Twelve minutes. That is how long a system of sensors and drones took to detect, locate and put out a fire in a demonstration by Dryad Networks, according to a statement the German company published on 4 November 2025. The company says the sequence was completed without direct human intervention [1].
It was a controlled trial and it does not set a universal benchmark: how a fire behaves depends on fuel, terrain and weather. But it points to the variable that weighs most in suppression and has the most room for improvement: time. In this article we show it with official data and with a simulator built on the same physical model used by fire behaviour prediction tools.
The battle of the first minutes
Spain’s 2025 season makes it clear. Of 8,199 fires, 5,601 were put out before they burned one hectare: these are known as incipient fires (conatos). At the other end, 63 large wildfires —0.8% of the total— burned 310,108 hectares, 87.4% of the entire area affected [6].
8,199 fires in 2025: 5,601 incipient fires (68.3%), 2,535 fires of 1 to 500 ha and 63 large fires (0.8%) that burned 87.4% of the area.
The operational takeaway is simple: most of the damage is caused by a few fires that escape initial attack. Every minute gained before that moment is area that does not burn.
Why every minute counts: fire grows with the square of time
With steady wind and fuel, a fire spreads at an almost constant rate and its outline approximates an ellipse stretched in the direction of the wind [7] [9]. If length and width grow in proportion to time, the area grows with its square: at 40 minutes, a fire does not cover twice the area it covered at 20, but four times as much.
Response time has two parts: the time it takes to detect and confirm the fire, and the time it takes for crews to arrive. The drone acts on the first. The simulator compares two scenarios with the same fire and different detection times. Change the fuel, the wind or the moisture and see how much area the first attack faces in each case.
Interactive simulator: enable JavaScript to compare the area of the same fire with two different detection times.
The simulator times are examples. In a real situation they depend on the territory, the surveillance network and the resources available. What does not change is the relationship between time and area. On an extreme-risk day, cutting minutes from detection can mean crews find an incipient fire instead of a wildfire.
Where the drone fits: from prevention to flare-ups
Detection is only one part. Integrated into procedures and under the coordination of fire services, a drone can provide information across the whole response chain. It also has limits, and they are worth knowing. Some initiatives focus on prevention, such as the Chilean company Drovid, which proposes using drones and artificial intelligence to detect risk situations [2]. Go through each step:
- Prevent: visible patrols at access points and detection of risky activities.
- Detect: thermal camera to find fires before smoke is visible.
- Confirm: live video and coordinates for the coordination centre.
- Support the attack: updated perimeter and hotspots, including at night.
- Watch for flare-ups: thermal sweeps after the fire is under control.
There is accumulated experience. The U.S. Forest Service UAS programme report records more than 17,000 flights in fiscal year 2024. The figure covers the whole programme, including fire management, resource management and training; it does not refer only to suppression flights [3].
More than 678,000 hectares burned in the European Union
The European Union had recorded 678,978 hectares burned as of 23 September 2026, above the 2006–2025 average, according to the Joint Research Centre [4].
In Spain, the 2026 national deployment includes 56 aircraft provided by the Ministry for the Ecological Transition and the Demographic Challenge (MITECO), in addition to regional resources and those of other agencies [5]. Drones should strengthen that system through defined missions, reliable communications and coordination with crewed aircraft.
Prolonged surveillance also requires organisation. An aircraft’s endurance does not, on its own, amount to permanent surveillance: it takes relays, power, personnel and a system able to receive and use the information.
The role of fixed-wing VTOL
Fixed-wing VTOL combines vertical take-off and landing with the flight of an aeroplane: it can operate without a runway and cover large areas with more endurance than a multirotor. It is the profile a preventive patrol over a large forest calls for.
IMS Drones works with platforms of this type, such as KITE and Falziot. An example mission: on a high-risk day, an aircraft takes off vertically from a forest access point and flies a scheduled patrol over hard-to-reach areas. With optical and thermal cameras, it can detect a possible fire or a risky activity, pinpoint it and send images and coordinates in real time to the coordination centre, which decides how to act.
The goal is for an alert to become useful information and reach those who must respond in time. To achieve this, technology has to be integrated with the professionals, procedures and resources already working on the ground. That is where drones add value: they help to see sooner, see better and decide with more information. You can find more applications on our page on drones for wildfires.
Sources
- Dryad Networks. Statement of 4 November 2025 on the detection and suppression demonstration. dryad.net
- Ministry of Economy of Chile. “Startup chilena es finalista de los WIPO Global Awards 2026 por tecnología de detección temprana de incendios forestales”, 26 May 2026. economia.gob.cl
- USDA Forest Service. UAS programme report, fiscal year 2024 (FY24). fs.usda.gov
- Joint Research Centre. “Current wildfire situation in Europe”, data as of 23 September 2026. joint-research-centre.ec.europa.eu
- Spanish Ministry of the Interior. “Interior adelanta al 1 de junio el comienzo de la campaña estatal estival contra incendios forestales”, 21 May 2026. interior.gob.es
- MITECO. “Los incendios forestales en España. Avance informativo, 1 de enero a 31 de diciembre de 2025”, May 2026. miteco.gob.es
- Rothermel, R. C. (1972). A mathematical model for predicting fire spread in wildland fuels. USDA Forest Service, Res. Pap. INT-115. fs.usda.gov
- Anderson, H. E. (1982). Aids to determining fuel models for estimating fire behavior. USDA Forest Service, Gen. Tech. Rep. INT-122. fs.usda.gov
- Anderson, H. E. (1983). Predicting wind-driven wild land fire size and shape. USDA Forest Service, Res. Pap. INT-305. biodiversitylibrary.org
- Albini, F. A. (1976). Estimating wildfire behavior and effects. USDA Forest Service, Gen. Tech. Rep. INT-30. fs.usda.gov
- Europa Press (via Infobae). “La Fiscalía constata un incremento de más de 8 puntos en los incendios intencionados en 2025, el 32% del total”, 10 September 2026. infobae.com
Sources accessed on 11 October 2026. Dryad Networks’ results are reported by the company itself. The simulator applies models [7]–[10] and has been checked against the reference tables in [8]; the cause figures in the “Prevent” step come from [11].

