B11D-0761
Interactive effects of fire, soil climate, and vegetation on CO2 fluxes in an upland black spruce forest and peatland in interior Alaska
Fire is an important control on the carbon (C) balance of the boreal forest. In addition to the immediate release of stored C to the atmosphere through organic matter combustion, fire has the capacity to alter controls on decomposition, through changes in soil climate and substrate quality. Here, we present findings from two complimentary studies that examine how fire modifies the physical properties of soil and how these modifications influence rates of decomposition and C exchange in Alaska's boreal forest. First, we conducted a laboratory study to evaluate the interactive effects of fire, soil temperature, soil moisture, and moss type on CO2 fluxes from organic soils. Second, we conducted intensive field measurements of ecosystem CO2 fluxes in a 3 year-old burn to evaluate the effect of fire on carbon exchange in an upland forest and peatland in interior Alaska. Incubation CO2 fluxes showed a significant interaction between burn status (burned, unburned sites), temperature (2 °C vs. 20 °C treatments), and moisture treatment (field moisture vs. saturated). Incubation CO2 fluxes in the unburned sites increased with temperature by a factor of 5 and 29, whereas incubation CO2 fluxes in the burned sites only increased by 6 to 8 times. Incubation CO2 fluxes in the unburned sites increased with moisture content between 19 and 24 times, whereas incubation CO2 fluxes in the burned sites only increased by a factor of 4. Incubation CO2 fluxes from unburned Sphagnum samples were nearly 3 times greater than fluxes from burned Sphagnum. In the field At 3 years post-burn, moisture content was higher in the burned upland forest and burned peatland relative the unburned sites. However, soil temperature was not significantly different between burned and unburned sites. Mean rates of net ecosystem exchange (NEE) showed greater rates of CO2 uptake in the unburned peatland site than in the burned peatland, averaging -1.51 and 0.16 g C m-2 d-1, respectively. NEE rates were not significantly different in the two upland forest sites, averaging 0.0005 g C m-2 d-1 across sites. Mean rates of ecosystem respiration (ER) were not significantly different between the burned and unburned upland forest sites, averaging 0.9 g C m-2 d-1 overall. ER rates were not different between the burned and unburned peatland sites, averaging 1.4 m-2 d-1 overall. Soil temperature and moisture content accounted for between 20 to 45 % of the variation in ER rates in unburned upland forest and unburned peatland, and less than 10 % in the two burned sites. These findings, together with the incubation study, suggest that while fire creates soil climate conditions more conducive to rapid decomposition, rates of C release from soils may be constrained following fire by changes in litter quality that slow rates of decomposition.
B11D-0762
Mercury Binding and Mobilization in Post-fire Soil Horizons
Fires affect watersheds in many ways, including through increased erosion and sediment transport, decreased water quality, and transport and cycling of nutrients and metals. This research addresses mercury (Hg) mobilization within post-fire stream systems, focusing on the influence of soil particle size on binding affinity and potential transport. Using a network of sampling sites within the Piru Creek watershed, affected by the Day Fire during September 2006, total mercury (THg) was measured in soils collected before, during, and after the 2006- 07 storm season. Unburned and burned soil samples from various levels of burn intensity were collected in one- inch increments to a depth of 6 inches and partitioned into fine, medium and coarse fractions. THg concentrations within each grain fraction were measured in triplicate using a Milestone Direct Mercury Analyzer. Initial findings indicate a loss of THg at the surface in the burned soils, as well as increased levels of THg at depths of 2-4 inches. We hypothesize this to be due to volatilization of Hg due to burning, which is either released upward into the atmosphere, or becomes bound to organic matter and settles just below the surface during the formation of a hydrophobic layer. Surface loss may also be attributed to post-fire erosional processes and storm transport. Additionally, analysis of the size fractionated soils reveals that the highest readings of THg in unburned soils occurred in the fine sediments in every case. THg concentrations in fine sediments in the burned soils were not significantly higher than THg concentrations in medium or coarse sands. More recently collected samples show evidence of continuing atmospheric deposition of Hg at the soil surface. Leaching tests are also being performed on this same set of soils to aid in evaluating potential mobilization of THg from transported soils during storm events.
B11D-0763
A Dynamic Soil Layer Model for Assessing the Effects of Wildfire on High Latitude Terrestrial Ecosystem Dynamics
Wildfire is considered an important disturbance to boreal ecosystems in North America. It can affect carbon dynamics directly through combustion emissions, and indirectly through vegetation succession and removal of the surface organic layer, which might accelerate the degradation of permafrost and hence the release of soil carbon. At the regional scale, the direct effects of fire have received a lot of attention, but the evaluation of the indirect effects has been more limited because the appropriate tools have not yet developed for application at the regional scale. In this study, we developed a dynamic soil layer model (DSLM) used in the Terrestrial Ecosystem Model (TEM) to investigate the effects of changes of surface organic layer on soil temperature, moisture, and carbon dynamics. The DSLM consists of (1) a simultaneous heat and water transfer scheme, in which a Two-Directional Stefan Algorithm was used to provide a stable and efficient simulation of soil thermal dynamics in both seasonal frost and permafrost regions; (2) an explicit vertical distribution of soil carbon; (3) a conceptual model of removal of organic layer by wildfire; (4) and a conceptual model of recovery of organic layer after wildfire. DSLM-TEM was calibrated for black spruce, white spruce, deciduous and tundra vegetation types. DSLM-TEM was first tested on a tundra burn site and two black spruce fire chronosequences for its performance on soil temperature and moisture simulation. Several sensitivity tests have then been performed to investigate the effects of different schemes of organic layer removal and recovery on permafrost and carbon dynamics. Initial results showed that: (1) soil temperatures and soil moistures were well simulated; (2) active layer depth was sensitive to the thickness of the organic layer; and (3) the simulated organic layer thickness can reasonably represent the dynamics of soil organic layer development after fire disturbance.
B11D-0764
Simulated Response of Conterminous United States Ecosystems to Climate Change at Different Levels of Fire Suppression, CO2,and Growth Response to CO2
The impact of fire management, CO2 levels, and the growth response to CO2 on the response of ecosystems to climate change scenarios produced by three different General Circulation Models (GCMs) was simulated for the conterminous United States by the MC1 Dynamic Global Vegetation Model (DGVM). Distinct regional trends in response to projected climatic change were evident across all combinations of the experimental factors. In the eastern half of the U.S., the average response to relatively large increases in temperature and decreases in precipitation was an 11 percent loss of total ecosystem carbon. In the West, the response to increases in precipitation and relatively small increases in temperature was an 5 percent increase in total carbon stocks. Simulated fire suppression reduced average carbon losses in the East to about 6 percent, and preserved forests which were largely converted to woodland and savanna in the absence of fire suppression. Across the west, unsuppressed fire maintained near constant carbon stocks despite increases in vegetation productivity. With fire suppression, western carbon stocks increased by 10 percent and most shrublands were converted to woodland or even forest. With a relatively high level of growth sensitivity to CO2, total ecosystem carbon pools at the end of the century were on average about 9-10 percent larger in both regions of the U.S. compared to a low level of CO2 sensitivity . The western U.S. gained enough carbon to counter losses from unsuppressed fire only with the high CO2 sensitivity, especially in conjunction with the higher CO2 level. In the eastern U.S., fire suppression was sufficient to produce a simulated carbon sink only with both the high CO2 level and high CO2 sensitivity. Considerable uncertainty exists with respect to the impacts of global warming on the ecosystems of the conterminous U.S., some of which resides in the future trajectory of atmospheric CO2, in the direct response of vegetation to increasing CO2, and in future tradeoffs among different fire management options, as illustrated in this study.
B11D-0765
Burn Severity Measurement Using LiDAR Data
LiDAR data were collected before and after a prescribed burn of sagebrush rangeland in eastern Idaho in the fall of 2005. These data were processed to filter vegetation points from bare earth and calculate the vegetation heights. The pre-fire and post-fire vegetation heights, as well as derived surface texture products, were compared to determine the amount of burned vegetation and calculate burn severity indices. The resulting classifications were then validated using field data from the burn site. While all products clearly reflected variations in burn intensity, the change in surface roughness was found to most accurately reflect the field- measured burn severity.
B11D-0766
Influence of Fire Severity on Watershed Nitrogen Cycling Using 15N Natural Abundance in Terrestrial and Aquatic Ecosystem Components
Fire is an integral component of ecosystem nitrogen (N) cycling in coniferous ecosystems of the Rocky Mountains. The objective of this study was to use N stable isotopes at natural abundance to study post-fire N cycling in small watersheds that experienced different fire severities. Within four wildfire, one spring test burn, and three spring prescribed burn sites we quantified and interpreted the N concentrations and N isotopic signatures of soil, plants, streamwater and in-stream moss, and the NO3- use by plants. We found short-term (1-3 y) post-fire increases of δ15N in these ecosystem N pools to be correlated with fire severity. After wildfires, δ15N significantly increased in all of the studied N pools (P < 0.05), whereas after spring burns only plant foliage δ15N significantly increased (P < 0.05), although with a smaller magnitude than after wildfire. For example, the δ15N of foliage of upland plants was enriched by 2.9 ‰\ (absolute difference between burned and unburned watersheds) in the first two years after wildfire, but only 1.3 ‰\ after spring burns. The simultaneous enrichments of both shoots and roots in wildfire-burned areas indicated that isotopic enrichment was caused by uptake of enriched soil N. This was corroborated by absolute increases in soil NH4+ δ15N by on average 4.6 ‰\ for the first two post-fire years relative to unburned areas (P < 0.05). In-stream moss δ15N in wildfire-burned watersheds was increased by 1.3 ‰\ relative to unburned watersheds, but there was no response in prescription-burned watersheds. A major difference between wildfire and spring prescribed burn effects was the substantial (two orders of magnitude), sustained (3 yr), and significant (P < 0.05) increase in streamwater NO3- concentrations after wildfire and the lack thereof after spring burns. The smaller or lacking isotopic response after spring prescribed burns likely reflected less volatilization of 14N during the lower-temperature burns and less altered N cycling processes (i.e., minor increases in net nitrification and subsequent lack of nitrate leaching) relative to wildfire. Thus, isotopic shifts in terrestrial plant foliage or in-stream moss after fire are a useful indicator of the magnitude and duration of fire effects and the fate of post-fire available N.
B11D-0767
Synchrony of Pandemics, Fire Reduction, and Reforestation in the Tropical Americas With Atmospheric Carbon Dioxide Changes During European Conquest
A new reconstruction of the biomass burning history of the tropical Americas is consistent with expanding fire use by Mesoamerican and Amazonian agriculturalists from 2000 to 500 years BP and a subsequent period of fire reduction due to indigenous demographic collapse. Our reconstruction synthesizes published data from stratigraphic charcoal accumulation records from lake and bog sediments and soil charcoal records, including soil charcoal obtained from archeological Amazonian Dark Earth sites. The charcoal data provide fire histories from over 40 localities and enable reconstruction of the Late Holocene regional biomass burning history of the tropical Americas. Synthesis of the stratigraphic charcoal records yields indexes of 1) the inter-site variability in charcoal accumulation; and 2) the mean rate of regional charcoal accumulation during 500-year increments since 3500 years BP. The age distribution of dated soil charcoal particles from non-archeological sites provides an independent measure of variation in regional charcoal accumulation; whereas the age distribution of soil charcoal dates from archeological sites records variation in charcoal accumulation related to anthropogenic biomass burning. The charcoal accumulation indexes derived from stratigraphic records begin to increase at ~2000 years BP, obtain maxima during the 500-year period just prior to European arrival, then decline to near-minimum values during the 500-year period subsequent to contact. Similarly, the age distribution of soil charcoal dated from non-archeological and archeological sites both indicate increases in charcoal accumulation from 2000 to 500 years BP followed by decline. We interpret the covariation between measures of charcoal accumulation derived from archeological and non-archeological sites as a consequence of the expansive influence of anthropogenic activity on the regional fire regime. The increase in regional charcoal accumulation apparent in the stratigraphic and soil charcoal records beginning at 2000 years BP correlates with expanding indigenous population, agriculture, and fire use in the tropical Americas. The rise in inter-site variability in charcoal accumulation after 2000 years BP is consistent with a demographic shift toward sedentary agrarian communities and localized increases in charcoal accumulation in densely populated centers. The declines in regional charcoal accumulation and inter-site variability after 500 years BP suggest a correlative cause related to reduction in anthropogenic biomass burning resulting from pandemic-driven population collapse. Published reconstructions of Pre-Columbian demography indicate that during European conquest, pandemics killed ~90% of the indigenous American population (~60 million), estimated to represent ~20% of the 16th century global population. Our predictive calculations suggest that fire reduction in the tropical Americas indicated in the charcoal record is associated with massive forest regeneration on ~1 x 106 km2 of land and sequestration of >10 Gt C into the terrestrial biosphere, which contributed to the ~2% global reduction in atmospheric CO2 levels and the 0.1‰ increase in δ13C of atmospheric CO2 from 1500 to 1700 A.D. recorded in Antarctic ice cores and tropical sponges.
B11D-0768
Improvement of wild fire detection algorithm with observation from commercial flight
The MOD14 algorithm is well tuned for fire detection world wide, and plays a significant role to detect wild fire in many countries, for its improved sensitivity and operability. However, we still need to improve wild fire detection algorithms for less commission or omission errors. To improve it, we built a database system for the fire observation by pilots of passenger aircraft. Then, we examined fire detection status to identify the bottleneck of fire detection, including non-hotspot pixels around every observed fire. Based on this examination, we proposed an improved wild fire detection algorithm modified from MOD14 algorithm. This algorithm detects around 30% more hotspots with not more false alarms comparing to original MOD14 algorithm. This algorithm utilize statistical test based on confidence ellipsoid for simultaneous distribution of brightness temperatures. We continue the analysis to find essential combination of efficient statistical test to detect wild fire and we will show you the latest result of new algorithm based on fire observation in 2007.
B11D-0769
Geostatistical Modeling of Forest Fire Burn Severity
Connecting remotely sensed measures of burn severity (i.e. Differenced Normalized Burn Ratio [DNBR]) with fuel properties during a burn is important for biomass consumption estimation. Results from a step-wise geostatistical analysis designed to measure the relative influence of physiographic and climatic factors affecting forest fire burn severity are presented. Universal and co-kriging inverse methods were used to assess spatial covariance and generate DNBR predictions and error assessments. Inputs to the model include topography, annual direct incident radiation, fire weather (i.e. temperature, relative humidity), and fuel loading. Annual direct incident radiation and fire weather exhibited correlations with burn severity implying a link with fuel moisture. Inclusion of mechanistic fuel moisture models is suggested to supplement the proximate measures used.
B11D-0770
Spatial variability in organic layer characteristics following wildfire in interior Alaskan black spruce forests
We measured residual organic layer characteristics at 194 test sites following 13 wildfire events in 2002 - 2005 in interior Alaska to further understanding of changes in post-fire variability in different landscape positions. North- facing and toe-slope forest organic layers were the least variable following wildfires (mean coefficient of variation = 64%). Flatter toe-slope forests had the highest mean cover of unburned and singed moss and Sphagnum moss species (33% of all sites surveyed) remaining after fire, and unburned organic matter patches were almost always associated with Sphagnum moss. We suggest that patchiness of moss dominated hummocks, which retain moisture effectively, largely determined variability in organic layer consumption in flat toe-slope forests. Across all sites, variation in post-fire organic layer depths increased as the fire season progressed, likely owing to the continued drying of surface fuels. Post-fire variability in organic layer depths was highest in wildfires occurring in the drier more severe fire years (2004-2005 vs. 2002-2003). These data suggest that variability in post-fire organic layer characteristics is not only mediated by biophysical properties at local scales, but is also largely influenced by climate variations within a fire season. We can therefore expect variability in post-fire organic layer depths to increase with increases in the duration of the fire season or with continued drying trends in a changing climate, especially in forests lacking significant cover of highly moisture-retentive Sphagnum mosses.
B11D-0771
Mercury Accumulation in the Forest Floor of the North Central United States
Atmospherically-deposited Hg has a strong affinity for soil organic matter. Fluxes of Hg in soil water of upland watersheds are generally small, but Hg stored in soil organic matter may be released when the forest floor is consumed by fire. The contribution of Hg from forest fires relative to other anthropogenic sources is an important unknown. The Forest Service, US Department of Agriculture, Forest Inventory and Analysis (FIA) program collects soil samples from forested areas across the United States as part of its Phase 3 sampling, and annual soils inventories are underway or completed in 45 of the 50 states (Alaska, Hawaii, Mississippi, New Mexico, and Oklahoma have yet to be sampled). Our objective here is to inventory and model the spatial distribution of forest floor Hg for a transect running across the north central United States. The collection of forest floor samples was accomplished as part of the standard FIA Phase 3 Soil Quality Indicator program. Field protocols include the measurement of the thickness of the forest floor and the collection of the entire forest floor found within a sampling frame having a diameter of 30 cm. We removed approximately 0.1 g of the sample for plots in our region of interest, and these were sent for Hg analysis by cold-vapor atomic absorption. Observations of mercury concentrations were joined with the Forest Inventory and Analysis Database to assign basic location information and associated inventory data. Mean plot-level values of Hg storage were tested against ecoprovince, forest type group, latitude, and longitude using analysis of variance and geographically weighted regression. Ecoprovince and forest type group were both significant predictors of mercury storage; conifer species tend to store more mercury than hardwood species. Different forest type groups store different amounts of forest floor Hg and in widely dispersed locations.
B11D-0772
Soil respiration from a boreal forest fire scar chronosequence
Climate change predictions suggest that warming is to be most pronounced at high latitudes, with a possibility of boreal forest warming of 4-6°C in the next 50-100 years. This has lead to the suggestion that changes in boreal forest soil carbon (C) storage could significantly alter the global soil C balance. Fire is the most significant factor controlling succession in the boreal forest biome and it is possible that climate change will lead to an increase in fire regime e.g. size, frequency, intensity, or any combination of these. Our research is investigating C flux dynamics in a Canadian boreal forest jack pine (Pinus banksiana Lamb.) dominated fire scar chronosequence. Fieldwork is carried out at Sharpsand Creek experimental burn site, near Thessalon, Ontario, Canada, where there are numerous fire scars of different ages. In June 2006 soil respiration (Rs), soil temperature (Ts) and soil moisture (Ms) were measured on three replicate scars selected from plot last burnt in 1948 and 1991. Rs values were later adjusted for Ts using a Q10 value of 2. There was no significant difference in mean adjusted Rs between the two scar age categories. It is likely that sample sizes were not large enough here to detect significant differences. In May 2007, large areas of the field site burnt as a result of wildfire; this provided an opportunity to take a large number of Rs measurements from recently burnt fire scars, as well as from those areas unaffected by the burn. Rs, Ts and Ms measurements were taken from 1948, 1975 and 1991 scar age categories (three replicate scars each) that were burnt in 2007, as well as a 1948 and 1991 fire scar that was unaffected by the wildfire. Soil samples were taken from three locations per fire scar surveyed and analysed in the laboratory for total C content. There was a significant difference in mean Rs adjusted for Ts and Ms (Rsadj) between the three pre-2007 fire scar age categories 1948, 1975 and 1991, that were all burnt in 2007. Mean Rsadj differed significantly between the 1948 and 1991 scars that were not burnt in 2007. A significant difference was apparent in mean Rsadj between the 1948 scar not burnt in 2007 and the three 1948 scars that were burnt in 2007. There was a significant difference in mean Rsadj between the 1991 scar not burnt in 2007 and the three 1991 scars burnt in 2007. Finally mean (Rsadj) differed significantly between the 1948 scar not burned in 2007, 1991 scar not burnt in 2007 and the 9 scars burnt in 2007. Our results indicate that Rs rates vary between fire scars of different ages and it seems that this is also apparent days after burning of scars that differ in their burn history. It appears that Rs from fire scars previously burnt in 1948 and subjected to fire in 2007, decreased as a direct result of the fire. In contrast, Rs from fire scars previously burnt in 1991 and subjected to fire in 2007, increased as a result of the fire. These results suggest that Rs is dominated by autotrophic components in mature forest (59 years post fire), but heterotrophic components in younger forest (16 years post fire). The possibility that Rs is increased directly after burning of younger fire scars is an important finding considering the view that climate change in boreal systems may increase the frequency of forest fires, and hence the proportion of younger forest. More research is needed in other boreal systems and in monitoring Rs for longer time periods after fire on scars with different burn histories.
B11D-0773
Long-Term Influence of Wildfire on Greenhouse Gas Fluxes in Southwestern United States Ponderosa Pine Forest Soils
The southwestern United States is home to the largest contiguous ponderosa pine forest in the world. Climate change and historical fire suppression have placed this forest at risk of stand-replacing wildfire. Ponderosa pine forest soils emit carbon dioxide (CO2) and consume methane (CH4) by biological oxidation. This study addresses how fire effects the source and sink strength of ponderosa pine forest soil. Fluxes of CO2 and CH4 were measured simultaneously in a dense unburned forest and a forest that experienced a high- severity fire 10 years previously using a static chamber technique and gas chromatography. We hypothesized that wildfire would reduce CO2 efflux due to a reduction in respiring belowground plant biomass and would increase CH4 consumption due to increased substrate diffusion into the soil. Our results supported these hypotheses. Additionally, CO2 efflux responds differently to changes in soil water content and soil temperature at the two sites, and models of the two environmental drivers produce significantly different results between sites. Ten years after fire, soil CO2 efflux is still depressed when compared to a dense forest; however, a reduction in net ecosystem production causes the burned site to be a net source to the atmosphere. The greater CH4 sink at the burned site does not offset the annual source of CO2 to the atmosphere. Wildfire in Southwestern ponderosa pine forests has lasting impacts on soil gas fluxes.
B11D-0774
Quantifying the Relative Importance and Potential Interactive Effects of Multiple Indices When Predicting Fire Risk and Severity in the Western US.
The national fire plan was implemented after the landmark fire season of 2000 as a response to a perceived increasing threat of severe wildfires. Subsequently, the Landfire project was initiated to develop a national dataset comprising vegetation condition, wildland fuels and fire regimes, and ecosystem status to support the national fire plan. A key product in this dataset to predict areas at risk for severe fires is the fire regime condition class (FRCC). The FRCC is an index of the degree of departure from the historical fire regime. This departure is a metric of the difference between current landscape vegetation composition and the range of historical reference vegetation characteristics; this difference can result from changes in vegetation characteristics and/or the spatial fire regime. The FRCC index is derived relative to simulated reference conditions, which in turn are based on Landsum, a landscape fire succession model. In addition to land management decisions, factors such as nitrogen deposition, ozone and climate affect both reference and current vegetation characteristics. It is an open question as to whether FRCC is sensitive enough to capture the full suite of potential effects on fire regimes. We are interested in examining vegetation change (via FRCC), nitrogen deposition, ozone concentration and climate variability in terms of their utility in predicting spatial variability in fire regime characteristics. Our analysis includes statistical examination of the multiple effects of nitrogen deposition, ozone, climate indices, and FRCC on fire frequency, size, and severity in California and the Western United States. We will assess how these four factors might act alone to influence fire, and their relative importance as co-determinants of fire risks. Our results will quantify how distinct FRCC is from climate and its efficacy as a predictor of fire risk and severity. Of particular interest is the extent to which FRCC predicts spatial variability in fire activity, and the extent to which climate and pollution patterns predict or influence the FRCC.
B11D-0775
The influence of rainfall, vegetation, elephants and people on fire frequency of miombo woodlands, northern Mozambique
Miombo woodlands are important in southern Africa as they occupy over 50% of the land and, their good and services support a large proportion of people in the region. Anthropogenic fires occur in miombo every year especially in the dry season (May – October). This study explores the influence of annual rainfall, elephant density, human density and corridors, and vegetation on the fire frequency. It was carried out in Niassa Reserve located in northern Mozambique, the largest and more pristine conservation area of miombo woodlands in the world. We used a time series analysis and statistical t-test of MODIS-derived Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI) to explore the relationship between biomass and fire frequency. The influence of rainfall, elephants, people and vegetation on fire return was explored using a stepwise logistic regression analysis. The results of this study indicate that fire frequency is higher in places with high biomass at beginning of the dry season. In these areas fire seems to be more intense and to strongly reduce biomass in the late dry season. Land cover is the strongest predictor of fire frequency, but elephant density, annual rainfall and human corridors are also important.
B11D-0776
Quantifying Biomass and Bare Earth Changes from the Hayman Fire Using Multi-temporal Lidar
Small-footprint multiple-return lidar data collected in the Cheesman Lake property prior to the 2002 Hayman fire in Colorado provided an excellent opportunity to evaluate Lidar as a tool to predict and analyze fire effects on both soil erosion and overstory structure. Re-measuring this area and applying change detection techniques allowed for analyses at a high level of detail. Our primary objectives focused on the use of change detection techniques using multi-temporal lidar data to: (1) evaluate the effectiveness of change detection to identify and quantify areas of erosion or deposition caused by post-fire rain events and rehab activities; (2) identify and quantify areas of biomass loss or forest structure change due to the Hayman fire; and (3) examine effects of pre-fire fuels and vegetation structure derived from lidar data on patterns of burn severity. While we were successful in identifying areas where changes occurred, the original error bounds on the variation in actual elevations made it difficult, if not misleading to quantify volumes of material changed on a per pixel basis. In order to minimize these variations in the two datasets, we investigated several correction and co-registration methodologies. The lessons learned from this project highlight the need for a high level of flight planning and understanding of errors in a lidar dataset in order to correctly estimate and report quantities of vertical change. Directly measuring vertical change using only lidar without ancillary information can provide errors that could make quantifications confusing, especially in areas with steep slopes.
B11D-0777
Temporal Trends in Post-fire Regeneration Patterns of Boreal Forests Using 1 km AVHRR NDVI
Forest fires are widely recognized as a major factor in global carbon budgets and are particularly important in boreal forests, where 37% of all carbon in the terrestrial biosphere is stored. Effects of climate change are evident in this region and will likely affect fire cycles, size, and severity. However, information is still relatively unavailable about the effects of burn characteristics on post-fire regeneration, and how burned areas differ from unburned areas in their responses to climate variation. Here, we analyzed the spatial variation of the NDVI temporal trends in post-fire regeneration as a result of burn severity, ecozone, land cover and latitude using a 1984-2006 1km AVHRR-NDVI time series for central Canada. The temporal trends were calculated using the seasonal Mann-Kendall trend test, a rank-based non-parametric test robust against seasonality, non-normality, missing values, and serial dependence. Burn severity was categorized in 4 groups from high to low: late-growing season fires (post July 20), early-season fires (pre July 20) in years of large fires, early-season fires in years of small fires and unburned areas. Negative temporal trends occurred in less than 4% of the sampled pixels. Increases of NDVI tended to be significantly lower at higher latitudes. Differences in NDVI were not significant across landcover types. Boreal Plains was the ecozone showing the highest positive trends in the NDVI; Boreal and Taiga Shields experienced moderate increases, while Taiga Plains showed minor increases. Our results show that the highest NDVI increases occurred in areas experiencing very severe fires during the late-growing season and in areas with non- severe, early-season fires. Unburned areas showed moderate NDVI increases.
B11D-0778
A Calibration of Temperature and Fire Frequency
A predictive model of fire frequency in North America using climate and human population density shows promise in estimating the importance of fire at broad temporal and spatial scales. Over 5 thousand fire scars from 120 sites and nine forest ecosystems were used to empirically derive and test a fire interval regression model. Three predictor variables were selected: a proxy of annual mean maximum temperature, annual precipitation, and human population density. The model was calibrated using mean fire intervals that document the presence of fire in a 1 to 3 km2 area during the two centuries prior to Euro-American settlement. This period allows for a more accurate calibration of temperature and fire frequency because of the reduced effects of land use, fire suppression, and other technological factors on fire events. Fifty five percent of the variance in mean fire intervals was explained by annual mean maximum temperature, 10 percent by annual precipitation, and an additional 11 percent by human population density (model r-square = 0.76). Although coarse (256 km2 cells), mean fire interval estimates provide a empirically derived and plausible depiction of the continental variability in historic U.S. fire frequency. Based on the regression diagnostics and spatial patterning in fire frequencies it is apparent that temperature is an important factor to U.S. fire regimes. We discuss the role of temperature as a 'master' variable for understanding multiple fire regime characteristics such as the rate of fuel combustion (the Arrhenius equation), the length of the fire season, and the broad scale variability of fire events. Modeling limitations, potential, and fire severity will be discussed. http://www.missouri.edu/~guyetter
B11D-0779
Applying Spatial Statistics to Isolate the Effects of Fuels, Topography, and Weather on Burn Severity
Fire severity datasets derived from satellite remote sensing data are now being used extensively in wildfire research and land management. Maps of burn severity based on the differenced normalized burn ratio (dNBR) are being produced and disseminated by the Monitoring Trends in Burn Severity (MTBS) project for all major wildfires in the United States from 1984 to present. This abundance of data presents unprecedented new opportunities for understanding how weather, terrain, and fuels interact to determine fire severity patterns, and for testing the effectiveness of fuel-reduction strategies for mitigating wildfire impacts. However, these datasets present challenges for statistical analysis because of their large sizes and the non-independence of spatially autocorrelated pixels. To explore the importance of spatial autocorrelation, we analyzed the spatial patterns of burn severity in two recent wildfires - the 2004 School Fire in the Blue Mountains of southeastern Washington and the 2005 Warm Fire on the Kaibab Plateau in northern Arizona. Conditional autoregressive (CAR) models were fitted with dNBR as the dependent variable and topography, fuels, and locations of recent fuel treatments as the independent variables. In both fires, elevation, slope, and aspect had strong effects on burn severity. Fuels had stronger effects on burn severity for the School fire than for the Warm Fire. In both fires, fuel treatments that combined thinning and prescribed burning resulted in statistically significant reductions in fire severity. The CAR models were then decomposed to isolate the spatial signal, which reflected spatially structured variability in dNBR that was not related to the independent variables. The spatial signals were correlated with the burn progression maps, reflecting spatial and temporal variability in weather and fire behavior (e.g. wind versus plume driven) over the course of the fire. These results suggest that spatial autocorrelation in the analysis of remotely- sensed burn severity datasets is not simply a nuisance, but in fact captures substantive and interpretable effects of weather and fire behavior on burn severity.
B11D-0780
Landscape estimates of heat release from prescribed fires: analysis and calibration of infrared imagery from aircraft
We used infrared images acquired from aircraft and ground-based calibration to produce sequential maps of fire radiative power (FRP, kW m-2) over the period of active flaming during prescribed fires. Ground sensors were calibrated for total radiant heat flux. Because the residence time of flaming combustion is short relative to the aircraft return time (4-5 min), integrating FRP over time to estimate fire radiative energy (FRE, kJ m-2) is not straightforward. Integration can be done for pixels where active flaming fronts can be identified and if the typical time course of FRP can be estimated independently. Heat flux data from ground sensors within the fires and from experimental fires were used to describe the time course of heat release. Experimental data and the literature were used to estimate the proportionality between energy release and total and rate of fuel consumption. Image analysis showed considerable variability among prescribed fires in spatial pattern and magnitude of heat release owing to fire weather and ignition method. http://www.cis.rit.edu/content/view/205/49/
B11D-0781
Effects of Pre-Fire Fuels Treatments on Post-Fire Burn Severity on the 2007 Fires in the Northern Rocky Mountains, USA
Climate change may be contributing to regional warming and drying trends that are increasing the size and severity of wildfires. Regardless if climate is a factor, the escalating costs of fire suppression and post-fire rehabilitation on the many large fires of recent decades have driven a national effort to reduce hazardous fuels across large areas, particularly those in the wildland-urban interface (WUI). Nationally, concern is especially focused on the numerous large wildfires currently burning in the Northern Rocky Mountains with a need for rapid science-based assessment of burn severity, even as fires and fire suppression efforts continue. Our objective is to assess if and how well various fuels reduction treatments applied pre-fire mitigated burn severity measured in the field immediately post-fire. We will obtain data from the incident command teams, including fire weather, daily fire progression maps, and where strategic and tactical fire suppression measures were applied. Location and type of fuels treatment as well as data on local vegetation type, structure, and fuels will be obtained from local management agencies and national databases. We will pair our sampled field plots in treated and burned areas with those not treated and burned in similar stand and topographic conditions across three or more large forest fires. Our analysis is both quantitative and qualitative, and linked with efforts to assess fuel treatment effects on fire behavior and ease of fire suppression. We report specifically on whether various fuels treatments are mitigating fire effects on soil (e.g., char, percent exposed, infiltration rate, water repellency) and vegetation (e.g., scorch, tree mortality, understory abundance, recovery). We discuss which fuels treatments work and which do not work, and the extent to which fire weather and other factors beyond the control of fire managers may determine whether or not fuels treatments are effectively mitigating severe fire effects.
B11D-0782
Impacts of prescribed fire on ecosystem C and N cycles at Fort Benning Installation, Georgia
A critical challenge for the land managers at military installation is to maintain the ecological sustainability of natural resources while meeting the needs of military training. Prescribed ground fire as a land management practice has been used to remove the ground layer plants at Fort Benning for two purposes: to facilitate access for military training, and to maintain and restore fire-adapted longleaf pine communities that are critical habitat for the federally endangered red-cockaded woodpecker (Picoides borealis). Nevertheless, the impacts of prescribed fire on ecosystem processes and health are not well-understood and quantified at the plot to regional scales. Frequent fire may result in ecosystem nitrogen (N) deficiency due to repeated N loss through combustion, volatilization, and leaching, threatening ecosystem sustainability at Fort Benning. On the other hand, N loss may be offset by enhanced symbiotic N2 fixation since fire favors herbaceous legumes by scarifying legume seeds and stimulating germination. Quantifying the impacts of prescribed fire on ecosystem carbon (C) and N cycles is further complicated by interactions and feedbacks among burning, nitrogen inputs, other land use practices (e.g. tree thinning or clear-cutting), and soil properties. In this study, we used the Erosion-Deposition-Carbon Model (EDCM), a process-based biogeochemical model, to simulate C and N dynamic at Fort Benning under different combinations of fire frequency, fire intensity, nitrogen deposition, legume nitrogen input, forest harvesting, and soil sand content. Model simulations indicated that prescribed fire led to nitrogen losses from ecosystems at Fort Benning, especially with high intensity and high frequency fires. Forest harvesting further intensified ecosystem nitrogen limitation, leading to reduced biophysical potential of C sequestration. The adverse impacts of prescribed fire and forest harvesting on C and N cycles were much higher in more sandy soil than in less sandy soil. N inputs from nitrogen deposition and legume N fixation helped replenish N losses to some extent. However, N losses due to fire and harvesting were not balanced or exceeded under current atmospheric N deposition and legume N input rates, suggesting additional N input (e.g., fertilization) may be needed to maintain the sustainability of current ecosystem states and management practices at Fort Benning.