Encyclopedia of Wildfires and Wildland-Urban Interface (WUI) Fires

Living Edition
| Editors: Samuel L. Manzello

Coupled Fire-Atmosphere Interactions

  • Mary Ann JenkinsEmail author
Living reference work entry
DOI: https://doi.org/10.1007/978-3-319-51727-8_77-1

Definition

Fire-atmosphere interactions are the result of coupling between the fire and the atmosphere. Atmospheric conditions impact wildfire behavior which in turn changes the atmosphere around it.

Introduction

To understand and study wildfire behavior in depth and to predict it, it is necessary to understand and be able to model “atmosphere-fire” interactions.

Wildland fire is the product of complex and coupled multiscale processes. At the very smallest scales are the processes of pyrolysis and combustion that produce flames. Heat, moisture, and radiant energy, supplied through the pyrolysis and combustion of ground and canopy fuel, generate extreme levels of buoyancy over the fire perimeter, while strong horizontal buoyancy gradients along the fire perimeter produce vortices of tornadic strength. These combine to produce fire-line winds through (vertical) convection and (horizontal) advection of hot gases. In addition to these fire-atmosphere interactions is the resulting turbulence that mixes combustion gases with ambient air to transport heat, moisture, and combustion products, such as smoke and firebrands, into the surrounding atmosphere. The atmosphere surrounding the wildfire is in turn affected by changing synoptic-scale weather and ABL (atmospheric boundary layer) conditions.

Although synoptic and ABL weather patterns influence temperature and humidity fields, impacting fuel moisture and therefore fire ignition and combustion processes, because wind drives fire propagation, much of wildfire behavior is the result of fluid-dynamical interactions between the fire, the fire-induced flow, and the surrounding flow field. Generally wind affects a wildfire’s spread by tilting the fire flames to ignite unburned fuel. But the buoyancy created by the heat from the propagating fire can cause intense updrafts, inducing very strong surface winds, that affect fire spread and, in turn, all other fire processes. The fire-atmosphere coupling is, therefore, two-way. Fire-induced updrafts can also generate pyro-cumulus (i.e., fire-generated cumulus cloud), precipitation, and fire storms (i.e., conflagration). When a large or intense wildfire significantly impacts local atmospheric conditions, it is sometimes said to be creating “its own weather.”

Our understanding of wildfire behavior is obtained from direct and comprehensive field and laboratory observations and from physical modeling. Physical modeling involves the simultaneous coupling between fire combustion, fire-induced flow, and the surrounding flow field to produce numerical-prediction fire-atmosphere models (e.g., see Mandel et al. 2011, and references therein). Wildfire research uses observations to improve these models, while the dynamical and prognostic modeling framework is used to interpret and understand laboratory and field observations.

In a coupled fire-atmosphere model, feedback between the fire and the atmosphere is through the injection of combustion-generated latent, sensible, and radiative heat into the model atmosphere, at each model time step. The model atmosphere responds, changing the flow in the fire’s environment. The fire-modified flow drives fire spread and the rate of combustion, which injects heat into the model atmosphere to create a dynamical fire-atmosphere feedback. Because the model’s fire-atmosphere feedback can be switched off, this modeling approach can be used to examine directly the importance of fire-atmosphere feedbacks that are believed to be responsible for unexpected and extreme wildfire behavior. By stopping the supply of heat to the model atmosphere, the flow driving fire spread is then the flow predicted by the atmospheric model unimpacted by the fire’s presence (i.e., the flow is modeled as if the fire was not there).

Between pyrolysis within the fire and the large-scale atmospheric processes in the fire environment, the range of physical and temporal scales is enormous. It is not possible to physically model and couple all wildfire processes at all scales. Despite this, in the same way a numerical weather prediction (NWP) model provides a reliable rain forecast without explicitly modeling droplet growth (a microscale weather event), it is not necessary to represent explicitly all fire processes over the entire range of scales to understand, and possibly predict, much of wildland fire behavior and propagation.

A crucial feature that a coupled fluid-dynamical fire-atmosphere model provides is the ability to simulate highly nonlinear dynamical flow features such as turbulence and the vorticity that can lead to the development of a fire whirl or vortex. A fire vortex is a rotating column of air formed in the pyro-convection of a wildland fire. This flow feature is created by a balance of forces which gives it the ability to persist unless other external, more dominant forces lead to its demise. Air pressure is lowest closest to the vortex’s axis of rotation, causing ambient air to accelerate and converge into its base, spiral upward, and then diverge (exhaust) at its top. Long-lived rotation is able to maintain greater horizontal wind speeds, enabling faster fire spread rates, as well as stronger updrafts, providing a mechanism for the creation and lofting of fire brands. If fire whirls stay approximately stationary or within a fire perimeter, they contribute little to fire spread. But fire vortices (and vorticity) are advected by the ambient winds, and fire whirls in close proximity to the fire perimeter can contribute to distinctive fire spread patterns. Moving vortices carry with them some angular and linear momentum, energy, and mass and are an important part of turbulent flow. All these properties associate pyrogenic convection and vorticity with extreme and/or unexpected fire behavior. Vortices within the fire perimeter are the primary coupled fire-atmosphere dynamic responsible for extremely strong surface winds that affect accelerated fire spread.

The following discussion deals with only a few of many existing studies that demonstrate the nature of fire-atmosphere interactions at the fire-line scale and greater on wildfire propagation and behavior.

Fire-Atmosphere Coupling and Pyro-Vortices

One of the first exercises performed with coupled fire-atmosphere models was to examine the impact of background wind speeds on fire-line behavior for the idealized conditions of homogeneous and simple fuel, flat terrain, under neutral stability, and steady background winds directed perpendicular to a fire line. Kochanski et al. (2013) extended this to investigate what effects different background upper-level wind profiles might have on the spread and behavior of this type of idealized fire. This study was partly motivated by Byram (1954). According to Byram certain ambient vertical wind profiles are conducive to erratic, or extreme, or “blow-up” fire behavior. Based on four idealized coupled fire-atmosphere simulations of small (20 m wide, 400 m long) fire lines, Kochanski et al. (2013) set out to test Byram’s hypothesis and to determine the importance of fire plume interaction with upper-level winds as the primary driver of surface fire spread.

In each simulation a different westerly, constant-in-time and direction, background vertical wind profile, all with the same surface wind speed, was imposed on the model’s inlet boundary (Fig. 2). The profiles were a constant speed (CONTROL with no shear), an easterly linear shear, a log-linear shear (LOG), and a hyperbolic tangent shear (referred to as TANH). The tangent wind profile was negative (decreasing with height) at low levels and then sign changing and asymptotic aloft. The surface roughness was virtually nil to eliminate any near-surface wind shear and horizontal vorticity generated by surface friction. These four wind profiles were chosen by Kochanski et al. (2013) to demonstrate that upper-level winds interacting with the fire plume convection not only impact the surface fire rate of spread but also can, under special background wind conditions, induce anomalous behavior of the fire line.

The CONTROL and LOG fires propagated as expected: forward, with a larger fire spread rate and intensity for a greater upper-level background wind speed (the background surface wind speed was identical in each fire), and each forming the near-parabolic fire-line shape characteristic of these types of fires and wind conditions (e.g., LOG fire, upper-right frame; Fig. 2), with the fire head eventually becoming more pointed under higher background wind speeds (e.g., CONTROL and LOG results; Fig. 1). A counter-rotating vortex pair developed at the fire head (e.g., LOG fire; Fig. 2), as observed in wildfires and typical for coupled fire-atmosphere simulations using axisymmetric heat sources (Cunningham et al. 2005, and references therein). Since Kochanski et al. (2013) eliminated frictional-shear at the surface, horizontal vorticity generated by strong buoyancy gradients around the edge of the fire updraft, tilted into the vertical direction by the fire plume, and then stretched and intensified in the fire updraft produced the counter-rotating vortex pair. The upper-level background winds, perpendicular to the fire line, were responsible for advecting the vortex pair and moving the fire head forward in an easterly direction.
Fig. 1

Horizontal cross sections at 2790 s and z = 6.1 m AGL for CONTROL, LOG, and TANH fires: (a) vertical z vorticity (ζz; s−1), (b) horizontal divergence (δ; s−1), (c) horizontal wind speed perturbations (m s−1), and (d) vertical wind speeds (w; m s−1). Magnitudes of each contour are indicated by colors in bar plots on the right. Vectors denote background plus perturbed wind components where vector scale is indicated in the top right corner of plot. Black dashed contours outline the burning surface area. Very bottom plot is energy release rate from moving fire perimeter, ERR (kW m−2), as a function of x (Kochanski et al. 2013)

Fig. 2

Vertical component of vorticity (s −1) for Kochanski et al. (2013)’s LOG (upper frame) and TANH (lower frame) fire simulations, at 500 and 1040 s, respectively. Upper right shows the vertical profiles of the background wind used in coupled fire-atmosphere experiments CONTROL (red pluses), LOG (green asterisks), SHEAR (blue squares), and TANH (dark purple triangles)

In the early stages, the TANH fire had a counter-rotating vortex pair at its head, but as Figs. 1 (right panel) and 2 (bottom right) show, once the fire plume penetrated the negative shear layer in the TANH wind field, the plume tilted upstream, and forward fire front propagation was suddenly eliminated. Figure 1 (right panel) shows a very different fire perimeter shape compared to the other fires, as well as high vertical vorticity, along and outside the fire perimeter. A three-dimensional view of the fire (Fig. 2) shows a highly perturbed wind field, with at least three coherent vertical vortices along and tied to the fire perimeter that are responsible for the unusual and unexpected fire spread at this time in the simulation. These vortices and other multiple weak vortices moved – either advected by the now turbulent flow or influenced by vortices nearby – throughout the fire environment. This unusual evolution of the TANH fire is a natural result of the dynamical instability in the TANH environmental wind shear (i.e., inflection point) being triggered by the pyro-convection. All these special features of the TANH fire happened very rapidly and serve as an idealized example of erratic, or extreme, or “blow-up” fire behavior.

Fire-Atmosphere Coupling and Pyro-tornadogenesis

On January 18, 2003, wildfire swept through the suburbs of Canberra in southeast Australia, producing a series of large (10 km in diameter) pyro-cumulonimbus (pyroCb) cells, each forming near the leading edge of the fire, reaching heights of 14–15 km, and lasting approximately 3 h. One cell produced sooty black hail. Eyewitnesses reported multiple vortices in the pyroCb, but one cell spawned a F2-strength tornado, with a damage path 450 across and 20 km long (see McRae et al. 2013, and references therein). Observations have the tornado close in time and space to the most rapid cell growth and pyroCb maturity, and spot fires were seen in the vicinity of the tornado. This tornado was the only known verified report of pyro-tornadogenesis in Australia and one of the few reported anywhere in the world, at that time. A breakaway fire-generated tornado is a primary example of extreme fire behavior.

This event inspired Cunningham and Reeder (2009) to investigate, using a coupled fire-atmosphere large-eddy simulation (LES) model, the dynamics of a pyroCb cell and how it might produce a localized region of intense vorticity (e.g., pyro-tornadogenesis). The model atmosphere was initialized with an upper-air sounding taken near Canberra on the morning of the fire, and the fire was represented by a simple (highly idealized) stationary surface source of heat and moisture. Model grid dimensions were 200 m in the horizontal and 150 m in the vertical, sized to capture a fire-spawned tornado and the dynamics of pyroCb convection.

Figure 3 shows views of the simulated convection and vorticity after 1 h of simulation. Surface heating by the fire produced a towering pyroCb, with a cloud top of 14 km, as observed, extending past the tropopause into the stratosphere. A strong, upright, tornado-like vortex formed on a side of the fire, extending from the surface into the cloud base, located at a height of 4 km. The pyroCb updraft exceeded 60 m s−1, powerful enough to generate and loft firebrands, which can lead to spot fires far from the initial source. Cunningham and Reeder (2009) analyses indicated that horizontal vorticity generated by strong buoyancy gradients, tilted into the vertical direction by the fire plume, and then stretched and intensified by the strong fire updraft produced this tornado-like vortex.
Fig. 3

Pyro-cumulus convection and vorticity after 1 h of simulation time: views from (a) north, (b) southwest, (c) south, and (d) northeast. Cloud hydrometeors (gray shading), precipitation (blue translucent shading), and intense localized vertical vorticity (green shading). The fire, defined by low-level temperature, is shown in red (Cunningham and Reeder 2009)

Moisture, liberated from the model fire, was crucial to the development of the pyroCb and associated tornadogenesis. Test simulations without moisture from both fire source and the background sounding produced a convective plume 7 km in height. Test simulations without the fire heat and moisture sources generated a relatively shallow cloud top height of 5 km. Only the simulation with heat and moisture from the fire produced a towering pyroCb exceeding 14 km.

In terms of coupled fire-atmosphere interactions, much more is needed to understand what, when, and where conditions, including background atmospheric conditions, combine to produce pyroCb and elusive tornadogenesis, and what impacts their predictability.

Fire-Atmosphere Interactions in the Atmospheric Boundary Layer (ABL)

Because wildfires reside in the ABL, understanding and predicting ABL flow and how it interacts with a wildland fire are crucial for understanding and predicting wildfire behavior. The ABL is the lowest few kilometers of the atmosphere that are modified by the Earth’s surface and that respond to surface forcings operating on time scales of about an hour or less (Stull 1988).

ABL flow is composed of ABL turbulence and waves, superimposed on the larger-scale mean wind field, and conditioned by topography. A major source of uncertainty in wildfire behavior and spread is the response of wildfire to the sudden changes in ABL flow in the fire’s environment. Two fire-atmosphere interactions in the ABL can dominate flow changes in the fire’s environment. One is the interaction or coupling between the fire and the fire-induced flow. The other is the interaction or coupling between the fire and the ambient flow.

Sun et al. (2009) investigated in detail these two fire-atmosphere interactions. Using a high-resolution atmospheric LES model coupled with a simple operational fire spread model, Sun et al. (2009) simulated several grassland fires set on flat terrain for two different ABL flows. One was a highly turbulent, buoyancy-dominated, convective boundary layer (CBL) and the other a roll-dominated ABL (RBL), both superimposed on the same constant mean background wind. The primary flow structure in a CBL is determined by a few powerful updrafts whose typical vertical scale is the CBL layer depth, while the RBL contains coherent horizontal convective rolls or eddy circulations confined to a smaller ABL depth.

The winds driving the fire spread model were determined by either a combination of ABL and fire-induced winds (two-way coupling) or ABL winds only (one-way coupling). Two-way coupling produced significantly greater variation in fire-line shape and behavior, larger burnt area, and faster fire spread in the downwind direction, demonstrating the importance of two-way coupling.

The strongest two-way coupling occurred with the simultaneous combination of low-level fire-induced convergence ahead of the fire’s front and fire-induced downdrafts behind the fire line. Propagation was abruptly increased when a stronger fire-induced downdraft behind the fire line colocated with a sudden ABL downdraft (e.g., Fig. 4). The greater the ABL turbulence, the greater the impact on fire propagation. Either coupling, i.e., one- or two-way, demonstrated considerable uncertainty in fire progression, suggesting that operational fire prediction be probabilistic to account for the random, turbulent nature of the ABL winds driving the fire.
Fig. 4

Vertical velocity (m s−1) at 147 m above ground level. Arrows denote horizontal wind vectors at 5 m about ground level at the end of a 5-min fire-CBL simulation. Black contours are the fire lines. See text for details (Sun et al. 2009)

Figure 4 illustrates one impact of fire-atmosphere coupling on fire spread in a CBL. Ahead of each fire, there is the convergence zone (denoted by light gray shading) that induces the flow tied to the spread of the head of the fire line. An enhanced downdraft behind the fire can be generated by the interaction between the fire plume and the large CBL eddies. The figure shows that behind the fire lines with the most significant growth and rapid spread rate are strong downdrafts (denoted by dark gray shading). An exception is the fire line at the top right of the simulation. Behind the head of the two small fires (left bottom fire and second fire from the bottom on the right), there is either an updraft or a weak downdraft. Multiple weak vortices and fire whirls developed and moved throughout the fire-CBL environment.

Fire-Atmosphere Interactions on the Mesoscale

Peace et al. (2015) demonstrate that a large, intense wildfire can indeed modify the dynamical structure of the surrounding atmosphere.

Four bushfires were burning in December 2007, on Kangaroo Island in South Australia. The MODIS satellite image of the afternoon of December 7 showed smoke plumes of similar sizes and opacities for all four fires (see Peace et al. 2015, their Figure 2). The MODIS satellite image from the next day showed that only one fire, the D’Estrees fire on the southeast corner of the island, had a much larger and opaque smoke plume, indicating increased fire activity. Peace et al. (2015) set out to determine why the D’Estrees fire, relative to the others nearby, experienced sudden and heightened fire activity. A coupled fire-atmosphere model was used to simulate the event. The model was initialized using two different global weather data analyses, and simulations were run with and without feedback from the fire. By comparing simulations with feedback on (fire) and off (no fire), it was possible to investigate how the fire-atmosphere interactions changed both fire behavior and the meteorological environment.

The northern extent of a weak cold front crossed the island during the afternoon of December 7 to enhance the sea-breeze wind shift for the D’Estrees fire. The feedback-on simulations placed the D’Estrees fire in a deep convergence zone between northwest winds and a sea breeze from the south, very different meteorological conditions compared to the other fires on Kangaroo Island. In the feedback-on simulations, fire-modified winds converged and focused the active head fire to a point in the early stages and then accelerated the forward spread. The cold frontal wind shift then opened up a long fire flank along the northeastern edge of the fire, resulting in a new larger and active fire front. In the end, compared to feedback-off, the feedback-on simulations burnt a different and larger fire perimeter and a greater overall fire area, all reasons for the increased smoke production that MODIS observed. Fire-modified winds were similar in strength to the background flow, and the fire’s influence on the near-surface flow extended several kilometers from the fire front.

A significant feature of the feedback-on simulations was a change in the sea-breeze front in response to the fire (Fig. 5). The fire plume’s impacts were to increase the circulation of the sea-breeze front and relocate it farther north, closer to the fire. The change in the timing of the sea-breeze front was attributed to prefrontal temperature increases in the presence of a fire plume. Gravity current and frontogenesis arguments suggest that, as the sea-breeze front approached the fire plume, the frontal line developed stronger upward motion compared to the no-fire scenario.
Fig. 5

y-z cross section of v (north-south), w (vertical) component of wind (vectors), and w component alone (shading: m s−1) for (top) feedback on and (bottom) feedback off. The left-to-right vertical cross section is taken from south to north along the blue line shown in the inset. The red line in the inset shows the fire perimeter. The red dot in the top panel shows the position of the fire front (Peace et al. 2015)

One of the feedback-on simulations produced a long-lived vortex. It was 1-2 km in diameter, extended 500-600 m in the vertical, and lasted 5.5 h (Fig. 6). It appeared first just behind, not coincident with, the active head fire, in a region of enhanced vorticity, and at a time of maximum vertical wind shear in both speed and direction. Twenty minutes later it had moved just in front of the active fire line. This “fire whirl” provided a mechanism for increased fire spread and activity. The fastest rate of spread along the fire front was adjacent to this fire whirl.
Fig. 6

Wind vectors and vertical vorticity (s−1; shaded) at 10 m about ground level. Fire line (feedback on) is shown by red line. The plot region (bottom right; green box) for other panels, with fire-line evolution shown by red lines (Peace et al. 2015)

The fact that only one of the feedback-on simulations produced a fire whirl illustrates two things: one, this type of nonlinear flow feature is a result of fire-atmosphere coupling; and two, flow instabilities of this type are sensitive to both random fluctuations in the coupled fire-atmosphere flow and to NWP initialization. Peace et al. (2015), in agreement with Sun et al. (2009), reinforce the argument that, for predictive modeling of extreme wildfire behavior, a coupled atmosphere-fire modeling approach is necessary and that probabilistic, rather than deterministic, methods are appropriate.

Fire-Atmosphere Interactions with Terrain-Forcing

Countryman (1971) observed that fire whirls that form immediately in the lee of a ridge line are more likely to move either laterally across the leeward slope or diagonally downslope when the background wind approaches the ridge at an angle rather than straight on. This dynamic mode of fire spread appears to be driven by the fire whirls that form over the leeward slope. Simpson et al. (2014) refer to this phenomenon as VLS, for vorticity-driven lateral fire spread.

In a series of sensitivity studies using coupled atmosphere-fire modeling, Simpson et al. (2014, 2016) were able to simulate and determine under what environmental conditions this phenomenon can occur. In each simulation, a westerly background wind approached the ridge of a two-dimensional mountain range, as a single small fire was ignited on the lee side at the base of the mountain. After ignition, the progress of the simulated fires moving upslope to the ridge top was tracked over a 2 h simulation time.

Simpson et al. (2014) found that lateral fire spread occurred only for fire-atmosphere coupling on and with grid spacings fine enough to resolve the fire whirls responsible for VLS. Fire-atmosphere coupling greatly increased the upslope rate of spread and initial fire development, particularly for higher model spatial resolution. As spatial resolution increased, the lateral spread began earlier and lower on the leeward slope during upslope fire spread. The fires then spread laterally across the leeward slope at an enhanced rate and continued to spread laterally and intermittently, with the highest spread rate close to the ridge line on the leeward slope.

Simpson et al. (2016) tested VLS for varying background wind speed (U0; units m s−1) and direction (δ; units degrees) relative to the ridge line for fixed lee-slope steepness (Fig. 7). Regardless of the wind direction (δ), no VLS developed below a threshold wind strength (left frames, U0 ≤ 2.5, Fig. 7). At very low wind speed, no lateral fire spread beyond that expected from the background wind and slope effects was simulated. For perpendicular (δ = 0) environmental winds, U0 ≥ 10 developed VLS. Under these conditions, Simpson et al. (2016) found that a strong pyrogenic updraft developed at ridge top, on the leeward slope, partly blocking the background wind, forcing winds to flow around the updraft. For sufficiently strong background winds, blocking initiated rotation, resulting in pyrogenic vorticity and fire whirls immediately to the lee of the ridge line. For δ = 10, intense pyrogenic vorticity developed more readily as ambient inflow angled toward the base of the pyrogenic updraft, and so the threshold U0 required for VLS when δ = 0 was lowered.
Fig. 7

Shading denotes ignition time (min) of coupled fire-atmosphere model grid cells and terrain height (line contour interval of 100 m) for the simulations with varying background wind speed (U0; m s−1) and deviation from perpendicular (δ degrees), for a fixed lee-slope steepness (θ = 35 degrees). Thick black line shows ridge line position. Dash-filled region shows the ignition region. Arrow (top right) of each plot indicates background wind. Colored dots show time and location of fire whirls identified for each simulation and their size scales with the magnitude of vertical vorticity (s−1) (Simpson et al. 2016)

But lateral fire spread took on different forms depending on the combination of U0 strength and δ. For example, in Fig. 7, for U0 = 5, and δ = 10 and 20, lateral spread was directed in the opposite direction (southward) to the background wind. The flow features responsible for moving the fire south along the ridge were fire spread providing inflow toward the furthest upslope region of the fire to initiate counterclockwise rotation and the formation of a fairly large-scale fire whirl that dominated the relatively weak northerly component of the background wind.

And for very high environmental wind speeds and deviations, the fire spread directly upslope, while the northerly component of the environmental flow pushed and angled the fire perimeter across the leeward slope (Fig. 7, U0 ≥ 7.5 and δ ≥ 20). The leading edge of the fire was located directly upslope from the ignition location, while the trailing edge was located further north and downslope. A pyrogenic updraft developed over the trailing edge, effectively blocking the southwesterly background wind and the formation of counterclockwise rotation, which aided a southeasterly upslope wind that continued to drive northward upslope spread. When the trailing edge of the fire front reached the ridge line, the opposing background and pyrogenic upslope winds converged. A southerly wind resulted in the lee of the ridge line and continued to drive the northward lateral spread, which became increasingly greater if either U0 or δ increased.

Simpson et al. (2016) also tested for VLS to varying lee-slope steepness (θ; units degrees), under increasingly stronger background winds, directed perpendicular to the ridge line (Fig. 8). A very complicated relationship between steepness θ and background westerly U0 emerged. For low slope (θ ≤ 10), there was no lateral fire spread. Under these conditions upslope rate of spread was low, and little heat was released over the leeward slope, producing weak pyro-convection, allowing the downslope background wind to dominate the weak pyrogenic upslope wind. At low wind speeds, there was also insufficient background vorticity immediately in the lee of the ridge line for fire whirls to develop.
Fig. 8

As in Fig. 7 but for background winds directed perpendicular to ridge line with varying speeds (U0) and lee-slope steepness (θ) (Simpson et al. 2016)

Otherwise a very complicated relationship between θ and U0 emerged (Fig. 8). For the fuel type used in these simulations, the fire rate of spread increased nonlinearly from around 0.06 to 0.47 m s−1 as θ increased from 10 to 40. For θ = 20, the upslope fire spread was still fairly low, and there was lateral fire spread provided the background wind was also fairly low (7.5 ≤ U0 ≤ 20). For a low fire spread rate, the pyrogenic updraft was fairly weak, and a balance between the background wind and pyrogenic winds was achieved for this range of U, allowing VLS to develop. But at greater steepness (θ ≥ 30), a much stronger U0 (≥ 10) was required to achieve this balance and lateral spread.

Simpson et al. (2014, 2016) found that during the initial period of lateral fire spread, when the fire size was small, the fire whirls that formed were typically large features that dominated the near-fire wind field near the ridge line. Early on in the simulations, a large fire whirl over the northern (southern) fire flank rotating clockwise (counterclockwise) about the x and z axes was seen (Fig. 9). Fire whirls, like the ones shown in Fig. 10, were formed nonsymmetrically in close proximity to the fire perimeter during the upslope fire spread stage and, in this simulation, significantly affected the lateral spread. As a fire grows in size and emits more heat, the atmospheric flow above and immediately upwind of the fire becomes more turbulent and complex. As a result of the increased heat and atmospheric turbulence, the fire whirls later in the simulations typically became more numerous and smaller in size (Figs. 7 and 8). Fire whirls were not always found in close proximity to rapid lateral spread. Pyrogenic winds without vorticity can align across the fire line, driving the lateral spread.
Fig. 9

Three-dimensional wind streamlines (m s−1, as shown in scale) over the leeward slope at a time of 40 min for U0 = 15 m s−1, δ = 0 degrees, and θ = 35 degrees. The solid red-filled area shows the extent of the fire area (Simpson et al. 2016)

Fig. 10

Vertical velocity, w (m s−1), at approximately 40 m above ground level, and horizontal mid-flame height wind vectors, where the wind speed is at least 5 m s−1, at simulation times 28, 30, 32, and 34 min. The expanding fire perimeter every 2 min is shown by thick black lines; dashed lines indicate terrain height at 100 m intervals. A reference westerly wind vector of 5 m s−1 is shown in the bottom right. Dashed pattern-filled region indicates the prescribed ignition region. For this fire, U0 = 15, δ = 0, and θ = 35 (Simpson et al. 2014)

Coupled Fire-Atmosphere Interactions in the WUI

Increasing encroachment by human development and communities into previously wild areas has created the wildland-urban interface (WUI), the fastest-growing land use type in the United States for the past two decades and the location where humans and their development meet or intermix with wildland fuel. At this time very little is known about how fire-atmosphere interactions cooperate with buildings and houses as fuel sources to produce severe fire spread and behavior (Mell et al. 2010). One observation is that many large WUI conflagrations, such as the California 2018 Camp Fire, spread from building to building by spotting of fire brands, which in a coupled fire-atmosphere situation is a highly random but potentially deadly process. There are currently no existing in-depth scientific studies on which to base a comprehensive discussion of the topic of fire spread in the WUI. There is research on each separate phase of spotting, from production of firebrands, transport of firebrands, and landing and possible ignition of ground fuel by firebrands, in very idealized and simple settings. But research using a coupled wildfire-atmosphere approach has yet to be applied to a realistic WUI setting that models fire spread through an urban community in a comprehensive manner. The difficulties are that exposure to fire and embers and subsequent ignitions in the WUI depend on multiple and coupled processes beginning with wildland fuels, terrain, and current and local meteorology, in addition to community features that include type, density, and layout of WUI terrain, buildings, roads, and landscaping.

Cross-References

References

  1. Byram G (1954) Atmospheric conditions related to blowup fires. Technical report, USDA Forest Service, Southeastern Forest Experiment StationGoogle Scholar
  2. Countryman CM (1971) Fire whirls … why, when, and where. Technical report, USDA Forest Service, Pacific Southwest Forest and Range Experiment Station, BerkeleyGoogle Scholar
  3. Cunningham P, Reeder M (2009) Severe convective storms initiated by intense wildfires: numerical simulations of pyro-convection and pyro-tornadogenesis. Geophys Res Lett 36:L12812CrossRefGoogle Scholar
  4. Cunningham P, Goodrick SL, Hussaini MY, Linn RR (2005) Coherent vortical structures in numerical simulations of buoyant plumes from wildfires. Int J Wildland Fire 14:61–75CrossRefGoogle Scholar
  5. Kochanski A, Jenkins MA, Sun R, Krueger SK, Abedi S, Charney JJ (2013) The importance of low-level environmental vertical wind shear to wildfire propagation: proof of concept. J Geophys Res Atmos 118:8238–8252CrossRefGoogle Scholar
  6. Mandel J, Beezley JD, Kochanski AK (2011) Coupled atmosphere-wildland fire modeling with wrf-fire. Geosci Model Dev 4:591–610CrossRefGoogle Scholar
  7. McRae RHD, Sharples JJ, Wilkes SR, Walker A (2013) An australian pyro-tornadogenesis event. Nat Hazards 65:1801–1811CrossRefGoogle Scholar
  8. Mell WE, Manzello SL, Maranghides A, Butry D, Rehm RG (2010) The wildland-urban interface fire problem – current approaches and research needs. Int J Wildland Fire 19:238–251CrossRefGoogle Scholar
  9. Peace M, Mattner T, Mills G, Kepert J, McCaw L (2015) Fire-modified meteorology in a coupled fire-atmosphere model. J Appl Meteorol Climatol 54(3):704–720CrossRefGoogle Scholar
  10. Simpson CC, Sharples JJ, Evans JP (2014) Resolving vorticity-driven lateral fire spread using the wrf-fire coupled atmosphere-fire numerical model. Nat Hazards Earth Syst Sci 14:2359–2371CrossRefGoogle Scholar
  11. Simpson CC, Sharples JJ, Evans JP (2016) Sensitivity of atypical lateral fire spread to wind and slope. Geophys Res Lett 43:1744–1751CrossRefGoogle Scholar
  12. Stull RB (1988) Boundary layer meteorology. Springer, DordrechtCrossRefGoogle Scholar
  13. Sun R, Krueger SK, Jenkins MA, Zulauf MA, Charney JJ (2009) The importance of fire-atmosphere coupling and boundary-layer turbulence to wildfire spread. Int J Wildland Fire 18(1):50–60CrossRefGoogle Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  1. 1.University of UtahSalt Lake CityUSA
  2. 2.York UniversityTorontoCanada

Section editors and affiliations

  • Sayaka Suzuki
    • 1
  1. 1.National Research Institute of Fire and DisasterTokyoJapan