Technical notebook 01 · Wildfire dynamics

How Far Can Wildfire Embers Travel?

The physics of spotting, firebrand showers and WUI ignition. A firebrand may cross a fuel break and fail to ignite anything. Another may travel a shorter distance, lodge in a dry joint and initiate a new fire after the flame front has passed. The operational question is therefore not distance alone, but the entire source–transport–receptor chain.

Research synthesis · August 2026
Two interactive laboratories
SI units · reproducible assumptions
Forestry · WUI · incident intelligence

01 · GenerateFuel fragments, heat, detach and enter an updraft.
02 · LoftThe plume lifts a shape-dependent population.
03 · TransportWind, turbulence and combustion reshape the trajectories.
04 · DepositParticles land, bounce, roll or accumulate in traps.
05 · IgniteA receptive fuel sustains smouldering or flaming combustion.

Executive brief

The maximum landing distance is not the effective ignition distance

Spotting is a conditional process. A model that calculates where particles land but assumes that every landing becomes a fire is a transport model with an ignition shortcut—not a complete spot-fire model.

P_spot(x) = P_gen · P_trans(x) · P_active(x) · P_dep(x) · P_ign(x | receptor)
Conceptual factorisation. The terms are generally correlated and should not be treated as independent without evidence.

A distribution, not one ember

Real sources emit populations with correlated mass, projected area, shape, temperature, combustion state and release height.

A changing particle

Mass loss changes terminal velocity and trajectory; orientation and tumbling change drag and lift during the flight.

A receiver, not bare ground

Needle beds, mulch, vents, roofs and timber joints retain heat and particles in radically different ways.

Safety boundary. None of the distances or indices on this page is a safety distance, an escape threshold or a forecast. Spotting can create fire behind crews, compromise egress and invalidate a geometry based only on the main flame front.

Terminology and measurement
Generation from forest and structural fuels
Flight physics
Laboratory 1: flight envelope
Deposition and ignition
Laboratory 2: receptor challenge
Vegetation and WUI interpretation
Operational and design implications
Validation and uncertainty
Primary references

01 · Define before calculating

Ember, firebrand and spotting are not interchangeable measurements

In fire science, firebrand normally denotes a hot, glowing or flaming solid particle capable of transport. Ember is used more broadly and often overlaps. Spotting is discontinuous fire spread caused when transported particles produce secondary ignitions ahead of, beside or behind the main fire perimeter. The distinction matters because a particle counter, a char cloth and an ignition tray measure different parts of the chain.

Metric	Symbol	Unit	What it tells us	What it does not tell us
Cumulative particle count	CPC	m⁻²	Total exposure received over an interval.	Instantaneous peaks or thermal state.
Particle number flux	N″	m⁻²·s⁻¹	Arrival intensity per unit area and time.	Mass or energy delivered.
Mass flux	ṁ″_p	kg·m⁻²·s⁻¹	Solid fuel loading delivered to the receiver.	Contact quality or combustion state.
Landing-distance quantile	x_p	m	Spatial distribution of simulated or observed landings.	Probability of ignition.
Spot-fire probability	P_spot	0–1	Conditional likelihood for a defined source, weather and receptor.	A universal property of a species or particle.

NIST’s three-dimensional firebrand diagnostics make this separation explicit: particle trajectories, size, shape and combustion state can be measured in time and space, while CPC and particle number flux allow one shower to be compared with another. That is a much stronger basis than describing an exposure as simply “light” or “heavy”.

02 · Source physics

A burning hectare does not emit a standard firebrand

Generation begins when a vegetative or structural element heats, dries, pyrolyses, chars and detaches through thermal cracking, section loss, aerodynamic loading, local collapse or contact with other elements. It then has to enter a flow whose vertical momentum can temporarily overcome its effective weight. Species name and burned mass alone cannot reconstruct that process.

p(m₀, A_p,0, ψ, T₀, χ₀, z₀, t)A source should be described as a joint distribution of initial mass, projected area, shape/orientation, temperature, combustion state, release position and time.

Source control	Physical pathway	Modelling consequence
Fuel architecture	Fine twigs, bark plates, cones, dead branchwood and attached litter create different detachable shapes.	Use source-specific particle distributions, not a single equivalent diameter.
Moisture and live condition	Change heating time, flame residence, charring and mechanical failure.	Any threshold belongs to its species, size and experimental protocol.
Wind at source	Changes ventilation, flame attachment, aerodynamic extraction and plume tilt.	Wind cannot be varied only in the transport stage.
Fire regime	Surface, torching, active crown and structural fire expose different materials and release heights.	Canopy and building sources may contribute larger or differently shaped particles.
Silviculture and maintenance	Alter ladder fuels, deadwood, crown separation and available fragment inventory.	Treatments modify the source distribution as well as spread continuity.

Full-scale tests show why one headline number is dangerous. In a specific still-air Korean pine experiment, more than 500 mostly cylindrical particles were measured, with an average geometry near 5 mm in diameter and 40 mm in length. Trees above roughly 35% dry-basis moisture did not produce significant quantities in that configuration. This is a source-specific observation—not a transferable ignition or generation threshold.

In another programme, 108 trees from four North American species produced about 15,700 collected particles and roughly 4,900 char marks on a defined receptor. The fraction that remained thermally active differed markedly between species. The useful conclusion is not a species ranking for every wildfire; it is that particle yield and active-particle yield are different response variables.

03 · Coupled aerodynamics and combustion

The atmosphere carries an object whose mass, shape and temperature are changing

A firebrand is not a passive tracer. Its trajectory follows the balance between gravity, buoyancy, drag, orientation-dependent lift and unsteady forces inside a velocity field altered by the fire itself. Simultaneously, oxidation and pyrolysis reduce mass and may change projected area. Aerodynamics changes combustion; combustion changes aerodynamics.

m_p dv_p/dt = m_pg − ρ_aV_pg + F_D + F_L + F_uLagrangian momentum balance. A complete model also evolves particle mass, temperature and geometry.

Terminal velocity is a scale, not a universal constant

For a fixed orientation and a steady relative flow, a useful first-order scale is:

v_t ≈ √[2m_pg / (ρ_aC_DA_p)]

A low ratio of mass to projected area promotes lofting. Thin bark flakes can therefore behave very differently from compact charcoal of the same mass. Rods tumble and reorient; recent canopy-flow experiments confirm that rod-shaped particles have distinct transport and dispersion behaviour. Current long-range models also show that the selected drag, lift and mass-loss formulations materially change landing patterns.

The seductive shortcut

If horizontal wind U, effective release height H_e and terminal velocity were constant, with no turbulence or vertical motion, then:

t_f ≈ H_e/v_t and x ≈ U H_e/v_tThis is a sensitivity scale. It is not a maximum spotting-distance equation.

The expression correctly exposes three levers—wind, release height and particle settling—but omits the intermittent vertical velocities that can repeatedly suspend a particle, the tilted plume, shear, canopy turbulence, terrain flows and the evolution of the particle. In coupled FIRETEC simulations, launch and deposition patterns were highly heterogeneous and controlled by fire–atmosphere interaction. In large-eddy simulations, plume turbulence was decisive for long-range dispersion.

Analytical scaleU H_e/v_t; transparent and useful for sensitivity.

Prescribed-flow particleLagrangian trajectory in a measured or modelled wind field.

Stochastic transportAdds unresolved turbulent velocity and parameter distributions.

Coupled fire–atmosphereThe plume and ambient flow evolve with the fire.

Reactive multiphaseTransport, heat, mass loss, morphology and ignition are connected.

Model hierarchy. More equations do not automatically produce more truth. Complexity is useful only when input characterisation, spatial resolution, numerical verification and independent validation improve with it.

Interactive laboratory 01

First-order firebrand flight envelope

Change one variable at a time. The model generates 1,200 reproducible synthetic trajectories and reports distribution quantiles—not a deterministic maximum.

Scope. The laboratory uses a terminal-velocity scale with lognormal aggregate variability. It does not resolve a 3-D plume, terrain, canopy flow, lift, particle rotation or receptor ignition. It is for teaching and sensitivity screening only.

Particle profile bark flake

Effective release height 60 m

Advective wind 12.0 m·s⁻¹

Initial mass 0.60 g

Projected area 700 mm²

Drag coefficient 1.17

Median active lifetime 140 s

Aggregate variability σ 0.55

Fuel-break width b 40 m

Terminal-velocity scale3.49 m·s⁻¹

Median landing distance195 m

90th percentile315 m

Active beyond b100.0%

b = 40 m

0 m600 m

Seeded sample · median flight time 16.6 s · 1,200 trajectories.

Output	Correct interpretation	Incorrect interpretation
Median	Half of this synthetic population lands before it and half after it.	The average distance of every wildfire.
90th percentile	A tail descriptor for the selected assumptions.	A safe outer boundary or physical maximum.
Active beyond b	Fraction crossing b before the sampled active lifetime expires.	Probability of a spot fire; deposition and ignition are absent.

Editor note. Kubio may not execute JavaScript inside the editing canvas. Save the page and open Preview in a new tab; the controls should update there.

04 · From landing to a self-sustained fire

The receiver decides whether transport becomes propagation

Landing is not the end of the trajectory. A particle can bounce, fragment, roll, be blown away or become trapped. Gutters, roof valleys, vents, cladding joints, fence bases, needle pockets, mulch beds and the lee side of rocks convert diffuse exposure into local accumulation.

dN_trap/dt = N″A_trapP_ret − N_trap/τ_resConceptual balance between arrival and loss from an effective trap area.

Multiple particles matter because they can cooperate: early arrivals preheat and dry the receptor, later arrivals add energy, and a sheltered geometry reduces convective losses. A continuous shower of small particles can therefore be more severe for a joint than one isolated larger firebrand.

Receiver	Dominant mechanism	Sensitive variables	Possible outcome
Needles and fine litter	Multiple contacts, porous ventilation and spread between fine elements.	Moisture, depth, compaction, wind and particle state.	Smouldering or flame.
Landscaping mulch	Accumulation in a porous bed with continuing air supply.	Product, particle pile mass, wind, moisture and proximity to structure.	Persistent smouldering and possible transition.
Timber joint or roof gap	Geometric trapping and reduced heat loss.	Gap width, substrate, airflow, number flux and exposure duration.	Hidden smouldering or flaming ignition.
Vent or open cavity	Particle entry transfers the ignition source behind the exterior envelope.	Aperture, mesh, pressure field and internal combustibles.	Internal ignition without direct flame contact.

Experiments explain apparently contradictory moisture results. In one Mediterranean protocol without wind, positive ignition across eleven firebrand–fuel-bed combinations occurred only with flaming particles, and bed moisture dominated the response. In a different NIST Dragon configuration, continuous showers accumulated in shredded wood mulch placed in a re-entrant corner and produced flaming ignition at much higher moisture under strong wind. The correct conclusion is not a universal moisture threshold: ignitability is a property of the complete exposure–receiver system.

No transferable “ignition temperature”. Receiver ignition depends on heat-flux history, contact area and resistance, ventilation, losses, drying, pyrolysis kinetics and the ability to sustain reaction after the firebrand source weakens.

Interactive laboratory 02

Firebrand accumulation and receptor energy challenge

This screening model separates particle arrival, geometric retention and a simplified thermal budget. The result Θ is an energy ratio—not an ignition criterion.

Receiver profile pine needles

Incident particle flux 0.40 m⁻²·s⁻¹

Exposure duration 300 s

Effective trap area 0.030 m²

Retention probability 50%

Mean particle mass 0.600 g

Particle temperature 650 °C

Receiver moisture, dry basis 12%

Receiver dry load 0.35 kg·m⁻²

Effective thermal coupling 25%

Cumulative arrivals120 m⁻²

Mean retained in trap λ1.80

P(at least two retained)53.7%

Energy ratio Θ0.16

Effective particle input

44 kJ·m⁻²

Dry-fuel heating scale

164 kJ·m⁻²

Water-heating/evaporation scale

108 kJ·m⁻²

Reading: waiting for calculation.

Assumptions: Poisson arrivals for the “at least two” illustration; fixed effective retention; lumped particle sensible/oxidation energy; a single coupling factor; receiver heating to a nominal 300 °C plus a water sink. Spatial hot spots, oxygen limitation, radiation, chemistry, losses, contact mechanics and sustained ignition are not solved.

Editor note. If the values do not update inside Kubio, test in WordPress Preview. The public page is the correct execution environment.

05 · Vegetation as source and receptor

Species matters through physical traits—not through a universal distance table

There is no defensible table in which “species × age” yields one spotting distance. Species, height and stand development influence detachable material, crown geometry, deadwood, moisture, release height and receptor structure. To compare vegetation scientifically, separate its two roles.

Vegetation as source

Measure detachable bark and twigs, dry and live mass, morphology, particle size/mass distribution, combustion state, crown height and release process.

Vegetation as aerodynamic layer

Measure canopy height, porosity, gap structure, roughness, sheltering and the interaction between the buoyant plume and canopy mixing layer.

Vegetation as receptor

Measure fine-fuel moisture, bed depth, bulk density, continuity, trap geometry, wind exposure and likelihood of sustained smouldering or flame.

Formation	Source-side questions	Receiver-side questions	Management interpretation
Mediterranean pine and shrub mosaics	Bark and branch morphology, ladder fuels, torching and active crown involvement.	Dry needles, shrub litter, stone-wall and terrace traps, WUI gardens.	Vertical separation and fine-fuel maintenance can alter both release and reception.
Eucalypt-dominated systems	Ribbon bark, leaf material, elevated release and strong convective events.	Dry litter, roof/gutter accumulation and wind-aligned corridors.	Do not transfer particle distributions from pine experiments without validation.
North American conifer WUI	Needle-bearing twigs, bark plates, crown-fire regime and structural fuel contribution.	Mulch, decks, vents, fences, receptive ornamental vegetation.	Community exposure can persist even where parcel vegetation is treated.
Boreal conifer and peatland edge	Crown involvement, dry branchwood and long residence in convective columns.	Feather moss, duff and organic layers capable of sustained smouldering.	Detection must continue after apparent passage of the front.

Spanish applicability. In Iberian landscapes, interpret Pinus halepensis, P. pinaster, P. pinea, P. nigra, eucalypt stands, shrublands and dehesa systems through measured fuel traits and fire regime. The international edition retains this Mediterranean lens but makes the method portable to other fire environments.

06 · Operational intelligence and engineering

A fuel break interrupts surface continuity. It does not put a roof over the atmosphere.

At least three crossing modes must be examined: direct flame or radiation across the gap, firebrand transport over the gap, and growth of any secondary fires that establish beyond it.

Question	Observable evidence	Decision value
Is the source producing an active shower?	Particle flux, char-cloth marks, thermal/video observation, source regime and combustion state.	Defines whether patrol and surveillance depth must extend beyond the nominal line.
Where is the tail going?	Wind profile, plume tilt, shear, terrain channeling, landing quantiles and observed spot locations.	Supports search-sector geometry; it does not define a safe boundary.
What will retain particles?	Needle pockets, mulch, roof valleys, vents, fences, equipment bays and lee-side accumulation.	Prioritises inspection by receptor vulnerability rather than distance alone.
Can a spot become independent?	Fine-fuel moisture, continuity, wind, slope, detection latency and access.	Connects ignition likelihood to consequence and containment opportunity.

A three-layer interpretation for treatment strips

Source sideReduce ladder fuels, detachable dead material and the probability of crown involvement.

Central stripReduce surface intensity and provide access, visibility and suppression opportunity.

Receiver sideRemove traps and receptive fine fuels; harden vulnerable assets and maintain detection.

Depth of patrolUse observed showers and forecast wind to define search sectors beyond the strip.

FeedbackRecord landings, thermal state, receptor and outcome to improve local models.

Telecommunications perspective. A useful sensing architecture synchronises time across optical/thermal cameras, weather stations, particle collectors, drone imagery and incident reports. Without clock discipline, known sensor locations and uncertainty on detection latency, a “spot distance” may combine incompatible timestamps and moving fronts. The data pipeline is part of the fire-science instrument.

For incident teams. A spot fire may appear behind the current position, close a route or establish in untreated fuel before it is visible through smoke. Quantiles from a screening model may support observation hypotheses; they never authorise approach, remaining in place or crossing.

07 · Make the model falsifiable

Separate numerical accuracy, physical validity and operational usefulness

A model can be mathematically converged and still be physically incomplete. Qin and Trouvé’s 2025 framework is valuable precisely because it separates numerical error from physical modelling error; their verification tests use an instantaneous-ignition simplification to obtain exact benchmarks, while explicitly recognising that this is not a realistic receptor model.

Layer	Minimum evidence	Failure signal
Source characterisation	Joint distributions of mass, area, morphology, state and release height by fuel and regime.	One assumed particle represents every source.
Aerodynamics/combustion	Benchmarks for settling, drag/lift, mass loss and sensitivity to time step.	Landing distance changes materially with numerical settings.
Transport field	Measured or validated wind/plume statistics, terrain and canopy representation.	Observed lateral spread or tail behaviour is systematically missed.
Deposition	Landing and retention observations on representative geometries.	All landings are treated as retained.
Ignition	Fuel-specific exposure matrices including moisture, flux, state, wind and duration.	Every retained particle becomes an immediate spot fire.
Field performance	Time-synchronised fire perimeter, spot detections and uncertainty envelopes.	A single maximum is reported without misses, false alarms or latency.

Uncertainty should travel with the result

For operationally relevant outputs, report at least the input range, source of each parameter, model version, quantiles or probability bands, validation domain and known omissions. A value such as “spotting distance = 380 m” is weak. A statement such as “under the defined source distribution and wind profile, 90% of simulated active landings were inside 380 m; ignition was not modelled” is auditable.

Reproducibility note. The first laboratory uses a fixed pseudo-random seed so that the same parameters return the same sample. That is useful for teaching and comparison, not evidence that nature repeats the trajectory set.

08 · What the evidence supports

Ten conclusions worth carrying into design and command

Spotting distance is a distribution conditioned on source, fire–atmosphere flow, particle evolution and terrain.
The furthest landing is not automatically the furthest effective ignition.
Particle mass, projected area and orientation must be retained separately whenever the model allows it.
Wind changes generation, plume geometry, transport, combustion, deposition and receiver ventilation.
Species effects are mediated by measurable traits and cannot be reduced to a universal lookup table.
Particle number flux, cumulative count, mass flux and combustion state describe different exposure dimensions.
Accumulation and receptor geometry can dominate the outcome of an ember shower.
A fuel break is not a firebrand barrier; its receiver side needs its own treatment and surveillance logic.
A simulator must state which links in the generation–transport–ignition chain it omits.
For field use, uncertainty, detection latency and the growth of secondary fires matter as much as landing distance.

Final idea. The rigorous question is not “How many metres can an ember fly?” It is: “What particle population does this source generate, what flow transports it, what fraction remains active, where is it retained, and which receptors can turn that exposure into independent fire during the decision window?”

Primary literature and institutional sources

Selected references

Priority is given to peer-reviewed primary research, validated government research programmes and foundational models. DOI links lead to the publisher record.

Tarifa, C. S., Notario, P. P. & Moreno, F. G. (1965). On the flight paths and lifetimes of burning particles of wood. Symposium (International) on Combustion, 10, 1021–1037. doi:10.1016/S0082-0784(65)80244-2.
Albini, F. A. (1979). Spot fire distance from burning trees—A predictive model. USDA Forest Service, General Technical Report INT-56.
Albini, F. A. (1983). Potential spotting distance from wind-driven surface fires. Combustion Science and Technology, 32, 277–295. doi:10.1080/00102208308923662.
Sardoy, N., Consalvi, J.-L., Kaiss, A., Fernandez-Pello, A. C. & Porterie, B. (2007). Numerical study of ground-level distribution of firebrands generated by line fires. Combustion and Flame, 150, 151–169. doi:10.1016/j.combustflame.2007.04.008.
Sardoy, N., Consalvi, J.-L., Kaiss, A., Fernandez-Pello, A. C. & Porterie, B. (2008). Numerical study of ground-level distribution of firebrands generated by line fires. Part II. Combustion and Flame, 154, 478–488. doi:10.1016/j.combustflame.2008.05.006.
Koo, E., Linn, R. R., Pagni, P. J. & Edminster, C. B. (2012). Modelling firebrand transport in wildfires using HIGRAD/FIRETEC. International Journal of Wildland Fire, 21, 396–417. doi:10.1071/WF09146.
Thurston, W., Kepert, J. D., Tory, K. J. & Fawcett, R. J. B. (2017). The contribution of turbulent plume dynamics to long-range spotting. International Journal of Wildland Fire, 26. doi:10.1071/WF16142.
Manzello, S. L., Maranghides, A. & Mell, W. E. (2007). Firebrand generation from burning vegetation. International Journal of Wildland Fire, 16, 458–462. doi:10.1071/WF06079.
Manzello, S. L. et al. (2009). Quantifying the ignition hazard of firebrands on wildland–urban interface structures. Fire and Materials, 33. doi:10.1002/fam.977.
Hudson, T. R. et al. (2020). Size and energy distribution of firebrands produced from burning trees. International Journal of Wildland Fire. doi:10.1071/WF19182.
Filkov, A. I., Prohanov, S. A., Mueller, E., Kasymov, D., Martynov, P., Houssami, M. E., Thomas, J. C., Skowronski, N., Butler, B., Gallagher, M., Clark, K., Mell, W. & Simeoni, A. (2017). Investigation of firebrand production during prescribed fires conducted in a pine forest. Proceedings of the Combustion Institute, 36. doi:10.1016/j.proci.2016.06.125.
Viegas, D. X. et al. (2014). Ignition of Mediterranean fuel beds by several types of firebrands. Fire Technology. doi:10.1007/s10694-012-0267-8.
Suzuki, S., Manzello, S. L. & Hayashi, Y. (2015). The size and mass distribution of firebrands collected from ignited building components exposed to wind. Fire Technology. doi:10.1007/s10694-014-0425-2.
Bouvet, N., Link, E. & Fink, S. (2021). A new approach to characterize firebrand showers using advanced 3D imaging techniques. Experiments in Fluids, 62. doi:10.1007/s00348-021-03277-6.
Bouvet, N., Wessies, S., Link, E. & Fink, S. (2023). A framework to characterize WUI firebrand shower exposure using an integrated approach combining 3D particle tracking and machine learning. International Journal of Multiphase Flow, 170, 104651. doi:10.1016/j.ijmultiphaseflow.2023.104651.
Bouvet, N. & Kim, D. (2024). Firebrands generated during WUI fires: a novel framework for 3D morphology characterization. Fire Technology. doi:10.1007/s10694-023-01530-4.
Alonso-Pinar, A., Filippi, J.-B. & Filkov, A. I. (2025). Modelling aerodynamics and combustion of firebrands in long-range spotting. Fire Safety Journal, 152, 104348. doi:10.1016/j.firesaf.2025.104348.
Alonso-Pinar, A., Filippi, J.-B., Nguyen, H.-N. & Filkov, A. I. (2025). A simplified model to incorporate firebrand transport into coupled fire atmosphere models. International Journal of Wildland Fire, 34(7), WF24200. doi:10.1071/WF24200.
Qin, Y. & Trouvé, A. (2025). A numerically accurate mathematical framework for simulations of firebrand transport in landscape-scale fire spread models. Fire Safety Journal, 152, 104343. doi:10.1016/j.firesaf.2025.104343.
Sunberg, L. K. C., Chung, H., MacDonald, E. S., Ouellette, N. T. & Koseff, J. R. (2025). Transport of rod-shaped particles in a canopy flow with a buoyant plume. International Journal of Multiphase Flow, 105331. doi:10.1016/j.ijmultiphaseflow.2025.105331.
NIST: Quantifying Firebrand Exposures for WUI Ignition Potential. Ongoing measurement-science programme, updated 2025.

Suggested citation: Navarro Bodoque, Ó. (2026). “How Far Can Wildfire Embers Travel? The Physics of Spotting and WUI Ignition.” Tecnobosque Technical Notebook 01, English edition, version 1.0, 7 August 2026.

From open knowledge to applied tools

Ember Exposure Technical Pack

This article remains free. A separate professional pack is in development for lecturers, fire services, WUI planners and engineering teams who need a reusable workflow rather than a one-off reading.

Calculation workbook

Documented inputs, unit control, scenario comparison and uncertainty fields.

Teaching deck

Editable diagrams, worked examples and discussion prompts for technical courses.

Field evidence sheet

A structured record for source, shower, receptor, time synchronisation and outcome.

Join the launch list

No operational certification or automated safety decision is implied. The pack will be released only after technical review and usability testing.

Technical and safety notice

This notebook is educational. Simplified equations, historical thresholds and interactive laboratories do not replace legislation, official forecasts, local fire-behaviour analysis, LCES, service procedures or incident command. The models do not calculate safe distances, evacuation triggers or guaranteed control opportunities.