Satellite Fire Detection 2026: VIIRS, Sentinel-3 SLSTR, MODIS, GOES-R/MTG, FY-3

August 2024, Sierra Nevada. The Park Fire grew from one ignition to a 100,000-acre front in 24 hours. VIIRS on NOAA-20 caught the first hot spot at 21:34 UTC and read 18.4 MW FRP — already partially saturated. Sentinel-3B passed 81 minutes later and read 26.7 MW with no saturation. GOES-18 watched every 5 minutes and saw FRP climb from 8 to 60 MW in 90 minutes. No single satellite saw the full fire.

Detection physics: why 3.7 micrometres

Every satellite fire product rests on Planck’s law: the spectral radiance of a blackbody shifts toward shorter wavelengths as temperature rises. For typical fire temperatures of 700-1,200 K, peak emission falls in the 2.4-4.1 micrometre range, while terrestrial backgrounds at 290 K emit primarily at 8-14 micrometres. This makes the mid-infrared channel near 3.7 micrometres an exceptional contrast band for hot-spot detection: at 11 micrometres a fire on background may show a 5-10 K rise above background, whereas at 3.7 micrometres the rise reaches 50-200 K.

Roberts et al. (2003) describe the radiative physics of fire detection in detail in their foundational BAMS paper and explain why all modern algorithms use a combined 3.7 + 11 micrometre core (Roberts et al., 2003, Bulletin of the American Meteorological Society).

Wooster (2002) developed the baseline Fire Radiative Power (FRP) algorithm, which converts radiative contrast into megawatts and then into combustion rate and emissions (Wooster, 2002, Geophysical Research Letters). FRP became the single intensity metric that allows comparison of data from different satellites.

VIIRS: the new gold standard

The Visible Infrared Imaging Radiometer Suite is NOAA’s primary fire instrument. VIIRS flies on three operational platforms: Suomi NPP (launched 2011, end-of-life expected 2026-2027), NOAA-20 / JPSS-1 (2017), and NOAA-21 / JPSS-2 (2022). JPSS-3 (NOAA-22) and JPSS-4 (NOAA-23) are planned by the end of the decade.

The instrument has two channel groups: I-Bands (375 m at nadir, 5 channels) and M-Bands (750 m at nadir, 16 channels). The VNP14IMG fire product uses I4 (3.55-3.93 micrometres) and I5 (10.5-12.4 micrometres), with M-Bands as context. Schroeder et al. (2014) describe the algorithm, and NASA documents it on the VIIRS Land page.

The key VIIRS advantage over MODIS: a 375 m pixel instead of 1 km gives an area 7.1 times smaller, which means much higher sensitivity to small fires. Schroeder et al. (2014) showed that the 375 m product detects roughly 4 times more small fires (1 acre and below) at the same 1.2% commission error rate (Schroeder et al., 2014, RSE).

VIIRS limitations:

• Channel I4 saturates near 367 K in standard mode (about 470 K in fire-detection mode for NOAA-20). Intense fires (Australia 2019, Canada 2023) receive understated FRP estimates.
• Polar orbit produces 4-6 passes per day in the tropics, 10-14 at high latitudes.
• The 3,040 km swath retains full 375 m resolution only out to a 56° zenith angle. Pixel size grows to 800 by 1,600 m at swath edges.

Csiszar et al. (2014) validated the VIIRS Active Fire Product against ASTER and ground observations, confirming high agreement with MODIS and better completeness for small fires (Csiszar et al., 2014, JGR Atmospheres).

MODIS: the veteran of fire science

The Moderate Resolution Imaging Spectroradiometer flies on Terra (launched 1999) and Aqua (2002). MODIS established the scientific fire-data foundation underpinning thousands of publications. The MOD14/MYD14 product (Collection 6.1) operates with a 1 km pixel and uses channels 4 (3.75 micrometres), 22 and 23 (also 3.7 micrometres at different dynamic ranges), 31 (11 micrometres), and 32 (12 micrometres) (NASA MODIS docs).

Giglio et al. (2016) describe the Collection 6 Algorithm Theoretical Basis Document (ATBD): tighter neighbourhood criteria reduced commission error by 50% relative to Collection 5 (Giglio et al., 2016, RSE).

Current MODIS status and issues:

• Terra (Aqua) has been operating since 2000 (2002), and both satellites operate beyond their design life. NASA plans to deorbit Aqua in 2026.
• The Terra orbit is degrading, with the nadir crossing time drifting from the original 10:30 a.m. toward 11:00 a.m. and later, complicating long-term records.
• The 1 km pixel makes MODIS inferior to VIIRS for small-fire detection, but the historical 25-year record (2000-2025) remains a unique scientific resource.

Kaufman et al. (1998) described the original MODIS Fire Product and its theoretical foundations in a classic paper that remains the basis for all subsequent work (Kaufman et al., 1998, JGR).

Sentinel-3 SLSTR: the European alternative

The Sea and Land Surface Temperature Radiometer flies on Sentinel-3A (launched 2016) and Sentinel-3B (2018). The third satellite, Sentinel-3C, is preparing for a 2026 launch, ensuring continuity to 2030+. The mission is described on the ESA mission page.

SLSTR has nine spectral channels from visible to far-infrared. Fire detection uses F1 (3.74 micrometres) and F2 (10.85 micrometres), purpose-built fire channels with a wide dynamic range. F1 spatial resolution is 1 km in standard mode and 0.5 km in fire mode. The 1,420 km swath in nadir-only view produces 1-2 passes per day at mid-latitudes for the two-satellite constellation.

Key advantage: F1 saturates near 650 K instead of 470 K on VIIRS. This makes SLSTR the only civilian sensor that delivers an undistorted FRP estimate for very intense fires. Wooster et al. (2021) validated SLSTR FRP against ground experiments and obtained an RMSE near 18% for typical large fires (Wooster et al., 2021, Remote Sensing).

Second advantage: a dual viewing angle (nadir plus 55° oblique) enables better atmospheric correction for thermal channels through an effectively extended optical path.

SLSTR limitations:

• Lower daily coverage compared with VIIRS or MODIS. For mid-latitude operations this means 3-4 passes per day for the two-satellite constellation.
• Distribution latency through Copernicus EUMETSAT is typically 60-180 minutes for the Near-Real-Time product.
• Less integration experience in North American fire systems, where NASA products have historically dominated.

GOES-R and Meteosat Third Generation: the geostationary revolution

Geostationary satellites observe one hemisphere continuously. This shifts the operational picture: instead of 4-6 images per day we get 100-1,000.

The GOES-R series includes GOES-16 (East), GOES-18 (West), and GOES-19 (launched 2024). The Advanced Baseline Imager (ABI) has 16 channels from visible to infrared. Channel 7 (3.9 micrometres) is the primary fire channel. Spatial resolution is 2 km at nadir, with pixel width growing to 4-6 km at high latitudes. Schmidt et al. (2017) describe the FDC (Fire Detection and Characterization) algorithm (Schmidt et al., 2017, Atmospheric Environment).

Temporal resolution: 5 minutes for the CONUS sector, 1 minute for mesoscale sectors (1,000 by 1,000 km, retargetable to an active fire). This makes GOES-R indispensable for tracking fire-front dynamics through the day.

Weaknesses: a 2 km pixel implies a detection threshold of 30-50 MW for most fires. Hall et al. (2019, IEEE TGRS) estimate a 50 MW threshold for Canadian boreal fires at summer noon. Geostationary geometry produces high zenith angles at high latitudes, which degrades contrast and complicates detection.

Meteosat Third Generation MTG-I1 launched in December 2022. The Flexible Combined Imager (FCI) has 16 channels, and the 3.8 micrometre channel is used for fire detection. Spatial resolution is 1 km in the visible and 2 km in the infrared. Temporal resolution is 10 minutes for the full disk and 2.5 minutes for the European mesoscale sector (EUMETSAT MTG). MTG-I2 and MTG-I3 are planned for 2026 and 2030.

Himawari-8 and Himawari-9 from JMA provide geostationary coverage of the Pacific. The AHI instrument has a 3.9 micrometre channel. Spatial resolution is 2 km, temporal 10 minutes for the full disk. Wickramasinghe et al. (2020) validated the Himawari fire product over Australia and Southeast Asia and obtained acceptable accuracy for fires larger than 0.5 ha (Wickramasinghe et al., 2020, Remote Sensing).

FY-3 and FY-4: the Chinese constellation

Feng-Yun (FY) is a Chinese meteorological system. FY-3D (launched 2017) and FY-3E (2021) are polar satellites carrying MERSI-II, which has a 3.72 micrometre fire channel with a 250 m pixel (full resolution) and 1 km (standard fire product). FY-4A and FY-4B are geostationary over East Asia, carrying AGRI with a 3.72 micrometre channel and a 4 km pixel.

The instruments are technically comparable to Western counterparts. The issue for Western users is data-distribution standards and policy. Products are not published through a FIRMS analogue; scientific access is possible through CMA after coordination. For Europe and Ukraine, the FY platform is not operationally integrated into fusion systems in 2026 due to the absence of open distribution standards. We mention it as a technical fact only.

Comparative table of key parameters

Algorithms: contextual versus absolute thresholds

Every modern fire algorithm is contextual. Rather than applying a simple absolute temperature threshold (“if T > 320 K, this is a fire”), each pixel is compared with its neighbours. This is required to exclude background heating, since sun-warmed sand in a desert can exceed 320 K without any fire.

The base algorithm logic:

1. Candidate selection: a pixel must have T_3.7 above a given threshold.
2. Background characterisation: T_3.7 and T₁₁ statistics in a window around the candidate (typically 9 by 9 pixels).
3. Contextual tests: the difference DT = T_3.7 minus T₁₁ must exceed a threshold plus delta times sigma_bg, where sigma_bg is the background standard deviation.
4. Artefact rejection: water mask, cloud mask, sun-glint mask, industrial hot-spot mask.
5. FRP computation using the Wooster (2002) formula.

Details vary: VIIRS uses more aggressive I-Band thresholds in favour of completeness, MODIS Collection 6 uses tighter neighbourhood criteria in favour of accuracy, and SLSTR uses dual viewing angles for an additional consistency test. Giglio (2003) gave the canonical multi-stage description that became the foundation for all subsequent work (Giglio et al., 2003, RSE).

Errors: commission and omission in real conditions

Each sensor has characteristic errors.

VIIRS: Schroeder et al. (2014) estimate a 1.2% commission rate and an 8% omission rate for fires larger than 4 acres in CONUS. For fires below 1 acre, omission rises to 30-40%.

MODIS: Giglio et al. (2016) estimate a 1.5% commission rate for Collection 6 (half of Collection 5). Omission depends strongly on landscape: 5-10% for African savannas and 25-40% for Siberian boreal forests in winter.

SLSTR: Xu et al. (2020) validated SLSTR against MODIS and VIIRS and obtained 1.8% commission and 12% omission for fires above 5 MW (Xu et al., 2020, RSE).

GOES-R FDC: Hall et al. (2019) estimated a GOES-16 ABI sensitivity threshold of 50 MW in Canadian boreal forests at high latitudes during summer noon (Hall et al., 2019, IEEE TGRS).

A separate case is industrial commission errors. Blast furnaces at metallurgical complexes (Kryvyi Rih, Zaporizhzhia, historically Mariupol, Kremenchuk refinery), oil and gas flares, and thermal power plants in startup-burn mode all produce a thermal contrast similar to a fire. Only static industrial fire masks allow filtering them in post-processing.

Operational distribution ecosystem

Raw data from each satellite is available through specific channels. Key access points for operational use:

• NASA FIRMS for VIIRS and MODIS, with 3-hour latency for the standard stream, 60 minutes for LANCE-MODIS, and 15 minutes for US/Canada Direct Broadcast (FIRMS).
• EUMETSAT EUMETCast for Sentinel-3 and Meteosat MTG, with 60-180 minute latency (EUMETCast).
• NOAA STAR for GOES-R FDC, with 5-15 minute latency (NOAA STAR).
• Copernicus EFFIS as the integrated European portal with VIIRS, MODIS, MTG, and Sentinel-3 fusion (EFFIS).
• JRC GWIS as the global system (GWIS).

Most operational services (USFS, NIFC, EU JRC, BoM) use internal fusion systems that ingest several streams and filter artefacts using regional masks.

Regional solutions: which platforms dominate where

FRP as an intensity and emissions metric

Fire Radiative Power is more than an observational descriptor; it is the foundation of emissions estimation. Wooster et al. (2005) demonstrated that FRP scales linearly with biomass combustion rate, with a coefficient near 0.453 kg of fuel per MJ of radiative energy (Wooster et al., 2005, JGR). Integrating FRP over time yields Fire Radiative Energy (FRE), which combined with emission factors gives the mass of CO, CO₂, NO_x, and PM2.5 emissions.

This is the foundation of modern fire-emission inventories: GFED (Global Fire Emissions Database), GFAS (Global Fire Assimilation System from Copernicus), and QFED (Quick Fire Emissions Dataset). All three use MODIS and VIIRS FRP as input. Andela et al. (2017) provide a detailed GFED4s overview and justify FRP as the primary metric (Andela et al., 2017, Earth System Science Data).

Future platforms: 2026-2030

The next generation is planned for launch within the next five years.

• JPSS-3 (NOAA-22) continues VIIRS, with launch planned for 2027.
• MTG-I2 is the second Meteosat Third Generation satellite, with launch in 2026.
• Sentinel-3C and Sentinel-3D extend SLSTR through 2030+.
• Sentinel-2C and Sentinel-2D are optical satellites with 10-20 m resolution used for post-event burned-area analysis (not active detection).
• Landsat Next continues the Landsat series with an improved TIRS-2 thermal channel, launch in 2030+.
• FireSat (commercial initiatives) are announced small-satellite constellations with a fire focus and pass frequencies above 30 per day. Not fully operational in 2026.

The general trend is multi-satellite constellations with pass frequencies above 1 per hour for most populated regions by 2030.

Ukraine: which platforms WildFiresUA integrates

Ukraine has no fire-class satellites of its own and depends entirely on international data streams. The technically available platforms for Ukraine include:

• VIIRS aboard NOAA-20, NOAA-21, and Suomi NPP, providing 4-6 passes per day for territories at 44-52° N.
• MODIS aboard Aqua and Terra, with 2-4 passes (legacy record for historical analysis).
• Sentinel-3A and Sentinel-3B SLSTR, with 1-2 passes. The F1 channel does not saturate for large fires.
• Meteosat Third Generation MTG-I1, a geostationary platform over Africa and Europe that covers Ukraine with 10-minute updates.

WildFiresUA, the national fire service built by the YourAirTest team in partnership with Oles Honchar Dnipro National University (DNU) and EcoCity, integrates VIIRS and MODIS through NASA FIRMS, with additional Sentinel-3 SLSTR for unsaturated intensity estimation of large fires. MTG FCI integration as a fourth input stream for high-frequency monitoring is in progress in 2026. The active-fire map fusing these streams is publicly available at partner.yourairtest.com/map.

Ukraine requires multiple simultaneous streams for four reasons: heavy winter cloud cover (a single pass may be obscured), high zenith angles in the north (resolution degrades to 1.5 km), persistent industrial hot spots (eastern metallurgy, thermal power plants, refineries require masking), and preservation of the archival MODIS record for long-term statistics.

Summary

No satellite platform delivers a complete fire picture on its own. VIIRS provides the best polar resolution at 375 m, MODIS holds the longest historical record, Sentinel-3 SLSTR yields the most accurate intensity estimate for large fires, and GOES-R and MTG offer sub-minute temporal resolution. The current operational approach is fusion of all available streams with contextual validation and regional artefact masks.

For Ukraine, the optimal architecture uses VIIRS as the small-fire detection core, MODIS for historical-record support, Sentinel-3 for large-front intensity calibration, and MTG for high-frequency dynamic monitoring. WildFiresUA implements this architecture. The next generation of platforms (JPSS-3, MTG-I2, Sentinel-3C, FireSat) coming in 2027-2030 will increase pass frequency and reduce publication latency, further shrinking the operational response window.

Ukrainian startup ecosystem: follow TechUkraine and AIN.ua — the two leading outlets covering Ukrainian deep tech, climate tech, and environmental startups.

Related reading on yourairtest.com

• WildFires page of the WildFiresUA project
• How the national air quality monitoring system is built
• YourAirTest data sources

MASK0

Related reading — other scientific reviews

• The global health burden of forest smoke: GBD 2024, Lancet Planetary Health 2025, regional differences
• Forest smoke emission factors: Andreae 2019, Akagi 2011, and the global GFED database
• The global fire detection ecosystem 2026: satellites, sensor networks, AI, drones
• Physical wildfire spread models 2026: Rothermel, FARSITE, ELMFIRE, FOFEM, CAWFE — how they compare

What to do today

1. Check the YourAirTest air quality map for your city — recent PM2.5 readings.
2. If the topic resonates, share with colleagues; we monitor referrals via Google Search Console.
3. To contribute data (sensor measurements, regional models) reach out via the contact form.

References

1. Roberts G., Wooster M.J., Perry G.L.W. et al. (2003). Annual and diurnal African biomass burning temporal dynamics. BAMS.
2. Wooster M.J. (2002). Small-scale experimental testing of fire radiative energy. Geophysical Research Letters 29(21).
3. Kaufman Y.J., Justice C.O., Flynn L.P. et al. (1998). Potential global fire monitoring from EOS-MODIS. JGR 103(D24): 32215-32238.
4. Giglio L., Descloitres J., Justice C.O., Kaufman Y.J. (2003). An enhanced contextual fire detection algorithm for MODIS. RSE 87(2-3): 273-282.
5. Wooster M.J., Roberts G., Perry G.L.W., Kaufman Y.J. (2005). Retrieval of biomass combustion rates from FRP. JGR 110(D24).
6. Schroeder W., Oliva P., Giglio L., Csiszar I.A. (2014). The New VIIRS 375m active fire detection data product. RSE 143: 85-96.
7. Giglio L., Schroeder W., Justice C.O. (2016). The collection 6 MODIS active fire detection algorithm. RSE 178: 31-41.
8. Wooster M.J., Xu W., Nightingale T. (2021). Sentinel-3 SLSTR active fire and FRP product validation. Remote Sensing 13(9): 1627.
9. Schmidt C.C., Hoffman J., Prins E.M. et al. (2017). GOES-R ABI fire detection algorithm. Atmospheric Environment 150: 98-113.
10. Csiszar I., Schroeder W., Giglio L. et al. (2014). Active fires from VIIRS. JGR Atmospheres 119(2): 803-816.
11. Hall J.V., Loboda T.V., Giglio L. et al. (2019). A MODIS-based burned area assessment for Russian croplands. IEEE TGRS.
12. Xu W., Wooster M.J., Polehampton E. et al. (2020). Sentinel-3 SLSTR fire detection. RSE 248: 111870.
13. Wickramasinghe C., Wallace L., Reinke K., Jones S. (2020). Validation of Himawari-8 fire product. Remote Sensing 12(6): 978.
14. Andela N., Morton D.C., Giglio L. et al. (2017). The Global Fire Atlas. Earth System Science Data 9: 697-720.
15. NASA VIIRS Active Fires product page.
16. NASA MODIS MOD14 product documentation.
17. ESA Sentinel-3 mission page.
18. EUMETSAT MTG mission documentation.
19. NOAA GOES-R series programme.
20. NOAA STAR — Center for Satellite Applications and Research.
21. NASA FIRMS — Fire Information for Resource Management System.
22. Copernicus EFFIS.
23. JRC GWIS.
24. EUMETCast distribution platform.
25. NOAA JPSS / VIIRS programme.