- Research Article
- Open Access
Intercomparison of Fire Size, Fuel Loading, Fuel Consumption, and Smoke Emissions Estimates on the 2006 Tripod Fire, Washington, USA
Fire Ecology volume 10, pages 56–83 (2014)
Land managers rely on prescribed burning and naturally ignited wildfires for ecosystem management, and must balance trade-offs of air quality, carbon storage, and ecosystem health. A current challenge for land managers when using fire for ecosystem management is managing smoke production. Smoke emissions are a potential human health hazard due to the production of fine particulate matter (PM2.5), carbon monoxide (CO), and ozone (O3) precursors. In addition, smoke emissions can impact transportation safety and contribute to regional haze issues. Quantifying wildland fire emissions is a critical step for evaluating the impact of smoke on human health and welfare, and is also required for air quality modeling efforts and greenhouse gas reporting. Smoke emissions modeling is a complex process that requires the combination of multiple sources of data, the application of scientific knowledge from divergent scientific disciplines, and the linking of various scientific models in a logical, progressive sequence. Typically, estimates of fire size, available fuel loading (biomass available to burn), and fuel consumption (biomass consumed) are needed to calculate the quantities of pollutants produced by a fire. Here we examine the 2006 Tripod Fire Complex as a case study for comparing alternative data sets and combinations of scientific models available for calculating fire emissions. Specifically, we use five fire size information sources, seven fuel loading maps, and two consumption models (Consume 4.0 and FOFEM 5.7) that also include sets of emissions factors. We find that the choice of fuel loading is the most critical step in the modeling pathway, with different fuel loading maps varying by 108 %, while fire size and fuel consumption show smaller variations (36 % and 23 %, respectively). Moreover, we find that modeled fuel loading maps likely underestimate the amount of fuel burned during wildfires as field assessments of total woody fuel loading were consistently higher than modeled fuel loadings in all cases. The PM2.5 emissions estimates from Consume and FOFEM vary by 37 %. In addition, comparisons with available observational data demonstrate the value of using local data sets where possible.
La gente que está a cargo de la gestión del territorio utiliza quemas controladas e igniciones naturales para el manejo de los ecosistemas y debe de buscar compensar los costos de transacción que emergen entre la protección de la calidad del aire, la captura de carbón y la salud de los ecosistemas. El manejo de la producción de humo es un reto actual para la gestión del territorio con enfoque de ecosistemas. Las emisiones de humo son un peligro potencial para la salud humana debido a la producción de partículas finas (PM2.5), monóxido de carbono (CO) y precursores de ozono (O3). Además, las emisiones de humo pueden afectar la seguridad en el transporte y generar problemas de visibilidad a escala regional. La cuantificación de las emisiones resultantes por fuego es un paso crítico para la evaluatión del impacto del humo sobre la salud y el bienestar humano y además es una variable requerida para producir modelos sobre calidad del aire y emisiones de gases de efecto invernadero. La modelización de emisiones de humo es un proceso complejo que requiere la incorporación de múltiples fuentes de datos, la aplicación de conocimiento de disciplinas cientifícas divergentes y la vinculación de varios modelos científicos en una secuencia lógica y progresiva. Típicamente, las estimaciones del tamaño de fuego, cargas (biomasa disponible) y consumo de combustibles resultan necesarios para calcular la cantidad de contaminantes producidos por un fuego. Aquí examinamos el Complejo del Incendio Tripod (Tripod Fire Complex) de 2006 como caso para estudiar bases de datos alternativas y la combinación de modelos científicos disponibles para calcular las emisiones por incendios. Especificamente, utilizamos cinco fuentes para el tamaño de los incendios, siete mapas de cargas de combustibles y dos modelos de consumo (Consume 4.0 y FOFEM 5.7), que también incluyen series de factores de emisión. Averiguamos que la elección de cargas de combustibles es el paso más crítico en la elaboración de modelos, con mapas que varían hasta un 108 %, mientras que el tamaño del incendio y el consumo de combustibles varían menos (36 % y 23 % respectivamente). Encontramos que es probable que los mapas de modelización de cargas de combustibles subestimen la cantidad de combustible consumido durante los incendios, ya que las mediciones reales fueron superiores a las cantidades modelizadas en todos los casos. Las estimaciones de emisiones de PM2.5 generadas por Consume y FOFEM varían en 37 %. Las comparaciones realizadas con datos empíricos demuestran el valor de utilizar bases de datos locales cuando sea posible.
Naturally ignited wildfires and prescribed fires play a key ecological role by regulating species composition, forest structure, hazardous fuels, and species regeneration in many wildland ecosystems (Weaver 1951, Cooper 1960, Agee 1998). In fact, following the 1995 Federal Wildland Fire Management Policy and Program Review, many local, state, and federal agencies emphasized the need to re-introduce fire into fire-adapted landscapes in their land management plans (NWCG 2001, Stephens and Ruth 2005). However, smoke emissions from adding more fire to the landscape constitute a potential human health hazard due to the production of harmful pollutants such as finegrained particulate matter (particles ≤2.5 microns in diameter, or PM2.5), carbon monoxide (CO), and ozone (O3) precursors (Hardy et al. 2001, Reinhardt and Ottmar 2004, Larkin et al. 2009, Ottmar et al. 2009, Koichi et al. 2010). Such fires also affect human welfare by contributing to regional haze, reducing visibility, and increasing concentrations of greenhouse gases. Smoke impacts on health and visibility are a growing concern due to the number of severe fire seasons that have occurred since 2000, largely as a result of warmer temperatures, drier environments, and increased fuel loadings from historic fire suppression efforts (Mutch 1994, Ferry et al. 1995, Westerling et al. 2006). A key challenge for land managers is balancing the need for fire to maintain ecosystem health while managing smoke emissions.
To address potential smoke impacts due to burning, land managers must first understand the magnitude of smoke emitted by wildfires on local, regional, and global scales (Knorr et al. 2012). To quantify smoke emissions, land managers have turned to scientific models that are intended to provide reasonable estimates of smoke emitted by biomass burning during both wildfire and prescribed fire events (Larkin et al. 2009, Ottmar et al. 2009, French et al. 2011, Knorr et al. 2012). Smoke emissions modeling is a complex process that requires the combination of multiple sources of data, the application of scientific knowledge from divergent scientific disciplines, and the linking of various scientific models in a logical progressive sequence.
Decision support frameworks have been developed to provide modeling pathways that guide users through the complicated smoke emissions modeling process in a series of steps, including (1) identifying fire size and location; (2) determining fuel loadings; (3) estimating fuel consumption; (4) producing realistic estimates of smoke emissions (Figure 1); and, in some cases, (5) predicting smoke plume transport, dispersion, and chemical properties. Examples of such frameworks include the BlueSky Framework (Larkin et al. 2009) and the Wildland Fire Emissions Information System (WFEIS; French et al. 2011). A variety of data sets and models are available within the framework for performing each step in the modeling pathway (Larkin et al. 2009, Goodrick et al. 2012, McKenzie et al. 2012), and the selection of data sets and models can have a significant impact on the resulting estimates of smoke emissions (N. Larkin, USDA Forest Service, Seattle, Washington, USA, unpublished data; Phase 1 of the Smoke and Emissions Model Intercomparison Project (SEMIP): test cases, methods, and analysis results; http://www.airfire.org/projects/semip).
The Smoke and Emissions Model Intercomparison Project (SEMIP) (N. Larkin. unpublished data) was designed to create an open standard for comparing smoke and emissions models against each other and against realworld observations. In this study, we apply the SEMIP intercomparison methods and framework to the 2006 Tripod Fire Complex, a large fire event that occurred in Washington state, USA (N. Larkin, unpublished data). We intercompare five fire reporting systems, seven fuel loading maps, and two fuel consumption models, which, when combined into a modeling pathway, lead to estimates of fire emissions. The results of each step in the modeling chain (Figure 1) are examined and, where possible, compared to observations. The goal of our study was to evaluate uncertainties in emissions estimates by quantifying the variability in results produced by various modeling pathways at each modeling step. The Tripod Fire Complex event provides a single large fire case study and illustrates how the variability at each modeling step can be quantified for individual fires.
Smoke emissions modeling studies. Three recent papers provide interesting insights into the potential for different outcomes when using models to generate smoke emissions. Ottmar et al. (2009) reviewed current knowledge, data, and types of predictive models commonly used for smoke emissions modeling. Knorr et al. (2012) discussed issues surrounding smoke emissions modeling on a global scale. French et al. (2011) discussed potential variability in smoke emissions modeling using carbon as a specific pollutant. From this literature, it is clear that uncertainty exists at each step in smoke emissions modeling pathways. Ottmar et al. (2009) and Knorr et al. (2012) suggest that the largest potential differences are found during the fuel characterization and fuel consumption step when modeling smoke emissions. These authors found large differences in fuel loading and fuel consumption based on vegetation type and fuel classification system. Large differences were also observed in the fire detection and fire size estimates in these studies. The Ottmar et al. (2009) review further suggests that reducing uncertainty in the fuel characterization and fuel consumption steps will reduce uncertainty in smoke emissions modeling across scales (global, national, regional, single fire) and fire types (wildfire or prescribed fire). In the French et al. (2011) study, estimates of carbon emissions from wildfires were found to vary regionally. Their study attributed the observed differences in carbon emissions to high variability in fuel loading across regions because modeled emissions were higher in regions with higher fuel loadings. Moreover, French et al. (2011) observed high variability in modeled fuel consumption across regions. In some regions, modeled fuel consumption outputs were very similar, while they varied widely in others. French et al. (2011) suggested a causal relationship, with higher fuel loadings leading to higher fuel consumption and ultimately higher carbon emissions.
In each of these studies, the observed differences in fire size, fuel loading, and fuel consumption were passed down the modeling pathway to affect the magnitude of modeled emissions outputs. These recent studies have improved our understanding of the complexities involved in modeling smoke emissions, yet the results indicate that the range of uncertainty at each step in smoke emissions modeling pathways must be evaluated further.
The Tripod Fire Complex. Lightning strikes on 3 Jul 2006 and 24 Jul 2006 ignited the Spur Peak Fire and the Tripod Fire in Washington state’s Okanogan-Wenatchee National Forest, USA. These fires burned together as the Tripod Fire Complex, one of the largest wildfire events in the state’s history. From July until November, the Tripod Fire Complex burned across a range of vegetation types, including grasslands, sage shrublands, Douglasfir (Pseudotsuga menziesii [Mirb.] Franco) forests, ponderosa pine (Pinus ponderosa Lawson & C. Lawson) forests, lodgepole pine (Pinus contorta Douglas ex Loudon) forests, and mixed-species forests. Fire severity (magnitude of environmental change, Lentile et al. 2006) was classified as moderate to high, with reports of the fire spreading as a mixture of variable-intensity crown and surface fires. Low fuel moisture contents and above-average fall temperatures enabled the fire to continue burning until it was extinguished by season-ending snow and rain events.
The size of the Tripod Fire Complex ensured that the event would be captured by numerous fire reporting systems, which made possible comparisons of reported fire location and size information. The multiple vegetation types burned during this large wildfire allowed for comparisons of fuel loading, fuel consumption, and smoke emissions across a range of vegetation types from grasslands to dense high-elevation, closed-canopy forests. In addition, fuels and fuel loadings had been extensively studied in the Tripod area, leading to the availability of several fuel loading maps for use in estimating biomass quantities. Moreover, a series of field-level fuels plots were sampled prior to the fire, allowing for direct comparison among modeled and observed fuel loadings, and post-burn field observations of fuel consumption were also available for comparison with modeled estimates. The availability of these data sources made the Tripod Fire Complex well-suited for evaluating the influence of fire size, fuel loading, fuel consumption, and smoke emissions analyses.
We focused on quantifying the impact of selecting various data sets and models for estimating fire size, fuel loading, fuel consumption, and emissions, because these were the most readily quantifiable steps in smoke emissions modeling pathways. The specific data sets and models compared for each of these steps were described.
Fire Size Information
A number of systems are in use by fire management personnel to identify wildfire size and location. For this analysis, we evaluated fire size information from the following five fire reporting systems (Figure 2):
ICS-209 incident reports. The ICS-209 report is the official form used to report fire size information for large incidents. These reports (available at http://fam.nwcg.gov/fam-web/hist_209/report_list_209) provide a set of coordinates indicating the fire ignition point; the spatial extent of the fire must be incorporated from other sources or approximated as a circle drawn from the fire origin point.
Monitoring Trends in Burn Severity (MTBS) maps. The primary objective of the MTBS project is to provide spatially explicit burn severity information for national analysis of fire severity trends (Eidenshink et al. 2007). MTBS maps (available at http://www.mtbs.gov) are produced using pre- and post-wildfire Landsat imagery to identify areas where the landscape burned during wildfires (Eidenshink et al. 2007).
NIFC and GeoMac. The National Interagency Fire Center (NIFC) keeps polygon records of fire perimeters constructed by geographic information system (GIS) specialists assigned to the incident. GeoMac is the interagency storage location for these maps, and the field-mapped perimeters are viewed as the official fire perimeter maps by local land management agencies (R. Harrod, USDA Forest Service, Okanogan-Wenatchee National Forest, personal communication).
Moderate Resolution Imaging Spectro-radiometer (MODIS) fire detection. MODIS uses Terra (AM; morning) and Aqua (PM; afternoon) satellite passes to provide products on fire location, energy emitted, flaming and smoldering ratios, and estimated area (number of pixels) burned (Giglio et al. 2009). For the Tripod Fire Complex, fire area was determined by aggregating all pixels detected by the MODIS fire detection product into a single raster file. If fire was detected within an individual pixel, it was assumed that the entire pixel burned.
MODIS burned area. The MODIS burned area product uses the land surface reflectance data from the Terra and Aqua satellite passes to determine the day of burning for each 500 m2 pixel. Fire size is estimated by combining all pixels where fire was detected during a particular set of days (Boschetti et al. 2009). If fire was detected within an individual pixel, it was assumed that the entire pixel burned.
We mapped fire size (area) perimeters in ArcGIS 9.2 for the Tripod Fire Complex from each of the data sources listed previously, and calculated the area covered by the fire for each reporting system. Percentage difference statistics were calculated for each fire size report using the NIFC fire perimeter as the standard fire perimeter.
Fuel Loading Maps
Fuel loading maps provide estimates of the biomass (fuels) available to burn across landscapes. For the Tripod Fire area, we used seven fuel loading maps to estimate the quantity of biomass available to burn during the fire event.
National Fire Danger Rating System (NFDRS) fire danger fuel model map. The NFDRS fuel model map provides a spatially consistent map of fuel loadings at a 1 km resolution for the continental United States (Burgan et al. 1997a, Burgan et al. 1998). The mapped NFDRS fuel models include fuel quantity information for live woody fuels (shrubs), herbaceous fuels, and downed and dead woody fuels. NFDRS maps do not include information on larger woody fuels, decomposed (rotten) woody fuels, canopy fuels, litter, or duff.
Hardy98. The Hardy98 vegetation cover and fuel loading map provides a spatially consistent map of fuel loadings at 1 km resolution for the western United States (Hardy et al. 1998). In this map, vegetation cover types comprise 18 broad categories created using an EROS Data Center LAND Characterization Class (USDI Geological Survey, EROS Data Center, Sioux Falls, South Dakota, USA) product. The fuel loadings by vegetation cover type are presented for live shrubs, live herbaceous fuels, downed woody fuels, litter, and duff.
Fuel Characteristic Classification System (FCCS1) original. The FCCS1 fuel loading map provides fuel loading information at a 1 km scale for the continental United States (McKenzie et al. 2007). In the FCCS1 fuel mapping process, one of 112 fuelbeds was assigned to each 1 km pixel in a national map using a rule-based assignment methodology (McKenzie et al. 2007). The FCCS fuelbed concept is the most comprehensive of the fuel maps and includes downed woody fuels, shrubs, herbs, grasses, canopy fuels, dead standing trees (snags), stumps, litter, moss, lichens, and duff.
Fuel Characteristic Classification System—Landscape Fire and Resource Management Planning Tools Project (LANDFIRE) crosswalk maps (FCCS2 30 m and 1 km). We refer to the FCCS-LANDFIRE crosswalk maps as FCCS2 to distinguish them from the original FCCS maps previously described. The FCCS2 30 m map provides standard FCCS fuelbeds and associated fuel loadings mapped at 30 m for the continental United States, Alaska, and Hawaii. The FCCS2 30 m map uses the LANDFIRE map product structure and is available as a LANDFIRE map product. To produce this map, the standard FCCS fuelbeds and LANDFIRE Existing Vegetation Types (EVT) map layer were cross-walked (D. McKenzie, USDA Forest Service, Pacific Northwest Research Station, personal communication). The 1 km FCCS2 map is an aggregated version of the 30 m FCCS2 map (McKenzie et al. 2012).
LANDFIRE Fuel Loading Models (FLMs). The LANDFIRE FLM map product provides fuel loading values for surface fuels at 30 m resolution across the continental United States (J. Herynk, Systems for Ecosystem Management, personal communication). The FLMs were produced from 4046 field plots where shrubs, herbs, downed woody fuel, litter, and duff were measured and classified into fuel loading models based on expected emissions (Lutes et al. 2009, Sikkink et al. 2009).
Okanogan-Wenatchee National Forest FCCS custom fuelbeds map (OkWen custom FCCS). The OkWen fuelbed map consists of a set of 83 custom FCCS fuelbeds mapped at 25 m resolution across the Okanogan-Wenatchee National Forest. The map was created by linking custom fuelbeds to a set of Forest Service vegetation layers (McKenzie et al. 2007).
To intercompare fuel loading maps, we prepared vegetation maps for each fuel loading data source. Additional maps were created for each fuel strata (total fuel loading, surface fuel loading, canopy, shrubs, herbaceous, woody fuels, duff, and litter) and for each of the fuel loading maps. We calculated summary fuel loading statistics (minimum, first quartile, median, third quartile, maximum) in megagrams per hectare for each fuel map, and calculated relative differences (with FCCS2 1 km set as the reference map for coarse scale fuel loading maps and FCCS2 30 m serving as the reference map for fine scale fuel loading maps) to identify potential variability across the landscape by data source.
Fuel consumption modeling systems estimate fuel consumption on the level of tree stands (Reinhardt and Dickinson 2010). Spatial assessments of fuel consumption during a single large wildfire event can be estimated using the fuel loading maps discussed previously. In addition to total fuel consumption, a time profile estimate of fuel consumption by phase (flaming, smoldering, and residual) is also considered because the quantity and mixture of emissions varies by phase.
We used the First Order Fire Effects Model (FOFEM) 5.7 and Consume 4.0 fuel consumption models to calculate fuels consumed by the Tripod Fire Complex. FOFEM (Reinhardt et al. 1997, Reinhardt and Dickinson 2010) and Consume (Prichard et al. 2006) contain fuel consumption algorithms that estimate fuel consumption based on fuel loading (fuel available to burn) and fuel condition (moisture of fuels, fuel state). Consume 4.0 is the Python recoding of Consume 3.0, completed by the Michigan Technical Research Institute, Ann Arbor, USA.
To model fuel consumption, FOFEM and Consume require a standard set of inputs, including fuel loading data, fuel moisture by fuel type (e.g., 10 hr, 100 hr, duff), and the percentage of canopy consumed and area blackened. In addition, we developed fuel moisture inputs to the models using the Wildland Fire Assessment System (WFAS; Burgan et al. 1997b) and the Fire Emissions Prediction Simulator (FEPS; Anderson et al. 2004). The WFAS produces and maps daily estimates of 10 hr, 100 hr, and 1000 hr fuel moisture content for the continental United States, Alaska, and Hawaii. For the period and location of interest, WFAS consistently estimated 1000 hr fuel moisture at 8 %. Because WFAS does not provide daily estimates of duff fuel moisture, and daily estimates of 10 hr fuel moisture varied greatly over the time period studied, we used an existing default FEPS fuel moisture lookup table (Table 1) to determine duff and 10 hr fuel moisture values that correspond to a 1000 hr fuel moisture value of 8 %. We set the duff and 10 hr values to 25 % and 6 %, respectively.
We ran Consume and FOFEM on each of the seven fuel loading maps to identify variability in fuel consumption resulting from model choice. We set the inputs for each model to reflect a severe wildfire, and we set canopy consumption to 60 %. For Consume to calculate shrub consumption, we set percentage blackened to 95 %.
We summarized and plotted the FOFEM and Consume outputs by fuel loading map to evaluate fuel consumption variability. Summary statistics, including digital maps of percentage difference and the minimum, first quartile, median, third quartile, and maximum values, were calculated.
FOFEM 5.7 and Consume 4.0 estimate smoke emissions by applying emission factors to fuel consumption estimates (Ward et al. 1993). Emission factors are empirically derived algorithms that quantify the production of gases and particulates by the fire, including total particulate matter, PM2.5, particulate matter ≤10 microns in diameter (PM10), CO, carbon dioxide (CO2), methane (CH4), sulfur dioxide (SO2), oxides of nitrogen (NOx), and non-methane hydrocarbons (NMHC).
FOFEM 5.7 estimates emissions of PM10, PM2.5, CO, CO2, CH4, NOx, and SO2 using a set of default emission factors (Table 2; Ward et al. 1993, Hao 2003). Default combustion efficiencies of 0.97 for the flaming phase and 0.67 for the smoldering phase are used to determine the proportions of fuel consumed during each phase. Proportions of fuel consumed during flaming and smoldering are determined using Burnup, the combustion physics model in FOFEM (Albini and Reinhardt 1995).
Consume 4.0 provides several alternative emission factors for modeling smoke emissions. For the fuels burned during the Tripod Fire Complex, we estimated smoke emissions using the Consume default emission factors (Table 2); these settings are commonly used for modeling fuels across landscapes. The Consume default emission factors are averages of all emission factors in Consume for natural fuels and are assigned by fuel type (Douglasfir, mixed conifer, ponderosa pine, hardwood, juniper (Juniperus spp.), and sagebrush (Artemisia spp.), and consumption phase (flaming, smoldering, and residual smoldering) (Prichard et al. 2006). Consume estimates emissions for total particulates, PM10, PM2.5, CO, CO2, CH4, and NMHC, and reports emissions separately for the flaming and smoldering stages of burning (Prichard et al. 2006).
We compared smoke emissions for each of the seven fuel loading maps by producing plots of pollutant-specific emissions estimates generated by each model, including plots of relative differences between model pathway results. We also compared smoke emissions across fuel loading inputs to identify potential influences of fuel loadings on later steps in the smoke emissions modeling pathway. In all cases, we compared emissions estimates for the five pollutants common to both FOFEM and Consume: CO, CO2, CH4, PM2.5, and PM10.
The Okanogan-Wenatchee National Forest provided observational plot data from the late 1990s from the Pacific Northwest Current Vegetation Survey (CVS; Johnson 2001; E. Peterson, USDA Forest Service, Okanogan-Wenatchee National Forest, personal communication). The CVS woody fuels data provided direct observations to compare with modeled fuel loading data from each fuel loading map. We summarized the CVS woody fuel data four size classes (10 hr, 100 hr, 1000 hr, and total woody fuels). We calculated the biomass available to burn for each size class following Brown (1974). We intercompared the modeled and observed woody fuel loadings using box and whisker plotting techniques (Wilkinson 1982).
We linked MTBS satellite observations of burn severity (unburned, low, moderate, high) to field estimates of fuel consumption from Justice et al. (2010). When coupled with composite burn indices (CBI) that feature field estimates of fuel consumed in burn areas (Key and Benson 2006), MTBS burn severity maps can be used to provide a remotely sensed estimate of fuel consumption across landscapes. For the Tripod Fire Complex, Justice et al. (2010) coupled CBI methodology with post-fire field sampling of vegetation and fuels data to produce indices relating fuel consumption to MTBS burn severity. In our study, we used the Tripod fuel consumption indices provided by Justice et al. (2010) and the mapped MTBS burn severity classes for the Tripod Fire Complex to produce a composite map of observed fuel consumption estimates across the Tripod Fire Complex burn area.
To evaluate potential differences in smoke emissions due to alternative fire information data, fuel inputs, fuel consumption modeling, or alternative smoke emissions computation, we analyzed model output at each level in the modeling pathway (Figure 1). We compared fire size for each of the five fire reporting systems, and compared fuel loading among the seven fuel loading maps. Using each of the seven fuel loading maps as inputs, we compared fuel consumption and emissions for the two fuel consumption and emissions models.
To evaluate how model or data choice at each step in the modeling pathway potentially influences the final smoke emissions estimates, we developed two hypothetical pathways from fire size to smoke emissions. For one pathway, we always selected the highest values; that is, we chose the largest fire size from the available fire reporting systems, then selected the fuel loading map that predicted the highest quantities of biomass available to burn. Next, we selected the fuel consumption model with the highest predicted rates of consumption, and then applied the highest emissions rates produced by either FOFEM or Consume to the highest predicted fuel consumption quantities. We compared this “maximum outcome” pathway with a “minimum outcome” pathway, in which we selected the smallest fire size reported, then the smallest potential fuel loadings, and applied the lowest emissions rates to the lowest fuel consumption estimates.
Reported Fire Size
The NIFC system estimated the final fire size for the Tripod Fire Complex at 70 837 ha. This value was matched very closely by the ICS-209 reports (70 895 ha) and the MTBS system (71 307 ha). However, the MODIS fire detection system reported a final fire size of 99 045 ha, a 36 % difference from the NFIC estimate we used as our standard fire perimeter (Figure 2). The MODIS burned area product did not detect the Tripod Fire Complex because the Tripod burn area was characterized as snow or high aerosol.
Modeled Fuel Loading
Vegetation types were fairly similar across each fuel loading map investigated in this study (Figure 3). The general pattern of grasslands, shrublands, and open forests in the west-southwest portion of the burn area and denser conifer forests in the east and east-central portion of the burn area is consistent across the fuel loading maps studied. Fuel loading trends followed the vegetation type patterns, with lower fuel loadings to the west-southwest and higher fuel loadings in the east and east-central reaches of the Tripod Fire Complex burn area (Figure 4).
Across the seven fuel loading maps studied, total fuel loadings ranged from 2.7 million Mg to 8.8 million Mg (Table 3) for the area covered by the Tripod Fire Complex, a 108 % difference. Percentage difference is calculated as:
The older generation of fuel loading maps (NFDRS, Hardy98, and FCCS1) predicted less fuel across the landscape than the more recent fine-scale fuel maps (FCCS2 and OkWen Custom Fuelbeds). Of interest, the fine-scale LANDFIRE FLM fuel map estimates of burnable fuel were consistently among the lowest for all fuel strata (Table 3). Total fuel loadings for the FCCS-based fuel loading maps (FCCS1, FCCS2, OkWen) were significantly higher than the NFDRS, Hardy98, and LANDFIRE FLM maps (P < 0.003). The FCCS-based fuel loading maps (FCCS1, FCCS2, Ok-Wen) were not significantly different from one another when intercompared.
Modeled Fuel Consumption
When we examined gross biomass consumption totals for the Tripod Fire Complex, the FOFEM model consistently estimated lower fuel consumption than the Consume model, as shown in Table 4 and Figure 5. Across all fuel loading maps evaluated (Table 4), Consume estimated total fuel consumption values ranging from 2 446 474 Mg (NFDRS 1 km) to 5 585 824 Mg (OkWen 25 m), while FOFEM estimated total fuel consumption values ranging from 1 796 054 Mg (LANDFIRE FLM 30 m) to 4 397 196 Mg (OkWen 25 m). When we directly compared FOFEM fuel consumption outputs with Consume fuel consumption outputs, FOFEM consistently produced fuel consumption estimates approximately 23 % (percentage difference calculated using Equation 1) lower than Consume across all seven sets of fuel loading inputs. For example, the FOFEM fuel consumption estimates for the OkWen fuel map (4 397 106 Mg) were 24 % lower than the Consume fuel consumption estimates (5 585 824 Mg) for the same fuel loading map.
Fuel consumption also varied by fuel type and fuel strata. For example, when estimating total downed woody fuel consumption, Consume estimated a higher rate of consumption in the more open forests, grasslands, and shrublands located in the west and southwest portion of the Tripod Fire Complex burn area compared to FOFEM (Figure 5). Consume-based woody fuel consumption rates were only slightly higher than FOFEM estimates in the east-northeast sections of the burn area, characterized by conifer forests with deeper duff layers (Figure 5).
FOFEM estimated higher consumption rates for duff in the western regions of the fire area, while Consume-based duff consumption rates were higher than FOFEM-based rates in the eastern sections of the fire (Figure 5). For litter, FOFEM-based consumption rates were consistently higher than Consume-based rates across the Tripod Fire Complex burn area (Figure 5).
Modeled Smoke Emissions
Smoke emissions results followed an opposite pattern to fuel consumption, with FOFEM consistently estimating higher CO, CH4, PM2.5, and PM10 emissions than Consume (Table 5; Figure 6). For CO2, Consume produced higher emissions estimates than FOFEM; however, this was not significant except for the Hardy98 fuel loading map (P < 0.001). The particulate (PM2.5 and PM10) emissions estimates showed greater variance by fuel type, as Consume estimated higher particulate emissions for the open lands in the west and southwest regions of the fire, while FOFEM estimated higher particulate emissions in the eastern reaches of the fire, which were dominated by closed conifer fuel types (Figure 6). Modeled emissions estimates for PM2.5 were different only for the NFDRS (P < 0.001), Hardy98 (P < 0.001), and the OkWen custom FCCS fuel loadings cases (P < 0.02); PM10 emissions were different only when fuels inputs were provided by the NFDRS and Hardy98 fuel loading maps (P < 0.001).
Modeled Results vs. Observations
Field assessments of total woody fuel loading were consistently higher than modeled fuel loadings in all cases. Analysis using box and whisker plotting techniques revealed that the OkWen custom fuel loading map was the only map to provide fuel loading estimates within the range of the field observations, as shown in Figure 7. The median values and the central tendencies for all other modeled fuel loadings were far below the central tendencies for the field observations (Figure 7). Moreover, when we compared percentage differences, only the central tendency for the OkWen custom fuel model included zero (no difference to observations) within its range.
Observed fuel consumption rates prepared using the MTBS fuel severity maps, in combination with field-level observations of fuel consumption, were generally similar to the modeled fuel consumption within the Tripod Fire Complex burn area (Figure 8). Spatially, observed fuel consumption estimates were lower in the western and southern areas, which were dominated by more open forests, grasslands, and shrublands (Figure 8). Higher rates of fuel consumption were observed in the eastern reaches of the fire, where closed canopy conifer forests dominated the fuel type (Figure 8). FOFEM- and Consume-derived fuel consumption estimates followed a similar pattern and were generally within ±20 % of the MTBS-derived fuel consumption estimates (Figure 8). Overall, total fuel consumption estimated by FOFEM (56.8 Mg ha−1) was slightly lower than the MTBS total of 59.7 Mg ha−1, while total fuel consumption estimated by Consume (71.4 Mg ha−1) was higher than the MTBS total. Comparisons of individual fuel types across the landscape indicated that Consume tended to estimate higher fuel consumption than MTBS for the closed conifer fuel types with more biomass available to burn, while Consume-based fuel consumption rates were generally lower than MTBS for the more open stand types. In contrast, FOFEM consumption rates were generally lower than MTBS-derived rates for most fuel types (Figure 8).
Smoke emissions varied considerably when the maximum and minimum outcome modeling pathways were followed. The smallest fire size for this fire, reported by NIFC, was 70 837 ha. This is about 30 % lower than the largest fire size, reported by the MODIS fire detection product, of 99 045 ha, with an absolute difference of 28 208 ha. The non-detection by the MODIS burned area product was not considered in this analysis.
At the fuel loading step, modeled fuel loadings varied by a factor of 3 when compared across the NIFC fire perimeters. The NFDRS fuel loading map reported a value of 38.6 Mg ha−1, and the OkWen custom FCCS fuel loading map reported a value of 128.7 Mg ha−1 (Table 3). The maximum absolute difference in fuel loading was 90.1 Mg ha−1.
Examining only consumption model differences, using the NIFC fire perimeter to estimate fire size and the FCCS2 as the reference fuel loading, FOFEM estimated 57.6 Mg ha−1 of consumption, while Consume’s estimate (72.6 Mg ha−1) was 15.0 Mg ha−1 (or 26 %) higher (Table 4). When we compared fuel consumption for all fuel loading input options for FOFEM and Consume, fuel consumption varied from a low estimate of 25.3 Mg ha−1 (FOFEM and the LANDFIRE FLM fuel loading map), to a high of 85.9 Mg ha−1 (Consume and the OkWen custom fuel loading map).
We intercompared smoke emissions using the NIFC fire perimeter estimate with FCCS2 as the reference fuel loading map. FOFEM estimated 0.91 Mg ha−1 of PM2.5 emitted, while Consume estimated 0.63 Mg ha−1 of emitted PM2.5 (Table 5; Figure 9). When we compared smoke emissions for all fuel loading input options for FOFEM and Consume, smoke emissions estimates for PM2.5 varied from a low of 0.25 Mg ha−1 (Consume and the LANDFIRE FLM fuel loading map) to a high of 1.1 Mg ha−1 (FOFEM and the OkWen custom fuel loading map). The percentage difference values for all other emissions produced by FOFEM or Consume are shown in Figure 6.
The differences observed at each step clearly impacted subsequent steps in the modeling pathway. Using the high and low values at each step from fire size through PM2.5 emissions, we found that the variances are compounded. For the “maximum outcome” pathway, a fire size of 99 045 ha (MODIS fire detection), a fuel loading of 12 322 606 Mg (Ok-Wen), and fuel consumption of 8 503 708 Mg (Consume) results in a total PM2.5 emissions estimate of 106 574 Mg using FOFEM emissions (Table 6). In contrast, for the “minimum outcome” pathway, a fire size of 70 837 ha (NIFC), a fuel loading of 3 017 456 Mg (LANDFIRE FLMFootnote 1), and a fuel consumption of 1 796 054 Mg (FOFEM) yields a total PM2.5 emissions estimate of 17 467 Mg when Consume is applied (Table 6). These estimates differ by a factor of 6.
We noted clear differences at each step of the modeling pathway when estimating smoke emissions for the 2006 Tripod Fire Complex. Moreover, it was clear that differences observed at one step were then propagated through the modeling process. Quantifying differences at various positions in smoke modeling pathways will provide information on areas in which improvements in the modeling process are needed.
Reported Fire Size
The differences observed here show that fire size information gathered from various fire detection systems may vary considerably. Selecting the optimal fire size and location information is easier for a single large fire, such as the Tripod Fire Complex, where good field-based observations were available and fire perimeters are maintained within publicly available databases such as GeoMac and MTBS. In the fire community, NIFC fire perimeters are considered the most accurate fire size information available (R. Harrod, personal communication). For smaller fires, precise fire perimeters may not be available.
At broad scales, fire detection systems that include multiple fires, such as the MODIS fire detection or burned area products, are commonly used (Knorr et al. 2012). Based on our results, the Tripod Fire Complex emissions would not be included in a large-scale emissions assessment if the MODIS burned area product were used as the fire area source of information (the Tripod Fire Complex burn area was characterized as snow or high aerosol; therefore, no fire was recorded in the system). Moreover, if the MODIS fire detection product were used to identify fire size and location, the emissions for the Tripod Fire Complex would be overestimated because fire size was overestimated by the fire detect product when compared with the NIFC fire perimeter data. Overestimating fire size using MODIS fire detection data is due, at least in part, to our assumption that each fire detect burned an entire pixel (Soja et al. 2006). Omitting fires or overestimating fire size will greatly influence the amount of fuel provided to the fuel consumption models, and under- or over-estimates of fuel consumption will directly lead to under- or over-estimating smoke emissions.
The MTBS system evaluated for this single fire case was in good agreement with the NIFC fire perimeter. This suggests that, at least for assessing large wildfires, MTBS would produce a result closer to reality as identified by fire managers. However, using MTBS does not address the problem of detecting small fires, which need to be greater than 404 ha in the western United States and greater than 202 ha in the eastern United States to be included in the MTBS database.
Modeled Fuel Loading
The large differences noted in fuel loading for the Tripod Fire Complex (Table 3; Figure 4) illustrate different approaches to mapping fuels across landscapes. The current generation of fuel loading maps (FCCS2, OkWen, LANDFIRE FLM; McKenzie et al. 2007, Lutes et al. 2009, French et al. 2011) has become increasingly comprehensive since the creation of the NFDRS as a tool for fire danger rating (Burgan et al. 1997a, Burgan et al. 1998). Much of the resulting variability in the mapped fuel loadings we observed was due to the omission of specific fuels strata such as canopy, shrubs, litter, or duff layers in the older fuel maps (NFDRS, Hardy98). Use of the current generation of fuel loading maps resolves these issues, as these fuels maps were specifically created to model fuel consumption and smoke emissions and now include canopy fuels and improved modeling of litter and duff.
The current fuel loading maps (bottom three panels in Figure 4) reflect our improved understanding of fuels in fire-prone landscapes, yet quantifying the natural variability in fuels across landscapes continues to be a problem (Ottmar et al. 2009, Keane and Reeves 2012). The significant variability in fuel loading we observed among the LANDFIRE FLM map and the two FCCS-based map products illustrates this issue. For this case, the approach used to produce the OkWen fuel loading map, for which local land managers interacted with fire researchers to quantify fuel loadings across an extensive range of vegetation types and then mapped the fuels by vegetation type (McKenzie et al. 2007, Berg 2007), provided the best fit to the local condition. The OkWen custom FCCS fuel loadings largely agreed with the field data; this result was related, at least in part, to how the OkWen custom FCCS fuel loading map was produced.
Modeled Fuel Consumption
The good agreement among the modeled data and the MTBS satellite observations linked to the field observations of fuel consumption for the Tripod Fire Complex indicates that either FOFEM or Consume will provide acceptable estimates of fuel consumption for wildfires in the Pacific Northwest, provided that the fuels information supplied to the model characterizes the fuels appropriately. Smoke emissions modeling accuracy could be improved if fuel consumption algorithms were improved; however, much greater improvements to the estimates of smoke emissions could be gained by improving the estimates of the amount of fuel available to burn as the biggest differences, and largest uncertainties, were found among the estimates of fuel loading provided by the fuel loading maps and the field observations of fuel loading.
Modeled Smoke Emissions
The differences in modeled smoke emissions rates illustrate the importance of the selection of emission factors for smoke emissions modeling. FOFEM produced higher emissions for CO, CH4, PM2.5, and PM10, while Consume produced higher CO2 emissions. These results are due to different emission factors used in the FOFEM and Consume emissions calculations and the way each model handles the partitioning of fuel consumption into the flaming and smoldering phases. Further understanding of the emissions process, and translation into an empirical description, will help to reduce these differences.
This study demonstrates that the differences observed at one step of a modeling pathway are passed on to the subsequent steps of the pathway and influence the final results of modeling exercises, such as estimating smoke emissions. For the Tripod Fire Complex case study, the difference between maximum and minimum values at each modeling step was 33 % for fire size, 121 % for fuel loading, 130 % for fuel consumption, and 144 % for emissions estimation (for PM2.5). The compounded differences between the maximum and minimum possible pathways demonstrate the importance of selecting the modeling pathway best suited for the fire location and region.
Best Options for the Tripod Fire Complex
As our results show, considerable uncertainty can be associated with modeling smoke emissions. This does not mean that we should abandon the modeling approach for estimating smoke emissions. Rather, our results show that, with some careful thought, managers can pick an appropriate option at each step in a smoke emissions modeling pathway. For example, the best option for estimating the area burned by the Tripod Fire Complex was the MTBS fire perimeter, which was easily accessible and spatially verified by the MTBS project team. Locally specific fuel loading maps such as the OkWen fuel loading map (custom FCCS fuelbeds) should be used when available, as this fuel loading map best matched the field observed fuel loadings available for this project. If custom, locally produced fuelbeds are not available, the standard FCCS fuelbeds (FCCS2) available from LANDFIRE (www.landfire.gov) should be used.
Our study results illustrate how FOFEM, when compared to Consume, provided fuel consumption estimates that were closer to the fuel consumption estimates derived from the combined MTBS fire severity CBI field observations analysis; FOFEM therefore appears to be better than Consume for modeling both fuel consumption and smoke emissions on severe fires in this region. However, resource managers can and should perform retrospective analyses on historic fires in their region, such as we performed here, to determine which fuel consumption model performs best for the vegetation types that they are using to model smoke emissions at their local scale.
As noted throughout this paper, our best options for which models to use within a smoke modeling pathway are not the only options, and the results of a single case study should be used with caution; however, a smoke emissions pathway as outlined in this section will provide useful information needed to understand the tradeoffs between the ecological benefits of burning and the hazards due to the smoke emitted by burning.
Quantifying wildland fire emissions is a critical step in evaluating smoke impacts. Smoke emissions estimates require the use of modeling pathways that combine multiple sources of data and the linking of scientific models in a logical, progressive sequence. Fire emissions are typically modeled using information on fire size and location, coupled with fuel loading maps, which are then processed through consumption models capable of producing emissions estimates. In a complex modeling process, numerous options are available, and managers, scientists, and others who model fire emissions need to understand the uncertainties and differences in this process. Intercomparisons, such as the one conducted here, provide insights into the sensitivity of smoke emissions estimates to the variability present at each step in a modeling pathway. Use of accurate fire information and local fuels data is critical to reducing the amount of uncertainty in the overall modeling chain. The first step to providing better estimates of smoke emissions is to provide better estimates of the fuels available to burn. Under-or over-estimating fire size can lead to inaccurate representations of the amount of fuel available to burn, the amount of fuel consumed during a fire, and subsequently unrepresentative estimates of smoke emissions. Additional uncertainty when modeling smoke emissions can be introduced by the available fuel loading maps. For example, we found that all modeled fuel loading maps studied underestimated total woody fuel loading when directly compared with field-sampled fuel loading data. Locally produced fuel loading maps such as the OkWen custom FCCS fuel loading map can reduce some of the uncertainty due to fuel loading inaccuracies; however, our results suggest that fuel loading, and subsequent smoke emissions, are commonly underestimated on wildfire in the Pacific Northwest. Examination of other fires as part of SEMIP (N. Larkin, unpublished data) shows similar results, suggesting that the uncertainties identified here are important areas for future research and development.
Although NFDRS reported the lowest absolute fuel loadings, the LANDFIRE FLM pathway was used in this stage of the minimum possibilities pathway because the NFDRS did not include multiple fuel strata.
Agee J.K. 1998. The landscape ecology of western forest fire regimes. Northwest Science 72 (Special Issue 1998): 24–34.
Albini, F.A., and E.D. Reinhardt. 1995. Modeling ignition and burning rate of large woody natural fuels. International Journal of Wildland Fire 5: 81–91. doi: 10.1071/WF9950081
Anderson, G.K., D.V. Sandberg, and R.A. Norheim. 2004. Fire Emission Production Simulators (FEPS). User’s guide. USDA Forest Service Fire and Environmental Research Applications (FERA). <http://www.fs.fed.us/pnw/fera/>. Accessed 19 Feb 2014.
Berg, E. 2007. Characterizing and classifying complex fuels: a new approach. Canadian Journal of Forest Research 37: 2381–2382. doi: 10.1139/X07-150
Boschetti L., D. Roy, and A.A. Hoffmann. 2009. MODIS Collection 5 Burned Area Product—MCD45. User’s guide version 2.0. <https://earthdata.nasa.gov/sites/default/files/field/document/MODIS_Burned_Area_User_Guide_2.0.pdf>. Accessed 19 Feb 2014.
Brown, J.K. 1974. Handbook for inventorying downed woody material. USDA Forest Service General Technical Report INT-16, Intermountain Forest and Range Experiment Station, Ogden, Utah, USA.
Burgan, R.E., C.C. Hardy, D.O. Ohlen, and G. Fosnight. 1997a. Landcover ground sample data. USDA Forest Service General Technical Report INT-GTR-368CD, Intermountain Forest and Range Experiment Station, Ogden, Utah, USA.
Burgan, R.E., P.L. Andrews, L.S. Bradshaw, C.H. Chase, R.A. Hartford, and D.J. Latham. 1997b. Current status of the Wildland Fire Assessment System (WFAS). Fire Management Notes 57(2): 14–17. <http://www.fs.fed.us/fire/fmt/fmt_pdfs/fmn57-2.pdf>. Accessed 19 Feb 2014.
Burgan, R.E., R.W. Klaver, and J.M. Klaver. 1998. Fuel models and fire potential from satellite and surface observations. International Journal of Wildland Fire 8: 159–170. doi: 10.1071/WF9980159
Cooper, C.F. 1960. Changes in vegetation, structure, and growth of southwestern pine forests since white settlement. Ecological Monographs 30: 129–164. doi: 10.2307/1948549
Eidenshink, J., B. Schwind, K. Brewer, Z.-L. Zhu, B. Quayle, and S. Howard. 2007. A project for monitoring trends in burn severity. Fire Ecology 3(1): 3–21. doi: 10.4996/fireecology.0301003
Ferry G.W., R.G. Clark, R.E. Montgomery, R.W. Mutch, W.P. Leenhouts, and G.T. Zimmerman. 1995. Altered fire regimes within fire-adapted ecosystems. Pages 222–224 in: E.T. LaRoe, editor. Our living resources: a report to the nation on the distribution, abundance and health of US plants, animals and ecosystems. US Department of the Interior, National Biological Service, Washington, D.C., USA.
French, N.H.F., W.J. de Groot, L.K. Jenkins, B.M. Rogers, E. Alvarado, B. Amiro, B. de Jong, S. Goetz, E. Hoy, E. Hyer, R. Keane, B.E. Law, D. McKenzie, S.G. McNulty, R. Ottmar, D.R. Pérez-Salicrup, J. Randerson, K.M. Robertson, and M. Turetsky. 2011. Model comparisons for estimating carbon emissions from North American wildland fire. Journal of Geophysical Research 116(G4): G00K05. doi: 10.1029/2010JG001469
Giglio, L., T. Loboda, D.P. Roy, B. Quayle and C.O. Justice. 2009. An active-fire based burned area mapping algorithm for the MODIS sensor. Remote Sensing of Environment 113(2): 408–420. doi: 10.1016/j.rse.2008.10.006
Goodrick, S.L., G.L. Achtemeier, N.K. Larkin, Y. Liu, and T.M. Strand. 2012. Modelling smoke transport from wildland fires: a review. International Journal of Wildland Fire 22: 83–91. doi: 10.1071/WF11116
Hao, W.M. 2003. Unpublished emissions factors report. On file at USDA Forest Service Rocky Mountain Research Station, Missoula Fire Lab, Missoula, Montana, USA.
Hardy, C.C., R.D. Ottmar, J.L. Peterson, J.E. Core, and P. Seamon. 2001. Smoke management guide for prescribed and wildland fire. Report PMS 420-2. NFES 1279. National Wildfire Coordinating Group, Boise, Idaho, USA.
Hardy, D., J.P. Menakis, D.G. Long, and J.L. Garner. 1998. FMI/WESTAR emissions inventory and spatial data for the western United States. USDA Forest Service Rocky Mountain Research Station, Missoula Fire Lab, Montana, USA.
Johnson, M.D. 2001. Field procedures for the current vegetation survey, v. 2.04. USDA Forest Service, Pacific Northwest Region, Portland, Oregon., USA.
Justice, E., B. Cheung, W. Danse, K. Myrick, M. Willis, S. Prichard, and J.W. Skiles. 2010. Effect of fuel treatments on carbon flux during a wildfire using satellite imagery: Okanogan-Wenatchee National Forest. Unpaginated. Proceedings of an annual conference. American Society of Photogrammetry and Remote Sensing, 26–30 April 2010, San Diego, California, USA.
Keane, R.E., and M. Reeves. 2012. Use of expert knowledge to develop fuel maps for wildland fire management. Chapter 11. Pages 211–228 in: A.H. Perera, C.A. Drew, and C.J. Johnson, editors. Expert knowledge and its application in landscape ecology. Springer Science and Business Media, LLC, New York, New York, USA.
Key, C.H., and N.C. Benson. 2006. Landscape assessment: ground measure of severity, the Composite Burn Index; and remote sensing of severity, the Normalized Burn Ratio. Pages LA 1–55 in: D.C. Lutes, R.E. Keane, J.F. Caratti, C.H. Key, N.C. Benson, S. Sutherland, and L.J. Gangi, editors. FIREMON: Fire Effects Monitoring and inventory system. USDA Forest Service General Technical Report RMRS-GTR-164-CD, Rocky Mountain Research Station, Ogden, Utah, USA.
Knorr, W., V. Lehsten, and A. Arneth. 2012. Determinants and predictability of global wildfire emissions. Atmospheric Chemistry and Physics 12: 6845–6861. doi: 10.5194/acp-12-6845-2012
Koichi, I., G.H. Donovan, P.A. Champ, and J.B. Loomis. 2010. The economic cost of adverse health effects from wildfire-smoke exposure: a review. International Journal of Wildland Fire 19: 803–817. doi: 10.1071/WF09077
Larkin, N.K., S.M. O’Neill, R. Solomon, S. Raffuse, T.M. Strand, D.C. Sullivan, C. Krull, M. Rorig, J. Peterson, and S.A. Ferguson. 2009. The BlueSky smoke modeling framework. International Journal of Wildland Fire 18: 906–920. doi: 10.1071/WF07086
Lentile, L.B., Z.A. Holden, A.M.S. Smith, M.J. Falkowski, A.T. Hudak, P. Morgan, S.A. Lewis, P.E. Gessler, and N.C. Benson. 2006. Remote sensing techniques to assess active fire characteristics and post-fire effects. International Journal of Wildland Fire 15: 319–345. doi: 10.1071/WF05097
Lutes, D.C., R.E. Keane, and J.F. Caratti. 2009. A surface fuels classification for estimating fire effects. International Journal of Wildland Fire 18: 802–814. doi: 10.1071/WF08062
McKenzie, D., C.L. Raymond, L.K.B. Kellogg, R.A. Norheim, A.G. Andreu, A.C. Bayard, and K.E. Kopper. 2007. Mapping fuels at multiple scales: landscape application of the fuel characteristic classification system. Canadian Journal of Forest Research 37: 2421–2437. doi: 10.1139/X07-056
McKenzie, D., N.H.F. French, and R.D. Ottmar. 2012. National database for calculating fuel available to wildfires. EOS 93(6): 57–58. doi: 10.1029/2012EO060002
Mutch, R.W. 1994. Fighting fire with prescribed fire: a return to ecosystem health. Journal of Forestry 92(11): 31–33.
NWCG [National Wildfire Coordinating Group]. 2001. Review and update of the 1995 federal wildland fire management policy. National Interagency Fire Center, Boise, Idaho, USA.
Ottmar, R.D., A. Miranda, and D. Sandberg. 2009. Characterizing sources of emissions from wildland fires. Pages 61–78 in: A. Bytnerowicz, M. Arbaugh, A. Riebau, and C. Andersen, editors. Developments in environmental science, volume 8. Elsevier Inc., Philadelphia, Pennsylvannia, USA. doi: 10.1016/S1474-8177(08)00003-X
Prichard, S.L., R.D. Ottmar, and G.K. Anderson. 2006. Consume 3.0 user’s guide. USDA Forest Service General Technical Report PNW-GTR-304, Pacific Northwest Research Station, Seattle, Washington, USA.
Reinhardt, E.D., R.E. Keane, and J.K. Brown. 1997. First Order Fire Effects Model: FOFEM 4.0 user’s guide. USDA Forest Service General Technical Report INT-GTR-344, Intermountain Forest and Range Experiment Station, Ogden, Utah, USA.
Reinhardt, E.D., and M.B. Dickinson. 2010. First-order fire effects models for land management: overview and issues. Fire Ecology 6(1): 131–142. doi: 10.4996/fireecology.0601131
Reinhardt, T.E., and R. Ottmar. 2004. Baseline measurements of smoke exposure among wildland firefighters. Journal of Occupational and Environmental Hygiene 1: 593–606. doi: 10.1080/15459620490490101
Sikkink, P., R.E. Keane, and D.C. Lutes. 2009. Field guide for identifying fuel loading models. USDA Forest Service General Technical Report RMRS-GTR-225, Rocky Mountain Research Station, Fort Collins, Colorado, USA.
Soja, A., J. Al-Saadi, B. Pierce, C. Kittaka, J. Szykman, D.J. Williams, T. Pace, L. Giglio, J. Kordzi, D. Randall, and T. Moore. 2006. How well does satellite data quantify fire and enhance biomass burning emissions estimates? Proceedings of the EPA 15th annual international emissions inventory conference: reinventing inventories—new ideas in New Orleans. US Environmental Protection Agency, 15–18 May 2006, New Orleans, Louisiana, USA. <http://www.epa.gov/ttn/chief/conference/ei15/session10/soja.pdf>. Accessed 19 Feb 2014.
Stephens, S. L., and L.W. Ruth. 2005. Federal forest-fire policy in the United States. Ecological Applications 15(2): 532–542. doi: 10.1890/04-0545
Ward, D.E., J. Peterson, and W.M. Hao. 1993. An inventory of particulate matter and air toxic emissions from prescribed fires in the USA for 1989. Pages 1–19 in: Proceedings of the Air and Waste Management Association 86th annual meeting and exhibition. Air and Waste Management Association, 13–18 June 1993, Denver, Colorado, USA.
Weaver, H. 1951. Fire as an ecological factor in the southwestern ponderosa pine forests. Journal of Forestry 49: 93–98.
Westerling, A.L., H.G. Hidalgo, D.R. Cayan, and T.W. Swetnam. 2006. Warming and earlier spring increase western US forest wildfire activity. Science 313(5789): 940–943. doi: 10.1126/science.1128834
Wilkinson, L. 1982. SYSTAT (Version 13). Cranes Software International, Chicago, Illinois, USA.
The authors thank the Joint Fire Science Program for providing funding for this project (08-16-10). We are grateful to E. Peterson and J. Restaino for providing the current vegetation survey data. R. Harrod and J. Croft kindly provided their time and answered many questions concerning the Tripod Fire Complex. S. Prichard, L. Gangi, R. Keane, J. Reardon, and D. Lutes provided useful insight into the inner workings of Consume and FOFEM. D. McKenzie and J. Herynk provided technical advice on the construction of the FCCS and LANDFIRE FLM maps, respectively. J. Menakis provided data on the Westar project. S. Reid, J. Schwartz, C. Jennings, M. Michaels, M. Teplitz, M. Protteau, and L. Chinkin provided useful comments and edits of the text, tables, and figures.
About this article
Cite this article
Drury, S.A., Larkin, N.S., Strand, T.T. et al. Intercomparison of Fire Size, Fuel Loading, Fuel Consumption, and Smoke Emissions Estimates on the 2006 Tripod Fire, Washington, USA. fire ecol 10, 56–83 (2014). https://doi.org/10.4996/fireecology.1001056
- fire size
- fuel consumption
- smoke emissions
- Tripod Fire Complex