The Bor Forest Island Fire Experiment
Fire Research Campaign Asia-North (FIRESCAN)
IV. Bor Forest Island Fire Behavior and Atmospheric Emissions
1. Preburn Fuel Sampling
For fuel sampling and fire behavior documentation purposes, Bor Forest Island was gridded with a three by four array of 12 grid points, each 250 m apart in a SW-NE direction, and 150 m apart in a NW-SE direction. At each of these locations a 15 m transect was established, and downed woody fuels were inventoried using the line-intersect method (McRae et al. 1979). Mean preburn ovendry fuel weights were determined to be 0.198 kg m-2 for fuels 0-3 cm in diameter, 0.417 kg m-2 for fuels 0-7 cm in diameter, and 1.088 kg m-2 for fuels >7 cm in diameter. The total preburn downed woody fuel load was therefore 1.505 kg m-2.
A total of 5 pins were located at fixed intervals along each line-intersect transect in order to determine an average depth-of-burn. A horizontal bar of each pin was placed flush with the top of the lichen layer. Following the fire, the distance between the bottom of this bar and the remaining organic layer would constitute the depth-of-burn measurement at each location.
In order to determine the bulk density of the forest floor a series of 3 randomly located 0.25 m2 (0.5×0.5 m) sections of forest floor were removed and sectioned into lichen and duff layers. All materials were then ovendried and weighed. The lichen layer had an average depth of 6.9 cm (ODW 1.390 kg m-2) and the duff layer averaged 4.0 cm in depth (ODW of 2.879 kg m-2). The total forest floor therefore averaged a depth of 10.9 cm with an ovendry weight of 4.269 kg m-2.
At each of the 12 grid points a thermocouple/data logger system was installed in order to determine the spread rate and residence time of the fire. The thermologgers are designed to measure and record flame zone temperatures and their durations, above and below the top of the forest floor litter (lichen) layer. Three thermocouple probes were mounted at +5 cm, -1 cm, and -3 cm with respect to the top of the lichen layer, and connected to a buried datalogger.
2. Fire Weather
Daily fire weather data, recorded at the Avialesookhrana (Aerial Fire Protection Service) Base at the Bor airport, was used to track the development of fire danger conditions in the region of Bor Forest Island from the beginning of the 1993 fire season through the conducting of the Bor Forest Island Experimental Fire. Component codes and indices of the Canadian Forest Fire Weather Index (FWI) System (Van Wagner 1987), a subsystem of the Canadian Forest Fire Danger Rating System (Stocks et al. 1989), were calculated daily based on noon Local Standard Time (LST) measurements of dry-bulb temperature, relative humidity, windspeed and 24-hour precipitation. The Russian Nesterov Index (Nesterov 1949), based on noon LST measurements of dry-bulb and dew-point temperature and precipitation, was also calculated during this period.
The 1993 forest fire season in the central portion of Krasnoyarsk Region began in typical fashion. Following the usual cold and dry winter, snow cover disappeared in early May, although the month of May remained quite cool with low fire danger conditions. June was a much warmer and drier period, with temperatures consistently above 20oC, windspeeds generally above 10 km h-1, and very scant precipitation. Fire danger conditions were generally extreme by late June, but moderated substantially after more than 25 mm of rain fell during the 28 June-1 July period. However, no further precipitation occurred prior to the Bor Forest Island Experimental Fire on 6 July, and fire danger steadily increased as temperatures rose and humidity levels decreased. By 6 July in Bor fire danger conditions were high to extreme. The Duff Moisture Content (DMC) and Buildup Index (BUI) of the Canadian FWI System were at levels indicating moderate to high fuel consumption, while the Fine Fuel Moisture Content (FFMC) and Initial Spread Index (ISI) levels also indicated moderate to high spread rates. Bor fire weather data for the 1 June to 6 July period is listed in Table 9. Weather measurements taken on Bor Forest Island on 5 and 6 July were generally consistent with observations in Bor, with the exception of slightly lower windspeeds.
Tab.9. Canadian FWI System codes and indices for Bor Airport (temperature inoC, windspeed in km h-1, rain in mm, FFMC=Fine Fuel Moisture Code, DMC=Duff Moisture Code, DC=Drought Code, ISI=Initial Spread Index, BUI=Buildup Index, and FWI=Fire Weather Index).
3. Fuel Moisture
As the moisture content of ground, surface and aerial fuels is critically related to fuel consumption and fire behavior, a large number of fuel samples were collected immediately prior to ignition of the Bor Forest Island Fire. These samples were subsequently oven-dried, and the average moisture contents of selected fuel strata determined as a function of ovendry weight. Fine downed and dead woody fuel moisture contents were consistent at about 7.5% for material under 3 cm in diameter. Forest floor fuel moisture content levels showed a strong demarkation between lichen and duff materials, with lichen values of 8.9% (0-4 cm) and 11.2 % (4+ cm) contrasting with duff values of 50.2% (0-2 cm) and 104.1% (2-4 cm). The moisture content of one year-old Pinus sylvestris needles averaged 74.7%. In general, these fuel moisture values agree closely with those forecast by the Canadian FWI System codes and indices, and indicate a high level of fuel consumption could be expected during the experimental burn.
4. Fire Behavior
(a) Ignition Procedure
Since the primary purpose of the Bor Forest Island Fire Experiment was the creating of a high-intensity, stand replacement fire, ignition along the windward side of the Island, with subsequent headfire development was considered essential. Winds were light and variable on 6 July, and weather conditions measured on-site at 1300 h were similar to those observed at the Bor airport weather station. With light winds (~7 km h-1) from the SE, ignition began along the east side of Bor Forest Island at 1420 h, using hand-held torches (Figure 9). By the time this ignition line was complete (1436 h), however, winds had shifted 90o to the SW. This sudden wind change turned the original ignition line into a backing fire, and it was necessary to begin a second ignition line along the west side of the Island in order to obtain a headfire effect. This line (approximately 500 m) was ignited between 1515 and 1520 h, and the two ignition lines began slowly moving together. Winds were still light and variable in direction, however, and a decision was made to complete ignition along the southern and northern edges of the Island, effectively creating a perimeter-ignited, convection-driven fire. This phase of the ignition process began at 1530 h, being completed along the souther edge at 1535 h, and along the northern edge at 1550 h (see Figure 9). The complete ignition phase was monitored and documented by helicopter-mounted cameras.
Fig.9. Ignition sequence of Bor Forest Island Experimental Fire, July 6, 1993.
(b) Rate of Spread
In the early phases of the fire spread rates were slow, with surface fire predominating although some isolated torching took place (Figures 10 and 11), particularly when the fire reached the crest of the sloped sides of the Island. As the fire progressed, however, an increasingly stronger convective influence took place as the fire lines accelerated inward. Crowning became more continuous, with a fire storm developing in the final phase of the burn. Although it was impossible to observe visually, thermologger measurements indicated the active phase of the burn was complete by approximately 1550 h. The fact that the Bor Forest Island Fire did not burn as a running headfire complicated the use of thermologger data as an aid to quantifying spread rate. However, an average spread rate was determined from points along the windward edge where crowning became continuous to the fire convergence zone. Over this distance the fire spread at an average rate of 25 m min-1 or 0.42 m s-1.
Fig.10. Aerial view of ignition along west side of Bor Forest Island (1522 h).
Fig.11. Ground view of typical high-intensity surface/intermittent crown fire behavior on north side of Bor Forest Island (1545 h).
(c) Residence Times
The thermologger data provided a clear picture of the combustion rate and residence time of the Bor Forest Island Fire, as usable data was retrieved from 11 of the 12 loggers. Figure xx illustrates temperature versus time recorded at a typical grid point exposed to crown fire, near the center of the burned area where convection-influenced firelines converged. At this point the +5 cm thermocouple recorded a peak temperature of 850oC at 1548 h, remaining above 500oC for 1.17 min., and above 100oC for 3.84 min. The mean residence times above 500oC (at 11 locations) for the +5 cm, -1 cm, and -3 cm thermocouples were 1.07, 1.18, and 1.18 min. respectively. Corresponding mean residence times above 100oC were 5.64, 4.72, and 5.63 min. respectively.
The Bor Forest Island Fire thermologger data represents primarily crown fire originating from a deep but low bulk density surface lichen layer. It can be inferred from Figure 12, which shows very little time delay in temperature peaks reached by the -1 cm and -3 cm thermocouples, that the lichen layer burned very quickly, probably almost entirely by flaming combustion. No firm estimate of smoldering combustion duration can be inferred from the thermologger data, under the assumption that the 5.64 min. mean residence time above 100oC for the +5 cm thermocouple represents flaming combustion. Smoldering in the forest floor behind the passage of the flaming front was observed to be minimal, however, with little residual smoke.
Fig.12. Typical temperature-time profile for Bor Forest Island Experimental Fire.
(d) Fuel Consumption
Following the fire, line intersects transects were re-inventoried to determine the quantity of downed woody fuel remaining on-site, and therefore the amount of surface fuel consumed. Fuel consumption averaged 0.194 kg m-2 for 0-3 cm size class fuels, 0.270 kg m-2 for 0-7 cm size class fuels, and 0.541 kg m-2 for fuels >7 cm in diameter, translating into surface fuel consumption rates of 98%, 65%, and 50% for these three size classes. Overall downed woody fuel consumed was therefore 0.811 kg m-2 (~54%).
Depth-of-burn pins along each transect were also measured immediately following the fire. Depth-of-burn was slightly variable, but averaged 8.36 cm, with a standard deviation of 2.23 cm. This translates into consumption of all of the surface lichen layer (1.390 kg m-2) and an additional 1.46 cm (1.051 kg m-2) of the underlying duff (organic) layer. Total ground fuel consumption was therefore 2.441 kg m-2 (Figures 13a and 13b).
The short length of the Bor Forest Island Fire Experiment did not permit sampling of aerial (crown) fuels. However, an aerial estimate of the portion of the Island where crown fire consumed crown fuels determined that approximately 57% of the Island fell into this category (Figure xx). Using the diameter distribution of the Bor Forest Island stand, along with crown fuel weights measured in jack pine (Pinus banksiana) in Canada (Stocks and Walker 1975), and assuming aerial fuel consumption included needles and fine dead twigs (<1 cm diameter), a figure of 0.460 kg m-2 was estimated for crown fuel consumption. Total fuel consumption (ground, surface, and aerial) during the Bor Island Fire was therefore determined to be 3.712 kg m-2 (38.12 t ha-1).
Fig.13. Prefire ground fuel profile showing well-developed lichen layer (~7 cm) and underlying organic material (~4 cm) over mineral soil.
The frontal fire intensity concept of Byram (1959) was used to approximate the intensity, or energy release rate , of the Bor Forest Island Fire. The formula I=Hwr, where H represents the low heat of combustion (~18,000 kJ kg-1), w is the amount of fuel consumed (in kg m-2), r is the forward rate of spread in m s-1, and I is the frontal fire intensity in kW m-1. Using the spread rate and fuel consumption values determined earlier, a frontal fire intensity level of 28,062 kW m-1 was estimated for the Bor Forest Island Fire.
This level of energy release represents an extremely intense fire within the boreal ecosystem, and is the result of high fuel consumption levels in the flaming stage of combustion, in combination with the strong convective fire activity generated by the ignition pattern of the Bor Forest Island Fire. This extreme intensity was reflected in the strong vertical development of the convection column above the fire, which was estimated to have reached 5000 m during the most intense stage of the fire (Figure 14). The development of this most intense phase of the Bor Island Fire is evident from the high levels of crown fuel consumption evident in the center portion of the Island (Figures 15 and 16).
Fig.14. Well-developed convection column (5 km) above Bor Forest Island (1555 h).
Fig.15. Postfire ground-level view within high-intensity portion of Bor Forest Island Fire.
Fig.16. Aerial view of Bor Forest Island one day after the fire. Note the heavily crowned-out center portion of the Island, where convection-driven firelines converged.
5. Atmospheric Emissions
(a) Radiatively Active Trace Gases
A high-volume sampling system developed by NASA Langley atmospheric scientists was installed on an Aeroflot MI-8 helicopter and used to collect smoke samples immediately above the Bor Forest Island Fire. Particle-filtered samples were drawn through a probe mounted on the nose of the helicopter. This probe was coupled to a high-volume pump inside the helicopter by flexible hose. Each smoke sample was collected in a gas sampling bag, and then transferred into a stainless steel bottle for subsequent laboratory analysis to determine levels of carbon dioxide (CO2), carbon monoxide (CO), hydrogen (H2), and methane (CH4). Smoke sampling was conducted at altitudes as low as safety would permit, as determined by fire intensity and smoke turbulence. Flight paths chosen during smoke plume/column sampling were based on visual keys such as smoke color, flame characteristics, apparent turbulence, and combustion stage. Samples from the high-intensity flaming phase and low-intensity smoldering phase of combustion were targeted and collected during the fire.
The results from 13 smoke sampling runs are presented in Table 10. for three fire stages: flaming combustion during the surface fire phase (F1), flaming combustion during the high-intensity crowning phase (F2), and smoldering combustion when no flames were visible (S3). Emission ratios presented in this table were determined by measuring excess (above background) trace gas concentrations in the smoke plume and normalizing these values against CO2 levels. This relationship can be used to define combustion efficiency, and to develop emission factors (g product/g fuel burned) if fuel consumption levels are known.
Of major interest from Table 10 is the fact that samples collected during the high-intensity phase (F2) of the Bor Island Fire revealed elevated carbon monoxide emission ratios, suggesting lower combustion efficiency than previously inferred from results obtained from Canadian boreal logging slash fires during flaming combustion (Cofer et al. 1990). Methane and hydrogen emission ratios, however, were similar to measurements obtained in the Canadian fires. During the smoldering combustion phase (S3) carbon monoxide emission ratios were almost three times higher than on Canadian logging slash fires.
It had been previously suggested that very high-intensity flaming combustion may significantly change the emissions chemistry associated with the flaming stage of combustion, leading to more incomplete combustion and correspondingly higher proportions of incompletely oxidized combustion products such as carbon monoxide (Cofer et al. 1989). The enhanced proportion of CO emissions during the vigorous flaming stage of the Bor Forest Island Fire seems to support this thesis, although additional data will be required to verify this. Trace gas emissions from this fire are analyzed in greater detail in Cofer et al. (this volume).
Tab.10. Mean CO2-normalized emission ratios and standard deviations (in %) determined for the Bor Forest Island Fire
* Combustion Phase (F=Flaming; S=Smoldering), Stage of Fire (1,2,3), and number of Samples ( )
(b) Compounds Affecting Stratospheric Ozone
To compliment the NASA trace gas emissions measurements, both helicopter and ground-based grab sampling (using stainless steel vacuum canisters) of emissions for specific analysis of methyl bromide (CH3Br) and methyl chloride (CH3Cl) was also carried out during the Bor Forest Island Fire. Decay products of these compounds are, like the longer-lived chloro-fluorocarbons (CFCs), known to induce depletion of stratospheric ozone. It should be noted here that bromine is much more efficient on a per atom basis than chlorine in breaking down ozone (by a factor of about 40) (WMO, 1992).
The emission ratios of CH3Br and CH3Cl measured in the Bor Forest Island Fire were in the range of 1.1-31×10-7 and 0.2-12×10-5 respectively. This was considerably higher than those found in savanna and chaparral fires or in laboratory experiments (cf. Manö and Andreae, 1994) . Highest values were found over smoldering surface fuels. This can be explained by the lower combustion efficiency of the smoldering process when compared to the prevailing flaming combustion of grass-type fuels.
Estimates of global pyrogenic emissions of CH3Br from all vegetation fires and other plant biomass burning falls in the range of 10-50 Gg yr-1, or 10-50% of the total source strength (Manö and Andreae, 1994). An accurate determination of the contribution of boreal fires to the atmospheric budget of methyl bromide requires further analysis of boreal fire emissions.
In connection with the lake sediment coring investigation discussed earlier, a study of the dispersion of particles emitted from the Bor Island Fire was also undertaken. A series of twenty-one 400 cm2 traps were arrayed along three transects radiating away from the burn into the surrounding fen, in order to estimate the production and transport of “large” (10o micron) particles. Traps were located at 5 to 10 m intervals, beginning slightly within the burn edge and extending to >60 m from the burn. Traps were placed on the fen surface prior to ignition, filled with deionized water, and collected the following day when smoldering emissions had ceased. Total particle fluxes and particle size distributions were determined using microscopy and image analysis (Clark and Hussey 1996).
The emission factor based on particle flux at the ground surface within the burn (72.9 g m-2) and average fuel consumption (3.712 kg m-2) is 0.212 kg kg-1, substantially higher than is typical for aerosols within a buoyant plume. This high value can be attributed to the fact that it was obtained for large particles trapped at the ground surface. Settling is an important removal mechanism for these large particles, so they are rapidly depleted within a plume and generally not considered in aerosol measurements. The ground-level traps collected particles just slightly elevated from the ground surface.
Particle deposition declines sharply within 5 m of the burn edge (Figure 17a). Although there is scatter in the data, there is no trend over the interval from 10 to 70 m, indicating that depletion of particles is minimal. It is clear from Figure 17a that traps are required at greater distances from the burn in order to adequately characterize the pattern of deposition. The high intensity level and well-developed convection column of the Bor Island Fire probably lofted particles sufficiently high enough that the gradient in particle deposition spans a much larger distance than was sampled here.
The interpretation that deposition is rather uniform within 100 m of the burn is supported by particle size distributions (Figure 17b). The largest (>101 mm) particles are restricted to traps from within the burn. Maximum particle size was lowest (100.8 mm) for the most distant (>20 m) traps. This relationship between maximum particle size and distance is consistent with removal of the largest particles by settling. But the distributions are nonetheless highly similar over most of the particle size range (Figure 17b). If settling was an important influence on deposition, we would expect to see the slope steepen with distance as large particles are preferentially depleted. We expect that traps at greater distances from the burn would have diameter distributions with much steeper slopes than those close to the burn.
Fig.17. a. Particle accumulation in traps/ distance from the edge of the Bor Forest Island Fire. b. Particle size distributions for all Bor Forest Island traps.