens the Firehose
While FireSat is purpose-built, NOAA is extracting new wildfire capability from its existing geostationary fleet. On January 26, 2026, the agency launched the Wildland Fire Data Portal, a public-access platform that delivers experimental products from the Next Generation Fire System. NGFS applies automated heat-anomaly detection algorithms to continuous feeds from the GOES-East and GOES-West satellites, providing minute-by-minute fire detection and intensity monitoring.
“Having minute-by-minute access to wildfire data is critical during an active fire,” said Mike Pavolonis, Wildland Fire Program Manager at NESDIS STAR, in a statement accompanying the portal launch. “It provides the situational awareness wildfire response teams need.”
The system has already demonstrated its value during real incidents, including the 2025 Dragon Bravo fire and recent Colorado detections, according to NOAA. The portal, hosted on the NESDIS Common Cloud Framework, provides geolocation-improved imagery integrated with high-resolution maps that allow users to zoom to local scales. NGFS is expected to transition from experimental to full operational status by late 2026.
NOAA’s approach differs fundamentally from FireSat in its architecture. Where FireSat builds a new constellation from scratch, NGFS layers smarter algorithms on top of existing GOES infrastructure, offering continuous temporal coverage at coarser spatial resolution. The two systems are complementary: GOES can flag a new ignition within minutes from geostationary orbit, and FireSat’s low-Earth-orbit passes can then deliver high-resolution characterization of the same fire.
OroraTech: The European Contender
The third major program comes from OroraTech, a Munich-based company that deployed eight wildfire-detection CubeSats in March 2025 via a Rocket Lab Electron launch from New Zealand. The satellites, placed into a 550-kilometer sun-synchronous orbit, carry thermal imaging sensors designed for near-real-time fire monitoring.
OroraTech’s roadmap calls for up to 100 satellites by 2028, which would reduce maximum revisit gaps to 30 minutes. The company raised 27 million dollars in Series B funding in October 2024 and holds a 72-million Canadian dollar contract with Spire Global for satellite manufacturing as part of Canada’s WildFireSat mission, with a deployment target of 2029.
The proliferation of commercial wildfire constellations signals a broader shift in how fire agencies source their data. Rather than relying solely on government instruments with multi-year procurement timelines, agencies can now tap commercially operated networks that iterate hardware on annual cadences.
AI Models: Promising but Not Yet Ready to Lead
The torrent of new satellite data will be useful only if analytical tools can process it fast enough to inform decisions. A study published in the journal Natural Hazards and highlighted by the University at Buffalo in March 2026 offers a sobering benchmark. Researchers led by Yingjie Hu, an associate professor of geography, evaluated five deep-learning architectures — LSTM, U-Net, U-Net with attention, ConvLSTM, and ConvLSTM with attention — against 221 wildfires in Hawaii spanning 2012 to 2023.
ConvLSTM, a model that combines convolutional neural networks with long short-term memory layers to capture both spatial and temporal patterns, achieved the highest precision among the AI models tested. Its attention-enhanced variant showed higher recall, meaning it was less likely to miss actual fire spread. Yet the physics-based fire modeling tool FARSITE still achieved a superior F1 score and precision overall.
“Wildfires are becoming more frequent and more destructive, and communities need tools that can anticipate how these events will unfold,” Hu said. The study’s main takeaway was that AI can complement, but not yet replace, established physics-based fire models — a finding that suggests the operational pipeline will likely combine satellite detection with hybrid physics-AI spread prediction rather than relying on neural networks alone.
Separately, researchers at Johns Hopkins University and the University of Florida received nearly 1.2 million dollars from the National Science Foundation’s Fire Science Innovations program in February 2026 to develop AI-enabled tools that predict human behavior during wildfire evacuations, addressing the often-overlooked downstream challenge of what happens after a fire is detected.
NASA’s Digital Twin: Fusing the Feeds
NASA’s Wildfire Digital Twin program represents perhaps the most technically ambitious piece of the puzzle. The initiative aims to use AI and machine learning to forecast potential burn paths in real time by integrating satellite feeds, weather data, terrain models, and vegetation maps. The target spatial resolution is 10 to 30 meters per pixel — an improvement of roughly two orders of magnitude compared to current global fire models.
The Digital Twin is designed to consume data from multiple satellite sources, making it a potential integration layer for FireSat, GOES, and OroraTech feeds. If successful, it could turn the patchwork of overlapping constellations into a coherent operational picture.
What Converging Coverage Means for Fire Response
The strategic significance of 2026’s satellite deployments lies in their overlap. No single system solves the wildfire detection problem alone. NOAA’s GOES satellites offer continuous temporal coverage from geostationary orbit but at coarse spatial resolution. FireSat delivers high spatial resolution from low Earth orbit but with gaps between passes. OroraTech occupies a middle ground with moderate resolution and a commercial iteration speed that government programs cannot match.
Layered together, these systems begin to approximate something fire agencies have long sought: persistent, high-resolution, global fire surveillance. The remaining bottleneck is analytical — translating raw detections into actionable spread forecasts fast enough to inform evacuation decisions and resource deployment. The University at Buffalo benchmarks and NASA’s Digital Twin program suggest that challenge is being attacked in parallel, even if the AI tools are not yet ready to operate autonomously.
The wildfire satellite race is not a competition in the traditional sense. It is a convergence, and its beneficiaries will be the fire response teams and communities that, for the first time, may learn of an ignition within minutes rather than hours.