21 August 2023, 03:47 Pacific time. Above the San Bernardino ridge in Southern California, three DJI Matrice 350 RTK drones with Zenmuse H30T thermal cameras hum across the night sky. The fire that erupted fifteen hours earlier from a power line technical failure has consumed about 1,800 hectares and broken through the first containment line. A crewed NIROPS aircraft is scanning the neighboring sector at the same time. The drones close the tactical gap: they have built an infrared map of the front line at 25 cm per pixel, stream it live to the CAL FIRE command center, and within 47 minutes the operation has shifted from evening chaos to ordered planning of morning attacks. The sector commander later called this the “first night when we knew where the fire actually was, instead of guessing”.
Drones in fire monitoring are not a new concept. The first experimental UAV flights for forest fire detection were conducted as early as the 1990s by NASA Dryden Flight Research Center and USFS. What has changed is the convergence of three trends: cheaper sensors (a thermal camera with NETD < 50 mK now costs $5,000-15,000 against $200,000 in 2010), mature flying platforms (DJI Matrice, Skydio X10, Quantum-Systems Trinity Pro), and the emergence of regulatory frameworks for BVLOS (Beyond Visual Line Of Sight). In this article we examine where drones deliver real operational value, where they do not, what global regulation looks like, and why WildFiresUA deliberately does not use drones as the primary detection layer in wartime.
Drone detection architecture: platform and sensor types
UAVs for fire monitoring can be categorized along three axes: platform type, sensor payload, and operational integration mode.
Platform type. Multirotor: most of the current fleet, including DJI Matrice 350, Autel Evo Max 4T, Skydio X10. Flight time 35-55 minutes, range up to 8-15 km, vertical takeoff. Fixed-wing: Quantum-Systems Trinity Pro, AeroVironment Quantix, Wingtra One. Flight time 90-180 minutes, range up to 50-100 km, requires a runway or hand launch and lands by parachute or net. Hybrid VTOL: a compromise with vertical takeoff like a multirotor and long flight like an aircraft, for example Aurelia X8 PRO or Wingcopter 198. Tethered drones: specialized solutions for prolonged observation with cable power, where flight time is limited only by electricity availability.
Payload. High-resolution RGB is the basic sensor, suited for daytime observation. Thermal LWIR cameras (long-wave infrared, 8-14 microns) are the foundation of nighttime detection, with typical detectors FLIR Boson, Vue, and Hadron. High-quality drone thermal cameras have NETD < 40 mK and calibrated radiometry for absolute temperatures. Multispectral sensors (MicaSense Altum, Sentera Quad) are used for post-factum burned area assessment. Hyperspectral cameras are rare due to weight and cost. LiDAR (Velodyne Puck, Hesai XT-32) is used to build 3D models of terrain and canopy, important for analyzing fire spread through canopy. Gas sensors are experimental platforms with electrochemical or PID detectors for sensing CO, CO2, NOx in smoke.
Integration mode. Tactical BVLOS as part of an active fire operation is the most mature scenario in the USA and Canada. Strategic monitoring of high-risk areas in the seasonal window is maturing in the Mediterranean. Edge AI on board with autonomous patrolling is still research. Drone swarms with networked coordination are at an early research stage.
Allison et al. (2016) gave an early systemic assessment of drone architectures for fire monitoring and formulated four key effectiveness metrics: detection latency, coverage completeness, operational hourly cost, and integration compatibility with existing ICS systems (Allison et al., 2016, ICUAS).
USA: fleet, regulation, operational practice
The United States has the most developed system of operational drone use in the fire environment. Coordination is provided by the USFS UAS Program (at the federal level) and regional agencies such as CAL FIRE, CO Division of Fire Prevention and Control, and AZ DFFM.
The regulatory environment is set by FAA Part 107 for commercial and government operators with aircraft under 25 kg (55 lb). Part 107 sets baseline requirements: daytime flights, within visual line of sight (VLOS) of the pilot, no higher than 122 m (400 ft) AGL, and outside controlled airspace without permission. For operational fire use these limits are often extended through special waivers: nighttime flights (109.29 waiver), BVLOS operations, and flights inside TFRs (Temporary Flight Restrictions) over active fires.
The USFS UAS Program describes its application doctrine in the program document USFS UAS. Main scenarios: front-line reconnaissance at night to update tactical maps, monitoring of point hot spots after the main attack, risk assessment for firefighters, and burned area mapping for post-operational analysis. Drones are NOT used for direct firefighting at large scale (this remains the role of CL-415 aircraft and C-130 MAFFS).
A separate regulatory layer is the limit on civilian drone flights over active fires. FAA TFRs are automatically established over active suppression zones, and an illegal private drone flight in such a zone can block air operations for hours. Fines run from $20,000 to $75,000, and in some cases criminal charges follow.
The NASA UAS Traffic Management (UTM) project is working on a standard for integrating drones into managed airspace. NASA Ames UTM describes an architecture with identification, route planning, and deconfliction. For the fire domain this opens the prospect of automatic coordination between drones from different agencies in a shared zone.
Canada and EASA: the European regulatory alternative
In Canada, UAV regulation at the federal level is set by Transport Canada in the document CARs Part IX. Key categories: Basic operations (low risk), Advanced operations (higher category, requires an RPAS pilot certificate), and the Special Flight Operations Certificate (SFOC) for non-standard operations such as BVLOS.
Natural Resources Canada (NRCan) and the Canadian Interagency Forest Fire Centre (CIFFC) integrate drones into fire operations using a model similar to the American one, but at a smaller deployment scale. The provincial agencies of British Columbia and Alberta have the most experience with operational BVLOS flights over fires.
In the European Union, regulation is unified by the European Union Aviation Safety Agency (EASA) in the document UAS Implementing Regulation 2019/947. Categories: Open (low risk, up to 25 kg, VLOS), Specific (higher risk, requires SORA assessment), and Certified (treated like crewed aviation). For operational fire monitoring the typical category is Specific with STS-01 (BVLOS in sparsely populated areas) or STS-02 (BVLOS in urban areas).
Greece, Spain, Portugal, and Italy have national programs for drone-based forest monitoring. After the catastrophic 2018 season (the Mati fire, 102 fatalities), Greek Civil Protection invested significant funds into tactical drone reconnaissance. Spain’s INFOCA (Andalusia) integrates drones into regional coordination centers. The Portuguese SCRAM project, in partnership with the University of Coimbra, pilots autonomous drone patrols in the Serra da Estrela mountains.
Market price: the European drone market for public safety and crisis response grows by 12-18% per year according to Drone Industry Insights, although exact figures for the fire segment depend on methodology.
Australia and the Pacific region
Australian regulation is set by the Civil Aviation Safety Authority (CASA) Part 101. Operational fire service operators need a Remote Pilot Licence and a ReOC. The Bushfire CRC (now Natural Hazards Research Australia) and CSIRO have run numerous research programs on drones for fire monitoring. Particular focus is on integration with the Spark and Phoenix predictive systems that model fire spread.
Australia, with its enormous territory and long fire season, is experimenting with high-altitude long-endurance (HALE) platforms. The Bluebottle Aviation program tested a commercial reconnaissance aircraft with thermal cameras for patrolling Western Australia. The NSW Rural Fire Service operates a tactical DJI Matrice fleet for front-line reconnaissance during the Black Summer of 2019-20 and following seasons.
A separate case is New Zealand, where the Department of Conservation experiments with drones for monitoring hard-to-reach areas of national parks.
Thermal drones: physics and limits
Why does a thermal camera see a fire better than an RGB camera? Active combustion radiates strongly in the mid-infrared range 3-5 microns and the long-wave range 8-14 microns. Planck’s law gives a radiation maximum at 800-1,000 K at a wavelength of 3-4 microns. Most drone thermal cameras work in the 8-14 micron band (LWIR), optimized for contrast against the ground background at typical surface temperatures.
The key metric is the Noise Equivalent Temperature Difference (NETD), the smallest temperature difference the sensor can detect. Modern drone thermal cameras have NETD < 40-50 mK at f/1.0. This is enough to detect a smoldering fire (300-400 °C) at 200-500 m depending on atmospheric conditions and fire size.
The main limits of drone thermal detection: atmospheric transmission drops in smoky conditions and at high humidity; calibration for absolute temperature requires periodic in-flight recalibration; cheap sensors without thermoelectric cooling have significant temperature drift in flight; distinguishing between an active fire and heated rocky surfaces requires multispectral analysis.
Restas (2015) gave an early review of optimal sensor bands and configurations for drone fire reconnaissance (Restas, 2015, Procedia Engineering). Tang et al. (2015) described integrated sensor networks with UAV nodes for early-stage fire detection (Tang et al., 2015, Sensors). More recent work by Yuan et al. (2023) gave a critical assessment of thermal drone detection accuracy in complex topographic conditions (Yuan et al., 2023, Remote Sensing).
Drones vs satellites vs ground cameras: when each one wins
A common mistake in discussions is the assumption that drones replace other detection layers. They do not. Drones, satellites, ground cameras, and crewed aircraft do different jobs and have different economics.
Time horizon. VIIRS/MODIS satellites give global coverage with a 3-12 hour gap between passes. Geostationary satellites (GOES, MTG) refresh every 5-15 minutes for their hemisphere. AlertWildfire cameras give continuous observation along their stations. Drones provide episodic coverage limited by 35-180 minutes of flight time and range.
Spatial coverage. One polar satellite per pass covers a 2,300×600 km area. A geostationary satellite covers half a hemisphere. An AlertWildfire camera covers a 15-30 km radius depending on terrain. A drone with a fixed launch point covers a 5-15 km radius for multirotor or 30-50 km for fixed-wing. Deploying drones to a new area requires logistics.
Spatial detail. VIIRS: 375 m pixel. Geostationary: 2 km. Sentinel-2: 10-20 m. AlertWildfire: variable, the effective pixel is typically 1-5 m at detection range. A drone with RGB delivers 1-5 cm per pixel; with thermal, 10-50 cm. Drones lead by a wide margin on detail.
Operational cost. MODIS/VIIRS/Sentinel satellite data are free for public users. Camera networks like AlertWildfire have capital costs of $50,000-200,000 per point plus operations. A drone in flight costs $200-2,000 per flight hour with pilot, full service, and logistics. A NIROPS crewed aircraft costs $5,000-15,000 per flight hour.
Regulatory limits. Satellites operate by natural order. Camera networks operate without flight regulation. Drones are limited by aviation rules and (in war zones) by special restrictions.
Conclusion in tabular form: drones win for tactical front-line reconnaissance at night, for post-fire assessment of local plots, for firefighter safety on terrain, and for infrastructure risk assessment. Drones do NOT win for large-scale early detection, for global monitoring, or for forecasting spread over a large area. These are different layers of the architecture, not competing solutions.
Regional comparison of operational use
| Country | Regulation | BVLOS | Tactical night flights | ICS integration |
|---|---|---|---|---|
| USA | FAA Part 107 + waivers | Established, via waivers | Standard practice | Mature, USFS UAS |
| Canada | Transport Canada CARs IX | Through SFOC | Regular (BC, AB) | Mature, CIFFC |
| EU (overall) | EASA 2019/947 | Specific category, STS-01/02 | Growing | Variable by country |
| Greece | EASA + HCAA | Expanding | Regular | Civil Protection centers |
| Australia | CASA Part 101 | ReOC + waivers | Regular (NSW, VIC, WA) | Mature, AFAC coordination |
| Brazil | ANAC RBAC-E 94 | Limited | Experimental | Spot pilots, IBAMA |
The pattern: the level of operational drone integration scales with (a) the fire service budget, (b) the density of fire events per territorial unit, and (c) the maturity of the regulatory environment for BVLOS.
Battery, weather, operator: physical limits
Beyond regulation, drones face hard physical limits that constrain their role as the primary detection layer.
Flight time. Lithium-polymer batteries have a specific energy density around 250-300 Wh/kg. This is a physical limit that new chemistries (solid-state, metal) may push by 30-50% over 5-10 years, but not by an order of magnitude. A 25 kg multirotor class can stay airborne for 35-55 minutes with a thermal payload. Continuous nighttime monitoring requires 4-6 battery swaps per fire cycle, or a flying relay scheme with several drones.
Weather limits. Strong wind (>10-12 m/s for most multirotors) makes flight unstable or unsafe. Storm, rain, and icing mean a full halt. High temperatures (>40 °C) lower motor efficiency and shorten flight time. The paradox: fires arise most often in those very weather conditions (heat, dry wind) that make drone operations the most difficult.
Operator costs. An operational BVLOS flight requires an RPL/PIC pilot, an observer, a technician, and an airspace coordinator. One accounting cycle takes 4-6 people for one drone point. Pilot training takes 6-18 months to full qualification. At the peak of the fire season, the human resource becomes a critical bottleneck.
Logistics. The drone has to be brought to the fire. On a typical Australian or Californian territory this takes 1-4 hours of preparation and transport to a remote point. In areas without road networks, helicopter or foot delivery is added.
Cumulative result: the operational cost of drone flight time in the fire environment is $500-3,000 per flight hour with all expenses, comparable to a helicopter and cheaper than a NIROPS aircraft, but more expensive than satellite detection (a near-zero marginal cost per fire episode).
Drones in a war zone: why Ukraine cannot rely on them
The Ukrainian context differs sharply from the Californian or Greek one. Since 24 February 2022, a significant part of Ukrainian territory has been under combat, and the rest under periodic air threats. This creates three fundamental limits on operational drone-based fire monitoring.
Anti-drone EW. The Russian side actively uses electronic warfare (EW) systems that suppress GPS signals, command channel control (typically 2.4 and 5.8 GHz), and the video downlink. Systems such as “Pole-21”, “Strepet”, and “Shipovnik-AERO” generate jamming at distances up to 30 km. A civilian fire-fighting drone without an anti-jam architecture simply will not work in such a zone.
Air defense systems. Both Ukrainian and Russian air defense systems may interpret an unidentified drone as hostile and shoot it down. Coordination with military structures for an operational civilian drone flight in any active zone is a complex bureaucratic process with delays of many hours. For operational fire response this effectively rules out drone reconnaissance in front-line oblasts.
Front-line restrictions. Under regulations governing restricted airspace in wartime, civilian UAV flights are forbidden or substantially limited within a significant radius of the line of contact. An operations permit requires coordination with the Air Force Command, the General Staff, and regional military administrations. Processing time runs from several days to weeks, which is unacceptable for operational detection.
Economic vulnerability. Loss of a $20,000-200,000 drone at a fire is not just a financial blow but a potential compromise to the detection system as a whole. In peacetime such risk is acceptable as insurable. In wartime, where the drone fleet is a strategic resource for defense needs, every civilian observation unit is a potential withdrawal from military use.
Personnel pressure. Qualified drone operators draw from the same talent pool that the Defense Forces actively need. Deploying a sizeable civilian drone fire surveillance service during a full-scale war competes with military demand.
For these reasons WildFiresUA deliberately does NOT use drones as an operational detection layer. Instead, we rely on: (1) satellite data from VIIRS, MODIS, Sentinel-3, and MTG-I1, with global coverage and no flight regulation; (2) ground sensor networks for front-line regions, less vulnerable to EW and not requiring airspace; (3) integration with DSES data obtained by ground methods; (4) the FLEXPART/HYSPLIT physical smoke forecast models for predicting where direct observation is impossible. This is a structural choice, not a technological limit.
Postwar perspective on drone integration
We do not exclude drones from the long-term WildFiresUA architecture. In the postwar period, when the regulatory environment stabilizes, the human resource frees up, and the priority of civilian drone monitoring is restored, drone integration will deliver significant benefits.
Specific postwar deployment scenarios: (a) tactical monitoring of large forest fires in Polissia and the Carpathians for tactical support of DSES; (b) post-factum mapping of burned areas in the Chornobyl Exclusion Zone with thermal cameras to assess long-lasting smoldering processes in peat; (c) reconnaissance of point fires in populated areas after missile strikes on infrastructure (a long-term concern for reconstruction); (d) piloting drone patrols in regions of elevated risk (the southern steppe oblasts, the Azov region).
From a regulatory standpoint this will require harmonizing Ukrainian law with EASA standards and developing operational integration protocols with DSES and Civil Protection. Technologically, it requires investment in professional-grade thermal sensors, training of pilot personnel, and deployment of charging and storage infrastructure in regional centers.
European partners from the Horizon Europe program and the EU Civil Protection Mechanism are ready to support such integration after the active phase of the war ends. Current cooperation with Polish, Romanian, and Czech civilian drone programs is laying the foundation.
Drones in combined architectures: NASA, USFS, NRCan
Modern best practices integrate drones into a multi-layer architecture, where each layer does what it is good at. The most formalized model is in the NASA “Multi-Asset Wildfire Management” document and in parallel in the USFS Aviation Strategy.
The architecture: a global layer (satellites, physical FRP algorithms, hot spot products), a regional layer (ground cameras, local sensors), a tactical layer (BVLOS drones, crewed aircraft), and an agency decision layer (a command center with an ICS system). Drones in this model are the fourth detection line and the second assessment line, not the foundation of detection.
NASA Earth System Pathfinder offers integrated tools for working with all these layers through open APIs. For ML model training, archival datasets from NASA Ikhana thermal flights and USFS NIROPS are available at earthdata.nasa.gov.
A growing research field is the coordination of drone swarms with an observing satellite orbit. Yu et al. (2024) described an experimental scheme where VIIRS detection triggers an automatic drone platform sortie for tactical verification; the latency from satellite pass to drone arrival on target runs 18-45 minutes depending on basing distance (Yu et al., 2024, International Journal of Remote Sensing).
Conclusion
Drones are a strong tool for tactical fire reconnaissance, well suited to front-line assessment at night, infrastructure risk assessment, and post-factum mapping. Drones are NOT the foundation of global early detection: there satellites with physical algorithms and camera networks remain better in cost and coverage.
The FAA Part 107 regulatory framework in the USA and EASA 2019/947 in the EU are mature and create a working environment for operational BVLOS. Australia, Canada, Greece, Portugal, and Spain have practical experience integrating drones into fire agencies. Brazil and other countries are at an early stage.
For Ukraine, the current war context makes operational use of drones by a civilian fire monitoring service impractical: EW, air defense, regulatory limits, and personnel competition with defense rule it out. WildFiresUA deliberately relies on satellites, ground sensors, and physical models as a more reliable layer under current conditions. The postwar prospect of drone integration remains open and we are preparing infrastructure for future deployment.
Ukrainian startup ecosystem: follow TechUkraine and AIN.ua — the two leading outlets covering Ukrainian deep tech, climate tech, and environmental startups.
Related reading
- Global Scientific Review of fire detection 2026
- Wildfires — our fire monitoring platform
- YourAirTest air quality monitoring system
MASK0
Related reading — other scientific reviews
- Satellite fire detection 2026: VIIRS, Sentinel-3 SLSTR, MODIS, GOES-R/MTG, FY-3
- AI fire detection in satellite data and video streams 2024-2026: deep learning state of the art
- Physical wildfire spread models 2026: Rothermel, FARSITE, ELMFIRE, FOFEM, CAWFE — how they compare
- Atmospheric smoke plume dispersion: HYSPLIT, FLEXPART, SILAM, NAME, CALPUFF — the Lagrangian vs Eulerian battle
What to do today
- Check the YourAirTest air quality map for your city — recent PM2.5 readings.
- If the topic resonates, share with colleagues; we monitor referrals via Google Search Console.
- To contribute data (sensor measurements, regional models) reach out via the contact form.
References
- FAA Part 107 — Small Unmanned Aircraft Systems. faa.gov/uas/commercial_operators/part_107.
- FAA Temporary Flight Restrictions documentation. faa.gov/uas/getting_started/temporary_flight_restrictions.
- EASA UAS Implementing Regulation 2019/947. easa.europa.eu/en/domains/civil-drones-rpas.
- Transport Canada CARs Part IX (Remotely Piloted Aircraft Systems). tc.canada.ca/en/aviation/drone-safety.
- USFS UAS Program. fs.usda.gov/managing-land/fire/uas.
- USFS Aviation Strategy. fs.usda.gov/managing-land/fire/aviation.
- NASA Ames UAS Traffic Management. nasa.gov/ames/utm.
- NASA Earth Science Data Systems. earthdata.nasa.gov.
- Allison, R.S., Johnston, J.M., Craig, G., Jennings, S. (2016). Airborne optical and thermal remote sensing for wildfire detection and monitoring. 2016 International Conference on Unmanned Aircraft Systems (ICUAS).
- Restas, A. (2015). Drone applications for supporting disaster management. Procedia Engineering, 107, 316-321.
- Tang, J., Yu, J., Liu, S. et al. (2015). A wireless sensor network for early forest fire detection: design and validation. Sensors, 15(1), 319-336.
- Yuan, C., Zhang, Y., Liu, Z. (2023). Forest fire detection accuracy of drone-based thermal imaging in complex terrain. Remote Sensing, 15(9), 2339.
- Yu, K., Wang, P., Hou, Q. et al. (2024). Coordinated satellite-UAV wildfire monitoring with adaptive triggering. International Journal of Remote Sensing, 45(8).
- NASA Earth Observatory feature on UAVs in wildfire response. earthobservatory.nasa.gov.
- NSW Rural Fire Service operational publications on UAS integration. rfs.nsw.gov.au.
- CSIRO Bushfire Research and operational drone studies. csiro.au/en/research/natural-disasters/bushfires.
- Natural Hazards Research Australia (formerly Bushfire CRC). naturalhazards.com.au.
- Greek Civil Protection — Drone integration for wildfire management. civilprotection.gov.gr.
- INFOCA Andalucia (Spain) operational documentation. juntadeandalucia.es/medioambiente.
- Portuguese Civil Protection ANEPC. prociv.pt.
- EU Civil Protection Mechanism. civil-protection-humanitarian-aid.ec.europa.eu.
- JRC EFFIS European Forest Fire Information System. forest-fire.emergency.copernicus.eu.
- NASA FIRMS Fire Information for Resource Management System. firms.modaps.eosdis.nasa.gov.
- CIFFC Canadian Interagency Forest Fire Centre. ciffc.ca.
- BC Wildfire Service operational drone documentation. gov.bc.ca/wildfire-status.
- EUROCAE WG-105 standards on UAS detection and avoidance. eurocae.net.
- ICAO Remotely Piloted Aircraft System framework. icao.int/safety/UA.
- Drone Industry Insights market reports on public safety UAS sector. droneii.com.