Pinus yunnanensis fire spread and structural ignition risk in mountainous wildland–urban interfaces of southwestern China

Wildland–urban interface (WUI) fires pose increasing risks in mountainous regions. However, fire-spread processes in southwestern China remain poorly understood. This study quantifies the effects of wind speed, slope, and forest–building separation distance on fire propagation and structural ignition in Pinus yunnanensi s-dominated WUI areas. Laboratory burning experiments combined with fire dynamics simulator (FDS) modeling were conducted across wind speeds of 0–3.5 m/s, slopes of 0–35°, and building separation distances of 4–10 m, measuring peak temperature and radiative heat flux. Crown fuels reached 505–713 °C and 7.85–10.00 kW/m 2, respectively. Higher wind speeds reduced ignition time (~300 to ~194 s), while slope amplified fire development, peaking at 25°. Increasing separation distance significantly lowered heat exposure, preventing ignition at 10 m under the simulated wind speed of 2.5 m/s, 0° slope, no-ember-transport condition within the 900 s simulation period and shifted heating from flame-dominated to radiation-dominated. Forest–building separation emerged as the most effective factor controlling structural ignition, indicating that defensible-space design can substantially weaken vegetation–structure thermal coupling. These findings provide process-based evidence for near-structure fuel management, settlement planning, and community wildfire adaptation in mountainous WUI regions facing increasing climate-driven fire risk.

Under the combined pressures of global climate warming and rapid urbanization, fires at the wildland–urban interface (WUI) have become a critical issue in disaster prevention and mitigation. The WUI is characterized by bidirectional fire transmission, in which wildland vegetation can ignite buildings and, conversely, structure fires can spread back into adjacent wildland fuels. WUI areas are widely distributed across fire-prone regions worldwide and continue to expand with settlement growth (Lampin-Maillet et al., 2010; van Wilgen et al., 2012; Schug et al., 2023; Radeloff et al., 2005, 2018). This expansion increases the spatial contact between combustible vegetation and structures and, when combined with climate warming and regional fire-occurrence drivers, can substantially amplify wildfire exposure and the likelihood of extreme-impact events (Balch et al., 2024; Kumar et al., 2025; Cappelluti et al., 2026). Consequently, continued WUI expansion is expected to intensify wildfire impacts and increase the need to understand the physical processes governing fire spread between wildland fuels and buildings (Radeloff et al., 2018; Ager et al., 2019).

Early WUI studies emphasized delineation and mapping, whereas recent research has increasingly focused on wildfire exposure and vegetation-to-structure ignition (Mell et al., 2010; Chen et al., 2024). WUI definitions and classification frameworks can influence mapped risk, exposed populations, and mitigation priorities (Stewart et al., 2007; Bar-Massada et al., 2023; Gonzalez and Ghermandi, 2024; Cappelluti et al., 2025). However, once wildfire reaches the vegetation–building interface, structure loss is governed by interactions among fuels, weather, topography, and structural vulnerability (Caton et al., 2017). At the parcel scale, construction materials, near-structure vegetation, defensible space, and surrounding fuel configuration jointly influence building survival (Syphard et al., 2019; Escobedo et al., 2025). Structural ignition may result from flame contact and thermal radiation from nearby fuels or from airborne embers and firebrands (El Ezz et al., 2022; Mahmoud and Chulahwat, 2020; Silva et al., 2024). Defensible space is therefore a multi-component mitigation concept involving fuel management, vegetation maintenance, building characteristics, ember-resistant measures, and spatial separation between vegetation and structures. This study focuses specifically on forest–building separation distance as a quantifiable component of defensible space. From a fire-physics perspective, separation distance directly regulates vegetation–structure thermal coupling: at short distances, structures may experience flame contact, convective heating, and intense radiation, whereas increasing separation weakens direct flame impingement and near-field convective transfer, shifting exposure toward lower-intensity radiation-dominated heating. Forest–building separation is therefore not merely a spatial planning parameter but a direct control on the dominant heat-transfer pathways and, consequently, on structural ignition potential. Recommended separation distances vary widely with fuel conditions, fire behavior, and thermal exposure criteria, typically ranging from 20 to 65 m (Cohen and Butler, 1998; Butler and Cohen, 2000; Castillo et al., 2022; Fang et al., 2019), while China currently lacks unified technical guidance for WUI defensible-space design.

These issues are particularly important in mountainous WUI environments. Warmer and drier fire seasons can increase fuel flammability and prolong hazardous thermal exposure near structures (Jolly et al., 2015; Abatzoglou and Williams, 2016; Schoennagel et al., 2017), while steep terrain and the close proximity of combustible vegetation to rural buildings can intensify fire spread and fire–structure interactions (Caton et al., 2017). Although community preparedness, risk perception, and communication can influence the implementation of physical mitigation measures (Meldrum et al., 2019; Spano et al., 2021; Calvani et al., 2025), process-based understanding of vegetation-to-structure fire spread remains essential for translating such broader adaptation efforts into effective defensible-space and settlement-planning strategies.

In China, WUI research has progressed rapidly but process-based understanding of vegetation-to-structure fire spread remains comparatively limited. WUI areas have expanded significantly since 2008, with fire occurrence showing strong spatial and seasonal variability driven by both climatic and anthropogenic factors (Gong et al., 2024). Similar long-term WUI expansion has been reported in other fire-prone regions has increased settlement–vegetation contact and reinforced the need to integrate wildfire risk into land-use planning (Godoy et al., 2022). This issue is particularly relevant in southwestern China, where mountainous and plateau landscapes dominate Yunnan Province and create complex combinations of slope, aspect, and elevation that can strongly influence fire behavior and heat-transfer pathways (Zhang et al., 2019). P. yunnanensis is a dominant conifer in these fire-prone landscapes and is widely distributed around settlements in central Yunnan. The species exhibits several fire-related traits, including thick bark, serotinous cones, and high foliar volatile content, which may influence its combustion behavior (Su et al., 2015). Such trait-based differences are increasingly recognized as important determinants of species-level flammability and vegetation–fire dynamics (Cappelluti et al., 2024). However, existing studies have focused mainly on prescribed burning, fuel characteristics, and surface fire behavior, confirming its importance for regional fire management (Wang et al., 2023; Hong et al., 2025), whereas quantitative evidence linking its combustion behavior to building heat exposure and ignition potential in WUI environments remain scarce. In particular, it remains unclear how wind, terrain slope, and forest–building separation modify vegetation–structure thermal coupling and the transition among flame contact, convective heating, and radiation-dominated exposure in mountainous WUI conditions. Addressing these gaps requires approaches that can characterize fuel thermal behavior while systematically comparing fire–structure interactions under varying environmental and spatial conditions.

The fire dynamics simulator (FDS), based on large-eddy simulation of low-speed fire-driven flows, has been widely applied to vegetation combustion, fire spread, and flame–structure heat transfer under controlled WUI scenarios (Mell et al., 2007, 2009; Meerpoel-Pietri et al., 2023; Ganteaume et al., 2023). Previous studies have demonstrated its utility for analysing radiative exposure and coupled fuel–terrain–meteorological effects on structural ignition (Cohen, 2004; Fiorini et al., 2023). Accordingly, this study combines laboratory combustion experiments with FDS-based simulations to investigate fire spread from forest fuels to buildings in a representative mountainous WUI setting in Kunming, China. The objectives are to (1) characterize the thermal behavior of P. yunnanensis crown combustion, (2) evaluate the capability of FDS to reproduce observed heat-exposure patterns, and (3) quantify the effects of wind speed, slope, and forest–building separation distance on structural ignition potential. By linking fuel combustion, thermal exposure, and building ignition outcomes, this study aims to provide process-based evidence for defensible-space planning, landscape-scale fuel management, and community wildfire adaptation in fire-prone mountainous settlements.

Laboratory canopy-fire radiation experiments

Sampling and objective

Laboratory burning experiments were conducted to characterize radiative exposure produced by canopy fuels and to support subsequent WUI fire modeling. Crown fuels of P. yunnanensis were collected in July 2025 from a representative WUI area near Hamazhe Village (Jindian, eastern Kunming, China). Samples were mounted vertically for testing. Basic geometric descriptors were recorded (bole height, crown height, total height), and fuel moisture was measured immediately prior to burning. Fuel moisture was reported as air-dry moisture content (wet-basis), calculated from fresh mass measured on the day of sampling and air-dried mass measured before testing. These laboratory experiments were conducted under controlled indoor conditions to characterize the baseline thermal behavior of P. yunnanensis crown fuels, rather than to reproduce the full complexity of field fire behavior.

Experimental setup and measurements

Experiments were performed in the Wildfire Behavior Research Laboratory at Southwest Forestry University using a modular single-tree burning rig (Figure 1). The setup comprised an exhaust hood for smoke removal, suspension wires to secure samples beneath the hood, a support frame for sensors, and a cylindrical ignition drum containing a 0.20-m litter bed to initiate crown ignition. Ignition was achieved using a piloted ignition mode, in which an external flame source guided the initiation of combustion in the litter bed. Tests were conducted under quiescent indoor conditions (no imposed wind), with the exhaust hood operating throughout. A front-facing camera recorded each burn.

Mass loss was quantified by measuring pre-burn mass and post-burn residual mass after self-extinction and cooling to ambient temperature; the residual was reweighed until consecutive measurements were stable, and mass loss was computed by difference. Temperature was measured using K-type sheathed thermocouples (WRNK-191) connected to a multichannel acquisition system (DAM-4501 with DAM-1101), recorded at 1-s intervals. Thermocouples were placed 0.5 m from the sample on the left side at 0.5, 1.0, and 1.5 m above ground. Radiative heat flux was measured using MR-5 radiometers placed 0.5 m from the sample on the right side at the same heights to capture vertical variation in radiative exposure. Thermocouples and radiometers were positioned on opposite sides to minimize sensor interference. Fuel higher heating value (HHV) was measured from dried, ground subsamples using an XRY-1C oxygen bomb calorimeter (approximately 1 g per replicate; three replicates per sample; mean reported).

Test conditions

Four P. yunnanensis crown samples were tested. Fresh mass and geometric dimensions were measured on the day of collection. Samples were air-dried prior to experiments and reweighed to determine air-dry moisture content. Test conditions are summarized in Table 1.

WUI fire dynamics modeling

Modeling platform, reaction scheme, and grid

WUI fire dynamics were simulated using FDS Version 6.7.6, with geometry and scenario configuration implemented in PyroSim. Outputs were visualized in Smokeview and post-processed in Tecplot. Combustion was represented using the built-in WOOD-PINE reaction, which is a one-step reaction scheme widely adopted for forest fuel combustion in FDS and has been validated in previous WUI fire modeling studies (Mell et al., 2009). A uniform Cartesian grid of 0.30 m × 0.30 m × 0.30 m was applied throughout the computational domain. This grid resolution was selected to balance the computational cost of multi-scenario simulations with the need to adequately resolve flame–structure interactions near the building facade.

Fuel, vegetation, and building representation

Surface fuels and trees

P. yunnanensis stand was represented using a layered surface fuel bed comprising pine needles, litter, and an underlying soil layer. Pine needles and litter were treated as the ignitable fuel bed, assumed spatially homogeneous in density and moisture, with a total thickness of 0.10 m. Trees were prescribed with a total height of 6.5 m, live crown height of 4.5 m, and bole height of 2.0 m. A total of 50 trees were arranged at 2.0 m spacing, based on the average tree spacing observed during the field survey. This fixed stand configuration was maintained across all simulations to isolate the effects of wind speed, terrain slope, and forest–building separation distance. The crown fuels were represented using discretized obstruction elements with homogeneous combustible properties, which simplified the heterogeneous internal structure of real tree crowns. These simplifications were adopted to provide a controlled reference configuration for comparative scenario analysis.

Buildings

Two identical buildings were modeled to represent typical brick–wood structures in the study area. Each building was 9 m × 6 m × 7.6 m. The use of identical building geometry was intended to represent the dominant local residential building form and to isolate the effects of the investigated fire-spread variables. Exterior walls were defined as masonry, and the roof was defined as pine wood; indoor furnishings were also represented as pine wood for material consistency. These material settings were simplified representations of typical local construction materials and were selected to approximate their thermal behavior.

WUI configuration

The WUI scenario combined the surface fuel bed, forest stand, and buildings to represent fire spread from forest to settlement, with structure exposure via flame contact and radiative heating. These assumptions allowed the model to focus on the relative influence of fuel moisture, wind speed, slope angle, and forest–building separation distance (Figure 2).

Fire-source specification, environmental conditions, and terrain

A line ignition source (22 m × 1 m) was placed at the upwind edge of the forest fuel bed. This ignition configuration was designed as a representative scenario rather than to reproduce a specific wildfire event. The source length spanned the full width of the forest fuel bed to generate a continuous fire front approaching the built environment, while the 1 m width defined a narrow ignition zone. Fire growth followed an unsteady t 2 curve with a total peak heat release rate of 6 MW, which was used as a scenario-based ignition intensity to initiate sustained vegetation combustion. The ignition source was removed at 250 s after the initial fire front had developed, allowing subsequent fire spread to be mainly controlled by the prescribed fuel, wind, terrain slope, and forest–building separation conditions. These ignition parameters were kept constant across all simulations. Total simulation time was 900 s. Terrain slope was applied within the forest region, while the building area remained flat to represent an upslope fire approaching structures. Wind direction was prescribed from right to left. Background conditions were fixed at 20 °C and 15% relative humidity, representing dry spring conditions associated with forest-fire occurrence in the study region.

Sampling strategy and outputs

Point outputs of gas temperature and radiative heat flux were obtained at nine locations along three downstream sections (Y = 6.8, 8.8, and 10.8 m) and three heights (Z = 2.0, 4.0, and 6.0 m) at a fixed lateral position (X = 10.5 m). Temperature slices were extracted at the same Y-locations to characterize the thermal field near the building facade, the flame-front region, and within the fire-core region. Figure 3 illustrates the locations of output sensors and slice planes, while Table 2 provides the coordinates and identifiers of temperature and radiative points used in the numerical model.

Scenario sets

Three scenario sets were designed using a one-factor-at-a-time approach to isolate the effects of wind speed, slope, and forest–building separation distance under controlled conditions:

Scenario set 1 (wind): slope 0°, separation 4 m, wind speed 0, 1.5, 2.5, 3.5 m/s.

Scenario set 2 (slope): wind speed 2.5 m/s, separation 4 m, slope 0°, 15°, 25°, 35°.

Scenario set 3 (separation): wind speed 2.5 m/s, slope 0°, separation 4, 6, 8, 10 m.

The 4–10 m separation range was selected based on field surveys of typical forest–building distances in mountainous WUI settlements of central Yunnan, where residential structures are commonly found adjacent to or within P. yunnanensis stands.

Laboratory burning experiments

Mass loss and energy release

The burning rate is a key parameter for quantifying combustion intensity and represents the mass loss of combustible material per unit time (g/s). During the combustion process, the specimen mass decreases monotonically with time and stabilizes once combustion ceases. By analysing the differential characteristics of the mass–time curve, distinct combustion stages can be identified, including the ignition stage, steady combustion stage, and decay stage. As shown in Figure 4, for the P. yunnanensis samples used in this study, the components that were completely consumed during combustion were primarily the crown needles and portions of the outer bark.

Effective energy represents the total amount of heat potentially released during the complete combustion of the fuel and is an important indicator for evaluating fire behavior. It directly affects key combustion characteristics, including fire intensity, fire severity, radiative heat flux, and overall fire development. In this study, the effective energy release was calculated as:

Where: E is the energy released by the fuel (kJ), F is the fresh mass of fuel (kg), H is the calorific value of the fuel (kJ/kg), A is the effective combustion coefficient, and M is the fuel moisture content expressed as a fraction rather than a percentage. The calorific value of P. yunnanensis needles was experimentally determined to be 18.752 MJ/kg. As the effective combustion coefficient of P. yunnanensis has not yet been measured in this study, A = 0.65 was adopted by referring to Madrigal et al. (2011), who reported that effective heat of combustion during flaming combustion generally accounts for 56–82% of gross heat of combustion (Table 3).

The results indicate that fuel moisture content and fresh mass jointly influenced burning behavior in naturally air-dried P. yunnanensis samples under laboratory conditions. Lower FMC generally promoted faster ignition, higher mass loss rate, and greater effective energy release. Experiment 2 exhibited the most intense combustion, while Experiment 3 showed the weakest energy output, likely due to its relatively high moisture content.

Gas temperature and radiative heat flux responses

Based on Figure 5, the thermocouple-measured gas temperature and incident radiative heat flux showed a consistent single-peaked (unimodal) temporal response across all four experiments, with a rapid increase immediately after ignition followed by a gradual decay toward background levels. Peak gas temperatures ranged from 505 to 713 °C and occurred at HAG = 1.0–1.5 m within 21 ~ 28 s after ignition, whereas peak incident radiative heat fluxes ranged from 7.85 to 10.00 kW/m 2 and were consistently observed at HAG = 1.5 m within 27 ~ 36 s, i.e., with a 3 ~ 13 s delay relative to the temperature peaks (Table 4). Spatially, gas temperature decreased with increasing horizontal distance from the fire source; vertically, the height of peak temperature occurred at 1.0–1.5 m depending on the experiment, while within the tested height range the incident radiative heat flux increased with sensor height, following crown top > crown middle > crown base. These laboratory measurements provide baseline thermal exposure characteristics and are used to evaluate whether the FDS setup reproduces the observed temporal evolution and vertical gradients.

Model–experiment comparison

To quantify the vertical variation in building thermal exposure and to enable comparison with the laboratory burning experiments, three sampling heights were set in the forest fuel bed. Gas temperature was extracted at T4–T6 and incident radiative heat flux at R4–R6, corresponding to the crown base, mid-crown, and crown top, respectively (locations collocated in plan view Figure 6, X = 10.5, Y = 8.8, Z = 2.0, 4.0 and 6.0, respectively).

Figures 6a1–a4 shows that the simulated gas temperature time series at the three heights exhibit a consistent unimodal evolution across wind speeds, with a rapid rise followed by decay. The vertical ordering is stable across the four wind cases, with T6 > T5 > T4, indicating higher gas temperatures at higher elevations. This height-dependent pattern is consistent with the trends observed in the laboratory experiments.

Figures 6b1–b4 presents the corresponding incident radiative heat flux responses. Across wind speeds, radiative heat flux also increases rapidly during fire development and then decreases as the fire decays. During the early stage (t < 300 s), higher elevations experience stronger radiative loading (i.e., the upper locations generally exceed the lower locations). At later times, the vertical ordering becomes less uniform; the simulations indicate that the mid-crown level (R5) can exceed the crown-top level (R6), while R4 remains the lowest. Overall, the simulated vertical gradients and their evolution with time are broadly consistent with the experimental observations at comparable heights.

Because the laboratory experiments and FDS simulations were conducted at different geometric scales, the comparison was not intended as a strict point-by-point validation of absolute temperature or heat-flux magnitudes. Instead, the model–experiment comparison was used to evaluate whether the FDS setup could reproduce the main thermal-response characteristics observed in the experiments, including unimodal temporal evolution, vertical gradients, peak occurrence, and the relative ordering of temperature and radiative heat flux among canopy positions. A summary comparison is provided in Table 5. Accordingly, the FDS model was used in this study primarily for comparative scenario analysis of wind speed, slope, and forest–building separation distance, rather than for absolute prediction of building ignition risk.

Heat release rate under different environmental and spatial conditions

Having established qualitative agreement with the experimental vertical profiles, we use the model to quantify the effects of wind speed, slope, and forest-building separation distance on fire structure exposure.

Wind speed

Figure 7a presents the temporal evolution of the HRR under different wind speed conditions. As wind speed increases, the time required to reach the peak HRR is progressively shortened, while the peak value exhibits an overall increasing trend. After reaching the maximum, HRR decreases rapidly and subsequently approaches a quasi-steady level. It is observed that the early-stage HRR is dominated by the combustion of forest fuels, whereas after ignition of the building, the forest fuels are largely consumed and the HRR is mainly governed by building combustion.

Slope

The variation of HRR under different slope conditions is shown in Figure 7b. The results indicate that slope has a significant influence on fire intensity in the WUI. Among the tested cases, the HRR reaches its maximum at a slope of 25°. When the slope increases from 0° to 25°, both the peak HRR and the fire intensity increase noticeably. However, further increasing the slope to 35° leads to a reduction in HRR, indicating a non-monotonic relationship between slope and fire intensity.

Forest-building separation distance

Figure 7c illustrates the influence of forest-building separation distance on HRR under a wind speed of 2.5 m/s. With increasing forest-building separation distance from 4 m to 10 m, the peak HRR decreases significantly, and the fire duration is correspondingly shortened. The case with a 4 m firebreak exhibits the highest HRR and the longest burning duration, while the 10 m firebreak results in the lowest HRR and the earliest fire extinction. These results demonstrate that increasing forest-building separation distance effectively reduces fire intensity and limits fire spread toward buildings.

Roof thermal exposure under different environmental and spatial conditions

We next relate these differences in heat release rate to roof-level thermal exposure, quantified using roof temperature and incident radiative heat flux. The thermal response of the building roof was evaluated using the temperature sensor T6 and radiative heat flux sensor R6 under different wind speed, slope, and forest-building separation distance conditions.

Effect of wind speed

Figures 8a1,a2 show the temporal variations of roof temperature and radiative heat flux under different wind speeds. In all cases, both parameters increased rapidly during the early fire stage and decreased after reaching peak values. Higher wind speeds resulted in faster increases in roof temperature and radiative heat flux. The maximum roof temperatures were 853.5 °C, 847.1 °C, 841.5 °C, and 998.3 °C for wind speeds of 0, 1.5, 2.5, and 3.5 m/s, respectively. The corresponding peak radiative heat fluxes were 32.3, 31.8, 16.8, and 22.9 kW/m 2.

Effect of slope

Figures 8b1,b2 present the roof thermal response under different slope conditions. Both roof temperature and radiative heat flux increased with slope up to 25°, where the highest values were observed. The roof temperature exceeded 1,000 °C at 25°, while lower slopes showed slower temperature growth. Peak radiative heat fluxes were 12.1, 23.4, 382.2, and 292.4 kW/m 2 for slopes of 0°, 15°, 25°, and 35°, respectively, indicating a non-monotonic relationship between slope and roof heat exposure. The exceptionally high heat-flux value at 25° represents a simulated, localized, short-duration point-sensor peak, likely caused by transient flame/plume attachment near the roof sensor, and should not be interpreted as a sustained roof-wide exposure.

Effect of forest-building separation distance

Figures 8c1,c2 show the influence of forest-building separation distance on roof temperature and radiative heat flux at a wind speed of 2.5 m/s. Increasing the forest-building separation distance significantly reduced both the peak values and temporal fluctuations. At distances of 4 m and 6 m, roof temperatures approached 1,000 °C and radiative heat flux exceeded 15 kW/m 2. When the distance increased to 8 m, both temperature and heat flux decreased markedly. At 10 m, roof temperatures remained below 100 °C and radiative heat flux dropped to 11.9 kW/m 2, indicating effective mitigation of thermal exposure.

Fire spread process in WUI

We examine facade thermal exposure and the fire spread process, which directly determine building ignition outcomes under WUI conditions.

Wind speed effects on facade heat flux and temperature fields

Wind speed significantly affected building ignition in the WUI, with higher winds accelerating fire spread and reducing ignition delay. Under zero-slope conditions, ignition time decreased from 300 s at 0 m/s to 272 s at 1.5 m/s, 223 s at 2.5 m/s, and 194 s at 3.5 m/s (Figure 9). Facade exposure was governed by transient radiative and convective heating, resulting in time-dependent and spatially non-uniform heat fluxes. Temperature fields were evaluated using a slice at Y = 6.8 m, tangential to the facade edge and located 4 m from the forest model. At t = 300 s (early fire stage), increasing wind speed intensified facade heating and expanded the high-temperature zone. The maximum wall-surface temperature increased from approximately 650 °C (0 m/s) to ~780 °C (1.5 m/s) and exceeded 1,000 °C at 3.5 m/s.

Slope effects on facade heat flux and temperature fields

Under a wind speed of 2.5 m/s, slope markedly accelerated fire spread toward the structure and reduced the time to building ignition. As slope increased from 0° to 35°, the ignition time decreased monotonically from 228 s to 186 s (15°), 163 s (25°), and 145 s (35°). The fire development at t = 350 s (Figure 10) further indicates a progressive enhancement of upslope fire advance with increasing slope: at 0° only local burning on the building was observed; at 15° the burning area expanded but the structure was not fully engulfed; at 25° most of the building was surrounded by flames and the burning was most severe; and at 35° ignition occurred earliest, yet the overall burning severity was lower than that at 25°.

The radiative exposure on the building facade increased substantially with slope, with the highest levels occurring at intermediate slope. At the ignition stage, the range of facade incident radiative heat flux was 0–21.88 kW/m 2 (0°), 0–25.79 kW/m 2 (15°), 0–59.5 kW/m 2 (25°), and 0–55.0 kW/m 2 (35°), with the 25° case exhibiting the largest radiative loading. Temperature-slice results show that the spatial distribution of high-temperature regions around the building was also slope-dependent. At low slopes, elevated temperatures were mainly confined to the roof with a limited affected area; at 15°, the high-temperature region expanded markedly and the maximum temperature approached 750 °C; at 25°, the temperature rise was most rapid; and at 35°, the high-temperature region covered the largest area, indicating an expanded extent of thermal exposure to the building.

Forest-building separation distance effects on facade heat flux and temperature fields

Under WS = 2.5 m/s and slope = 0°, increasing the forest-building separation distance reduced thermal exposure to the structure, delayed building ignition, and ultimately prevented ignition at the largest separation. For forest-building separation distances of 4, 6, and 8 m, the building ignition times were 223 s, 256 s, and 285 s, respectively; at 10 m, no ignition occurred within the 900-s simulation for the tested scenario. The fire development at t = 300 s (Figure 11) shows that the burning area on the exterior facade was largest for 4 m, decreased from 6 m and 8 m, and was absent for 10 m, consistent with a progressive reduction in fire impact with increasing separation.

Facade incident radiative heat flux exhibited clear distance-dependent differences. At the time of ignition, the facade incident radiative heat-flux ranges were 0–21.5 kW/m 2 (4 m, t = 223 s), 0–22.5 kW/m 2 (6 m, t = 256 s), and 0–19.06 kW/m 2 (8 m, t = 285 s). For the 10 m case (no ignition), the facade incident radiative heat flux at t = 300 s ranged from 0 to 15.29 kW/m 2. The spatial distribution further indicates that, for the 4 m case, the mid-facade region experienced the highest radiative loading rather than the roof (Figure 11a).

Temperature-slice results corroborate the mitigation effect of increased separation distance, showing both lower peak temperatures and a reduced extent of high-temperature regions in front of the building. The maximum temperature in front of the structure was close to 1,000 °C for 4 m, decreased to approximately 950 °C for 6 m, and dropped to <60 °C for 10 m, indicating a substantially moderated thermal environment at larger forest-building separation distances.

Coupled exposure pathways controlling vegetation-to-structure ignition

The present study shows that WUI fire spread from Pinus yunnanensis fuels to buildings is controlled by coupled exposure pathways rather than by a single thermal process. Laboratory experiments demonstrated that naturally air-dried P. yunnanensis crown fuels produced short-duration but intense thermal pulses, with peak gas temperatures of 505–713 °C and peak radiative heat fluxes of 7.85–10.00 kW/m 2. These values indicate that crown combustion can generate hazardous near-field exposure even under quiescent indoor conditions. Similar transient thermal responses have been reported in experimental and numerical studies of burning tree and shrub fuels, where rapid flaming combustion produces short but intense heat-flux peaks (Mell et al., 2009; Meerpoel-Pietri et al., 2023).

The FDS simulations further showed that wind speed, slope, and forest–building separation distance changed both the magnitude and duration of building exposure. This is consistent with the general WUI ignition framework in which flame contact, radiation, convection, and ember transport jointly determine structure ignition potential (Caton et al., 2017; El Ezz et al., 2022; Silva et al., 2024). However, the present results add process-level evidence from a mountainous P. yunnanensis WUI setting in southwestern China, where the interaction between terrain, wind, and forest–building spacing can rapidly shift exposure from flame-dominated heating to weaker radiation-limited conditions.

Effects of wind and slope on structural ignition

Wind strongly accelerated vegetation-to-structure fire spread. As wind speed increased, building ignition occurred earlier, indicating that wind promoted flame tilt, plume attachment, and forward heat transfer toward the structure. This agrees with previous WUI modeling studies showing that wind modifies flame geometry and increases structural exposure by enhancing convective and radiative coupling between fuels and buildings (Mell et al., 2010; Fiorini et al., 2023). Nevertheless, the roof-level heat flux did not increase monotonically with wind speed, suggesting that local exposure depends not only on fire intensity but also on the relative position of the flame front, plume, and structural surface.

Slope shortened the time to building ignition, but the strongest fire–structure interaction occurred at 25° rather than at the steepest slope of 35°. This non-monotonic response suggests that earlier ignition does not necessarily correspond to stronger or more sustained structural exposure. As slope increased, upslope flame attachment and preheating of fuels intensified, accelerating fire spread toward the building. However, at very steep slope, the fire front passed the structure more rapidly, reducing the residence time of the most intense flame and plume near the building surface. In contrast, the 25° slope appeared to provide a more favorable balance between enhanced upslope spread, flame attachment, plume development, and exposure duration, producing the strongest roof and facade thermal loading. Therefore, slope should be interpreted not only as a factor controlling ignition timing, but also as a factor regulating the spatial alignment and residence time of heat transfer pathways near structures. Similar conclusions have been drawn in recent WUI heat-transfer simulations, where terrain affects not only spread rate but also the spatial distribution and persistence of heating on structures (Fiorini et al., 2023; Purnomo et al., 2024). Therefore, slope should be evaluated together with wind direction, fuel continuity, and building position rather than treated as an isolated hazard factor.

Forest–building separation and defensible-space effectiveness

Forest–building separation distance was the clearest controlling factor in this study. Buildings ignited at 4–8 m but did not ignite at 10 m within the simulated duration, indicating that increasing separation distance effectively weakened vegetation–structure thermal coupling. This non-ignition result at 10 m is scenario-specific and reflects the combined effects of the prescribed fuel configuration, wind speed, terrain slope, building materials, and simulation duration. In addition, ember transport and ember-induced ignition were not explicitly represented in the present model, which may influence ignition outcomes in real WUI conditions. Therefore, the 10 m result should be interpreted as indicating substantially reduced flame impingement and near-field thermal exposure under the tested scenario, rather than a universal defensible-space threshold. This agrees with earlier work showing that defensible space reduces structure ignition by limiting flame contact, radiant exposure, and near-building fuel continuity (Cohen and Butler, 1998; Cohen, 2004; Syphard et al., 2014). The present results further suggest that separation distance should not be interpreted simply as a way to reduce radiative heat flux. At shorter distances, direct flame influence and plume residence near the facade were important, whereas at larger distances the exposure regime shifted toward weaker radiation-limited heating. This transition supports the view that structure ignition depends on both heat-flux magnitude and exposure duration (Caton et al., 2017). However, the 10 m non-ignition result should be interpreted as scenario-specific rather than a universal safety threshold, because defensible-space requirements vary with fuel type, fuel loading, wind, slope, building materials, and ember exposure (Butler and Cohen, 2000; Castillo et al., 2022; Fang et al., 2019).

Regional implications for wildfire adaptation in southwestern China

The findings are particularly relevant to mountainous WUI settlements in southwestern China, where fire occurrence is strongly associated with the spring dry season rather than the summer fire season commonly observed in many Mediterranean, Australian, and western North American WUI regions (Gong et al., 2024). In Yunnan, prolonged dry-season conditions before the summer monsoon can reduce fuel moisture, while agricultural activities, traditional burning practices, and other human ignition sources further elevate fire risk (Ying et al., 2021). This regional pattern should also be interpreted within a broader climatic context, as heatwaves, droughts, and compound dry hazards can increase the likelihood of low fuel moisture and severe wildfire behavior across different ecoregions (Ruffault et al., 2020; Elia et al., 2024). Thus, the experimental finding that lower fuel moisture promoted faster ignition, higher mass loss rate, and greater effective energy release is consistent with both local dry-season fire mechanisms and wider climate-related drivers of severe wildfire risk.

The WUI literature has emphasized that wildfire risk is shaped by the combined effects of climate change, settlement expansion, and exposure of human communities to flammable landscapes (Chen et al., 2024). In this context, the present study provides a physical basis for locally adapted mitigation in rural mountainous settlements of Yunnan Province, where defensible-space planning should be combined with near-structure fuel reduction, interruption of horizontal and vertical fuel continuity, use of less flammable landscaping, protection of vents and eaves, and improvement of roof and facade fire resistance. At the landscape scale, forest–building separation should be considered together with road-network planning, evacuation access, and fuel-break placement. These measures should also be integrated with community engagement and local adaptation planning, because community perceptions and preparedness can affect the implementation of proactive wildfire mitigation (Copes-Gerbitz et al., 2022).

Several limitations should be considered when interpreting these implications. The laboratory experiments were conducted under controlled indoor conditions, and the simulations simplified the spatial heterogeneity of forest stand, crown structure, fuel distribution, building materials, and settlement layout. In addition, the one-factor-at-a-time design enabled the individual effects of wind speed, slope angle, and structure-to-fuel distance to be isolated, but did not fully explore potential interactions among these factors. Therefore, the results should be interpreted as a first-order assessment of individual controlling factors rather than a complete factorial analysis, and future studies should use fully factorial or coupled scenario designs to quantify interactions among wind, slope, fuel moisture, and forest–building separation distance. Ember transport and ember-induced ignition were not explicitly resolved, although firebrands are widely recognized as an important pathway for structure ignition in WUI fires (Caton et al., 2017; El Ezz et al., 2022; Silva et al., 2024). Future research should therefore incorporate a wider range of fuel moisture conditions, vegetation structures, wind directions, building materials, and ember exposure scenarios. Overall, effective WUI fire mitigation in mountainous southwestern China requires the integration of defensible-space design, near-structure fuel management, landscape planning, building hardening, and community adaptation, consistent with studies highlighting the influence of spatial layout, distance to fire, suppression accessibility, and structure properties on WUI exposure (Wei et al., 2023; Purnomo et al., 2024).

This study investigated the effects of wind speed, terrain slope, and forest–building separation distance on wildfire-driven structure ignition in a representative mountainous WUI scenario of central Yunnan, China, dominated by P. yunnanensis, using physics-based numerical simulations. The results demonstrate that all three factors significantly influence ignition dynamics, but their effects operate through distinct physical pathways. Wind speed primarily controls flame spread rate and convective heat transport, while terrain slope enhances upslope flame attachment and accelerates fire–structure interaction. Forest–building separation distance governs the transition from direct flame impingement to radiation-dominated heating regimes, thereby strongly influencing ignition occurrence.

Overall, the findings highlight that structure ignition in WUI fires is governed by coupled interactions between fire behavior, local topography, and spatial configuration under the specific fuel conditions of Pinus yunnanensis forests in central Yunnan. These results provide a mechanistic basis for defensible-space design and wildfire mitigation planning in mountainous settlements, particularly in regions characterized by seasonal dry conditions and increasing exposure to vegetation–structure interfaces.