H43F-1681
Synthesis of Post-wildland Fire Sediment Yields in Different Rainfall Regimes in the western United States
A variety of methods have been used to measure post-fire erosion across the western United States. We grouped these measurements into four methods: (1) point measurements on hillslopes, (2) hillslope plot measurements, (3) suspended-sediment and bed load sampling in channels, and (4) channel volume measurements of erosion or deposition. We classified features identified as "incised tributaries" or "gullies" as channel erosion. Only measurements made within two years of a fire were included in the synthesis. The measurement methods were stratified into seven different rainfall regimes based on seasonal rainfall type and rainfall intensities within each seasonal type. Post-wildfire sediment yields varied over five orders of magnitude for the different rainfall regimes. While differences in sediment yields exist across different rainfall regimes, a major source of variance is a result of the different methods used to measure post-fire erosion. A median sediment yield value for each measurement method was calculated. The average median sediment yield based on the channel volume method of 190 T/ha was greater than the sediment yields based on point measurements on hillslopes (29 T/ha), on hillslope plot measurements (14 T/ha), and on channel suspended-sediment and bed load sampling (2.5 T/ha) methods. This suggests that the channels are more important than the hillslope as sources of available sediment after wildfire. The lack of correlation of sediment yields with topographic slope and soil erodibility suggests that sediment availability may be more important than slope or erodibility in determining the sediment yield after wildfire
H43F-1682
Post-Fire Sediment Flux of the Day Fire, California
Hillslope transport rates in steeplands often increase dramatically following wildfires. Hillslope material transported by post-fire ravel processes is the primary source for sediment entrained in hazardous debris flows, which tend to initiate during intense rainfall. We used sediment traps to assess post-fire hillslope transport rates in steeplands burned by the 2006 Day fire in the Transverse Ranges of southern California. We installed seven sediment traps flush with the ground surface on relatively planar hillslopes ranging in gradient from 0.3 to 0.76 within two geologic parent materials: Mesozoic plutonic rocks and Miocene conglomerate and sandstone. Traps were installed within 2 to 4 weeks of fire containment while ravel was on-going. In the first month following fire containment, we observed widespread dry ravel and grain flows of cohesionless granular material occurring in response to localized turbulent wind bursts and solar-driven thermal variations. We collected sediment at <1 month intervals between October 2006 and July 2007, air dried samples, and then manually sieved and weighed them for particle size determination. The sediment mass acquired from each trap emptying interval was converted to a sediment flux per unit contour width. A nearby anemometer provided estimates of wind direction and speed. Particle size distribution of sediment accumulated in the traps remained relatively uniform throughout the monitoring period with no apparent immediate spike in post-fire, fine-grained ashy material. Ravel from the Miocene sediment is largely gravelly sand, while that derived from Mesozoic plutonic rocks is a sandy gravel grus. Sediment flux from all seven traps ranges over 2 orders of magnitude from 0.002 to 0.5 m3m-1yr- 1, lower on gentle slopes and higher on steeper slopes, independent of parent material. Plots show an exponential relation between cumulative sediment flux and local hillslope gradient, consistent with a non-linear relation between flux and gradient found by others. The departure from a flux law that is linear with slope occurs at ~0.8. Four of seven sediment traps recorded the highest sediment fluxes per trap-emptying interval prior to the onset of rain, with 6 to 32% of their total mass for the entire monitoring period accumulating during dry conditions. Sediment accumulation at these four traps correlates closely with both average wind speed and time- integrated wind speed. The remaining three traps record their highest fluxes in response to the first post-fire storms, despite their minimal precipitation, representing only 11% of seasonal precipitation. For all the sediment traps, these relatively small early-season rainstorms generated between 20 and 55% of the total accumulated sediment mass. In contrast, the largest rainfall event, producing 50% of the total seasonal rainfall, generated only moderate sediment fluxes, and late-season rain and wind events, albeit larger, did not have commensurately high sediment fluxes. These observations indicate that large volumes of sediment were transported under dry conditions and in response to the first rain storms. As a result, many slopes >0.65-0.7 were stripped to bedrock during the course of monitoring. Soil pits downslope in valley bottoms indicate aggradation of 0.25 m with a maximum fill exceeding 0.5 m. The combination of observed bedrock emergence and decreases in sediment flux recorded by the traps through the high rain and wind events indicates that these hillslopes converted from transport-limited to supply-limited conditions within months following the fire. Much of the sediment transported off the steep hillslopes aggraded the low-order channel network, which provides ample material for potential entrainment in future debris flows or floods.
H43F-1683
High-Resolution Rainfall From Radar Reflectivity and Terrestrial Rain Gages for use in Estimating Debris-Flow Susceptibility in the Day Fire, California
Constraining the distribution of rainfall is essential to evaluating the post-fire mass-wasting response of steep soil-mantled landscapes. As part of a pilot early-warning project for flash floods and debris flows, NOAA deployed a portable truck-mounted Shared Mobile Atmospheric Research and Teaching Radar (SMART-R) to the 2006 Day fire in the Transverse Ranges of Southern California. In conjunction with a dense array of ground- based instruments, including 8 tipping-bucket rain gages located within an area of 170 km2, this C-band mobile Doppler radar provided 200-m grid cell estimates of precipitation data at fine temporal and spatial scales in burned steeplands at risk from hazardous flash floods and debris flows. To assess the utility of using this data in process models for flood and debris flow initiation, we converted grids of radar reflectivity to hourly time-steps of precipitation using an empirical relationship for convective storms, sampling the radar data at the locations of each rain gage as determined by GPS. The SMART-R was located 14 km from the farthest rain gage, but <10 km away from our intensive research area, where 5 gages are located within <1-2 km of each other. Analyses of the nine storms imaged by radar throughout the 2006/2007 winter produced similar cumulative rainfall totals between the gages and their SMART-R grid location over the entire season which correlate well on the high side, with gages recording the most precipitation agreeing to within 11% of the SMART-R. In contrast, on the low rainfall side, totals between the two recording systems are more variable, with a 62% variance between the minimums. In addition, at the scale of individual storms, a correlation between ground-based rainfall measurements and radar-based rainfall estimates is less evident, with storm totals between the gages and the SMART-R varying between 7 and 88%, a possible result of these being relatively small, fast-moving storms in an unusually dry winter. The SMART-R also recorded higher seasonal cumulative rainfall than the terrestrial gages, perhaps indicating that not all precipitation reached the ground. For one storm in particular, time-lapse photographs of the ground document snow. This could explain, in part, the discrepancy between storm-specific totals when the rain gages recorded significantly lower totals than the SMART-R. For example, during the storm where snow was observed, the SMART-R recorded a maximum of 66% higher rainfall than the maximum recorded by the gages. Unexpectedly, the highest elevation gage, located in a pre-fire coniferous vegetation community, consistently recorded the lowest precipitation, whereas gages in the lower elevation pre- fire chaparral community recorded the highest totals. The spatial locations of the maximum rainfall inferred by the SMART-R and the terrestrial gages are also offset by 1.6 km, with terrestrial values shifted easterly. The observation that the SMART-R images high rainfall intensities recorded by rain gages suggests that this technology has the ability to quantitatively estimate the spatial distribution over larger areas at a high resolution. Discrepancies on the storm scale, however, need to be investigated further, but we are optimistic that such high resolution data from the SMART-R and the terrestrial gages may lead to the effective application of a prototype debris-flow warning system where such processes put lives at risk.
H43F-1684
Impact of Fire on Streamflow in Southern California Watersheds
Post-fire streamflow dynamics in Southern California have primarily been studied using small watershed experiments. These studies have concluded that increases in streamflow are a consequence of an increase in soil hydrophobicity, along with a decrease in transpiration rates associated with less vegetation. Extrapolation of the results from these studies to large watersheds (>50 km2) has been limited because large watersheds may not burn completely and other processes may emerge at these scales. In this study, six paired watersheds were used to test the hypothesis that there is an increase in streamflow following fire in large California watersheds (54-632 km2). The percentage of area burned in these watersheds ranged from 23 to 100%. The effects of fires on streamflow were examined at annual, seasonal, and monthly time-steps for the five years following fire. In addition, this study attempted to address fundamental regression assumptions that are commonly ignored, and create uncertainty bounds for evaluating the changes in streamflow before and after fire. Results of this experiment indicate that differences in pre and post-fire streamflows, at all time scales and in all the test catchments, were generally within the 95% uncertainty bounds of the regression equation. It is uncertain whether the apparent lack of significant difference between the pre and post-fire streamflow reflects no actual change in streamflow or is a consequence of the errors and uncertainties in the streamflow data. Furthermore, persistent drought in the years following fire made it challenging to interpret differences in pre and post-fire flows using the paired watershed methodology. The effects of hydrophobicity on post-fire streamflow may have been reduced by a limited number of storm flow events during these drought years. Under these dry conditions, soil moisture was the dominant control over transpirational losses, minimizing the effects of a reduction in vegetation cover. These results indicate that the consequences of fires are likely to vary depending on the post-fire meteorological conditions. The study addresses the challenges of using non-experimental watersheds for paired watershed studies.
H43F-1685
Fire Effects on Runoff Generation and Sediment Yield from a Coarse-Textured Sagebrush- Dominated Landscape
Post-fire increases in runoff and sediment yield from sagebrush rangelands are commonly attributed to fire- induced soil water repellency and/or reduction in canopy and ground cover. Recent research has demonstrated the strength of soil water repellency and its influence on runoff and sediment generation on burned and unburned sagebrush landscapes may exhibit significant annual fluctuation. The transient nature of soil water repellency complicates assessment of fire effects on hydrologic processes and determination of post-fire hydrologic recovery. Rainfall simulation and concentrated flow (rill) methodologies were applied on a coarse-textured sagebrush site in the Reynolds Creek Experimental Watershed, southwestern Idaho, USA. Simulations were conducted immediately pre- and post-burn and in each of three years following fire to determine fire effects on runoff and erosion. The influences of soil water repellency and changes in ground cover on hydrology and erosion processes were assessed using water drop penetration test and point frame methodologies. Runoff doubled following burning at the large plot scale (32.5 m2) and on coppice microsites at the small plot scale (0.5 m2). Runoff from rill processes was 3 times greater on burned than unburned plots immediately post- fire. Fire-induced increases in runoff generation were insignificant 1 year post-fire. Sediment yield from burned large and concentrated flow plots was 10 and 4 times greater respectively than on unburned plots immediately following and 1 year post-fire. Sediment yield on burned plots at the small plot scale was 70% greater immediately post-fire, but was greatly reduced 1 year following fire. Soils on burned and unburned hillslopes were strongly water repellent the year of the fire. The strength of water repellency was reduced by 50% 1 year post-fire and soils were slightly water repellent on burned large plots 2 years post-fire. The fire reduced litter cover by 50% and bare ground was increased from 20% to over 70%. Litter cover and percent bare ground on burned plots returned to near pre-burn levels 3 years post-fire. The greatest impact of burning was on sediment yield from rill processes. The combined effect of strong background water repellency and the increase in bare ground parentage allowed overland flow to concentrate into rills where rill flow width decreased and flow depth and velocity increased. Fire impacts on rill processes were persistent 2 years post-fire. The results indicate soil water repellency facilitates runoff on burned and unburned hillslopes, but the presence of ground cover mitigates sediment entrainment. These data suggest runoff and erosion from burned coarse-textured sagebrush hillslopes may require two to three years to return to background levels with or without the presence of strongly water repellent soil conditions.
H43F-1686
Erosion Rates of Volcanic-ash Derived Soils in the Blue Mountains of Eastern Oregon, USA: A Comparison Across Sales in Space and Time.
We examined present day rates of erosion in the Blue Mountains of eastern Oregon to quantify background erosion rates to provide standards for assessing possible accelerated rates of erosion resulting from wild fire or from land-management activities such as prescribed fire. The Skookum Creek watersheds, where stream discharge and sediment yield have been recorded continuously since the watersheds were gauged in 1992, provided a watershed-scale estimate of erosion rates. We installed hillslope erosion plots on north- and south- facing slopes within the watersheds in 2002 and collected data for three years to estimate short-term, hillslope- scale erosion rates. We also collected soil samples and analyzed them for 137Cs to get a 40-yr time- integrated estimate of hillslope erosion rates. Our results showed large differences between whole-watershed sediment yields and hillslope erosion rates measured from plots, suggesting that episodic processes dominated sediment production and transport and therefore controlled watershed-scale sediment budgets. At the hillslope-scale, short-term erosion resulted primarily from digging by small mammals and trampling by elk. Visual observations at the plots suggested that annual down-slope sediment movement was usually less than one meter. There were no significant difference among slope positions, but erosion rates were significantly higher on south-facing aspects and positively correlated to the amount of bare ground. In contrast, the 137Cs data suggested that erosion rates differed with slope position. Higher erosion rates were measured in toe- and mid-slope positions, with little erosion occurring on upper slopes and ridge tops. We examine these results in light of the present-day pattern of surface soils resulting from redistribution of volcanic ash from upper- slope to lower-slope positions and the effects of disturbance, including wildfire and the preferential grazing of riparian and lower-slope positions by domestic livestock.
H43F-1687
Soil Erosion in a Burned Mountainous Watershed, Kootenay National Park, British Columbia
In late July 2003, lightning ignited wildfires in Kootenay National Park, southeastern British Columbia. Following the wildfires, a study was initiated to investigate post-fire soil erosion in Hawk Creek watershed, located with the National Park. Hillslope gradients are moderate (generally less than 30 degrees) in the lower third of the basin and become steeper with increasing elevation. Summer rainfall is delivered by convectional thunderstorms, and average annual rainfall is about 340 mm. The mean fire return interval in this region is about 165 years and fire frequency has changed over time in response to climate. Silt fences were installed to measure soil erosion in the summers of 2004 and 2005. Results indicate a threshold rainfall intensity of 1.6 mm in 15 minutes is required to mobilize sediment and based on this value, a rain day herein refers to days when the maximum 15-minute rainfall intensity exceeds 1.5 mm. Sediment transport data show many rain days producing no sediment, while relatively small amounts of soil erosion occurred during several rain events in 2004 and 2005. An important finding is that post-fire soil erosion rates are very small, and considerably lower than many other results reported in the literature. The sandy soils exhibited no signs of soil hydrophobicity, but rather displayed high infiltration capacities that precluded the development of overland flow. No evidence of hydrophobic soils or rilling was evident in the hillslope plots during sediment collection after the rain events and it is probable that the majority of soil erosion occurred by rainsplash. The increased soil erosion measured on steeper plots would be in accordance with the increased rates in downslope rainsplash erosion expected on higher gradient slopes. Grain sizes trapped by the silt fence relative to the soil profile are based on soil samples collected from silt fences and nearby hillslopes. Most of the sediment collected in the silt fence traps fell in the size range less than 2 mm, indicating stresses associated with rainsplash were generally not sufficient to entrain larger particles.The less than 2 mm fractions for silt fence and hillslope samples were extracted for comparison, with the hillslope samples containing 45-58 percent of particles less than 354 microns, and silt fence percentages ranging from 72-90 percent. The higher percentage of fines in the silt fence samples suggests that fines less than 354 microns are being selectively entrained, transported and deposited in the silt fence. Selective transport of particles less than 0.5 mm occurred, with minimal mobilization of coarse sand. Soil erosion data collected herein provides the necessary data to derive gradient-driven transport equations, which can be combined with a stochastic algorithm for wildfire return interval and precipitation events (such as storm duration, interstorm duration, and average storm rainfall intensity) to explore implications of rainsplash erosion on burned hillslopes in this region over longer time scales. A brief window of opportunity exists after a fire for effective rainsplash or overland flow transport, and high energy storms are required in these same post-fire years to enable such an occurrence.
H43F-1688
Comparing Burned and Unburned Forest Conditions Using Simulated Rill Experiments
Despite the dominance of concentrated flow or rill erosion in the erosion processes in disturbed forests, few studies have quantified the effects of different types of forest disturbance on rill erosion. This study quantified the effects of four forest conditions—natural (recently undisturbed), low soil burn severity, high soil burn severity, and skid trails—on rill runoff quantity, runoff velocity, and rill erosion. Simulated rill experiments were conducted at sites in eastern Oregon (Tower Fire) and in northern Washington (North 25 Fire) on forested slopes with granitic and volcanic soils, respectively. The natural and skid trail conditions were established near each burned area in unburned forest. For each rill experiment, concentrated flow was applied at the top of the plot through an energy dissipater at five inflow rates for 12 min each. Runoff was sampled every 2 min and runoff volume and sediment concentration were determined for each sample. The runoff velocity was measured using a dyed calcium chloride solution and two conductivity probes placed a known distance apart. Runoff volume, runoff velocities, and sediment concentrations increased with increasing levels of disturbance. The natural plots had very low runoff rates and sediment concentrations at both the Tower and North 25 sites. The low soil burn severity plots had greater responses than the natural plots, but the responses in the two sites were different as a result of variability in effect of burning and differences in time between burning and the rill experiments. The high soil burn severity and the skid trail plots had the highest runoff ratios, runoff velocities, and sediment concentrations and the responses were similar at both sites. These results suggest that any differences in responses related to soil type or other site factors were masked by the increase in response resulting from the high levels of disturbance.
H43F-1689
Effects of Boreal Forest Fires in Interior Alaska
The summer of 2004 in Interior Alaska was characterized by enormous and devastating boreal forest fires. In addition, the effects of fires on suspended load in streams, and channel development in zones affected by fires or fire suppression activities are not well understood or quantified in areas underlain by discontinuous permafrost (i.e., perennially frozen soils). To improve our knowledge in these areas, small streams draining water from areas affected by fires in different proportions (i.e.: unburned, partially, and severally burned) were systematically sampled during the summers of 2005, 2006, and 2007. All the streams were located in watersheds underlain by discontinuous permafrost in Interior Alaska. Autosamplers were deployed in the streams after spring breakup to collect daily water samples. Pressure transducers and dataloggers in conjunction with velocity measurements were used to estimate water discharge in the streams. Human influence is negligible in the study areas, with the exception of modifications caused by fire suppression activities. Thus, data collected from these areas can be considered as a natural system response to forest fires. Field observations in areas affected by fire suppression activities indicate active processes of erosion and sedimentation, in terms of a) channel formation, b) channel widening, and c) alluvial fan formation and evolution. However, vegetation is recovering in other areas. A comparison between suspended sediment concentrations collected in 2005, 2006, and 2007 will be presented at the meeting.
H43F-1690
Overlandflow and Rainsplash Erosion Rates of Scoria Cone Hillslopes Affected and Unaffected by Wildlandfire
Data on linear diffusion-type degradation by rainsplash and on non-linear, non-diffusion type degradation by overland flow were collected during summer monsoon season in Arizona from 1996 to 2000. Splashboards, erosion pins, rain gauges, GPS, and cross-sections were major instrumentation used in this field based study conducted on natural, undisturbed slope and rainfall conditions. Effective diffusion coefficients, erosion rates due to raindrop detachment and overlandflow estimated empirically were investigated with this data set. The rainsplash data are consistent with a long-term diffusion model of slope degradation. However, the data are in harmony with model results of more arid climate than with the semiarid climate of the San Francisco Volcanic Field. This indicates that previous diffusion-type modeling of scoria cones may have overestimated the contribution of quantity of rainsplash erosion and underestimated the contribution of overland flow processes to overall erosion. The short-term slope modification by overlandflow compares well with results from a cellular aautomaton, smooth particle hdrodynamics model (CASPH). Detailed measurements of horizontal and longitudianl profiles on the slopes give in depth insight to the processes shaping these landforms. A wildland fire damaged the vegetation (Ponderosa Pine) of one of the two hillslope observation sites in 1996. This event resulted in a great opportunity to observe the post-fire geomorphological response on rainsplash and overlandflow processes on a weathered scoria cone slope surface. The rate of degradation of the burned cone was orders of magnitude larger than of the unburned cone. Both diffusive and non-diffusive erosion measurements proofed that the first post-fire rainy season in areas affected by wildlandfires are severe.
H43F-1691
Fire Induced Shallow Landslides Triggering
Forest fires can influence the hydrologic response and shallow landslides susceptibility of upland catchments. Many factors, interacting one with each others, play a role in watershed response altered by forest fire forcing, such as fire intensity, vegetation cover, soil properties, soil moisture content, rainfall intensity and time interval between fire and the rainfall event. Fire can change soil properties, inducing the development of a water repellent layer at or near the soil surface, that reduces soil infiltration capacity, increases overland flow and accelerates surface erosion. These hydrological changes can increase shallow landslides susceptibility. In order to study the effect of forest fire on shallow landslides triggering, at first the factors that pertain to the triggering of shallow landslides from a basin recently burned have to be examined. Then, via hydro- geomorphologic model, different scenarios have to be studied for analyzing the role played by each factor. As study area we choose a basin located in Northern Piemonte (Italy) where on 18th July 2005 an intense rainfall event triggered a large shallow landslide. This rainfall event happened after a forest fire burned about 50% of the basin area. The role of forest fire in shallow landslide triggering was investigated by accurate field surveys and field data measurements concerning hydrological and geological soil properties. They were carried out both in the burned and unburned portion of the study area. The acquired measures have then been used to parameterized an hydro-geomorphic model aimed to study different shallow landslides triggering scenarios.
H43F-1692
Storm Rainfall Conditions for Floods and Debris Flows from Recently Burned Basins in Southwestern Colorado and Southern California
Debris flows generated during rainstorms on recently burned hillslopes have destroyed lives and property throughout the Western U.S. Field evidence indicates that, unlike debris flows that start as landslides, these events have no discrete initiation source and can occur with little or no antecedent moisture. Using rain gage data and information on basin runoff response from five burned areas in Colorado and southern California, we documented the rainfall conditions that triggered post-fire debris flows and developed empirical rainfall intensity- duration thresholds for the occurrence of debris flows and floods following wildfires in these areas. This information can provide guidance for warning systems and planning for emergency response under similar conditions. Debris flows were produced from recently burned basins in Colorado in response to convective storms. Debris flows were triggered after as little as six minutes of storm rainfall. Most of the storms that generated debris flows lasted less than one hour, ranged in average intensity between 1.0 and 32.0 mm/hr, and had recurrence intervals of two years or less. Threshold rainfall conditions for floods and debris flows sufficiently large to pose threats to life and property from recently burned areas in south-central Colorado are defined by I = 6.5D-0.7, and for southwestern Colorado by I = 9.5D-0.7, where I = rainfall intensity (in mm/hr) and D = duration (in hours). Debris flows were generated from recently burned areas in southern California in response to frontal storms. The flows occurred after as little as two hours, and up to 16 hours, of low intensity (1-20 mm/hr) rainfall. The storms had recurrence intervals of two years or less. Threshold rainfall conditions for life- and property- threatening floods and debris flows during the first winter season following fires in Ventura County are defined by I = 12.5D-0.4, and for the San Bernardino, San Gabriel, and San Jacinto Mountains as I = 7.2D-0.4, respectively. The thresholds defined here are lower than most identified for unburned settings, perhaps because of the difference between extremely rapid, runoff-dominated processes acting in burned areas and longer-term, infiltration-dominated processes on many unburned hillslopes.
H43F-1693
What Causes Runoff and Sediment Yields to Increase After Wildfires?
Runoff and sediment yields can increase by several orders or magnitude after high severity wildfires. These increases have been attributed to soil water repellency, loss of surface cover, and soil sealing by either mineral or ash particles, but the relative effects of these factors have rarely been isolated. The objectives of this study were to: 1) isolate the effects of burning by measuring soil water repellency, surface cover, and sediment yields from 21 hillslopes burned in high-severity wildfires, 13-34 unburned hillslopes, and 3 hillslopes where the surface cover was removed by raking; and 2) use rainfall simulations to determine whether surface sealing is more prevalent on bare soils or soils covered with varying amounts of ash. The field measurements were made over a five-year period in ponderosa pine forests in the Colorado Front Range. The burned hillslopes generally had stronger soil water repellency than the unburned hillslopes only for the first summer after burning, but the mean cumulative sediment yield from the burned hillslopes was 31 Mg ha-1 as compared to minimal sediment yields from the unburned hillslopes. The raked hillslopes had very similar sediment yields to the burned hillslopes when they had comparable surface cover, rainfall erosivity, and soil water repellency. The rainfall simulations on bare soil generated much more runoff and sediment than the simulations on ash-covered soil, and both bare soils developed a thin, structural soil seal. Runoff and sediment yields decreased as ash thickness increased, but successive simulations quickly eroded the ash cover and increased runoff rates to the levels observed for bare soil. The results indicate that: 1) post-fire sediment yields are primarily due to the loss of percent cover rather than fire-enhanced soil water repellency; 2) surface cover is important because it controls the extent of soil sealing; and 3) ash temporarily prevents soil sealing and reduces post-fire runoff and sediment yields. The results have important implications for forest management and mitigating post-fire erosion.
H43F-1694
An Experimental Study of the Effects of Litter and Duff Consumption and Ash Formation on Post-Fire Runoff.
Consumption of the litter and duff layers in forest wildfires can lead to substantial increases in the frequency and magnitude of overland flow. These increases result from the loss of storage in the organic surface layer, reduced surface roughness, and from sealing of the exposed mineral soil surface. The presence of an ash layer may accentuate surface sealing by providing an additional source of fine material, or it may reduce runoff by storing rainfall and by protecting the soil surface from raindrop impacts. We used simulated rainfall experiments to assess the effects of litter and duff consumption and the presence of ash layers of varying thickness on post fire runoff at two forested sites in western Montana, one with sandy loam soils formed out of granodiorite and the other with gravelly silt loam soils formed out of argillite. At each site we measured the runoff from simulated rainfall in replicated 0.5 m2 plots before and after application of the following treatments: 1) burning with a fuel load of 90 Mg ha-1, 2) manual removal of the litter and duff layers, 3) addition of 0.5, 2.5 and 5 cm of ash to plots from which the litter and duff had previously been removed, and 4) addition of the same depths of ash to burned plots at the sandy loam site. In the burned plots the surface litter and duff layers were completely consumed and a <1cm layer of black and gray ash and char was formed, indicating a moderate severity burn. The mean soil temperature in the upper 1 cm of the mineral soil was 70° C, and there was no detectable increase in water repellency. The mean final infiltration capacity of the burned sandy loam plots was 35 mm hr-1 compared to a pre-fire mean of 87 mm hr-1, while in the gravelly silt loam plots the pre- and post burn infiltration capacities (27 and 31 mm hr- 1) were not significantly different. Manual removal of the litter and duff layers reduced the mean final infiltration capacity in the sandy loam plots from 64 mm hr-1 to 40 mm hr-1 and in the gravelly silt loam plots from 23 mm hr-1 to 16 mm hr-1. We attribute decreases in infiltration due to the burning and duff removal treatments primarily to surface sealing. In the sandy loam plots, burning may have had a greater effect on infiltration than duff removal because the thin ash layer in the burned plots provided an additional source of fine material. In the gravelly silt loam plots, macropores located around rock fragments helped to minimize sealing effects. The addition of 0.5 cm of ash to the burned granitic plots resulted in a 20 mm hr-1 decrease in the final infiltration rate, and this was also probably due to surface sealing. However, the overall effect of ash addition was to increase the cumulative infiltration in proportion to the ash thickness and to maintain a higher average infiltration rate, indicating that while thin (<1 cm) ash layers may promote sealing, thicker ash layers help to reduce the runoff rate by providing additional storage for rainfall and by protecting the soil surface from raindrop impacts.
H43F-1695
Does wildfire ash block soil pores? A micromorphological analysis of burned soils.
Increases in runoff and erosion after forest wildfires are often attributed to the removal of surface cover, the formation of water repellent soils, and sealing of the soil surface by ash. The latter process involves clogging of pores by ash as well as rainsplash induced compaction of the ash layer. However, few studies have directly addressed the hydrologic role of ash and no studies have documented ash sealing in a forest fire environment. In an attempt to determine whether ash contributes to reduced infiltration after fire we conducted a micromorphological analysis of soils collected before and after three controlled pile burns at the Lubrecht Experimental Forest in western Montana. The burns were conducted with a fuel load of 90 Mg ha-1 on sites dominated by Lodgepole pine ( Pinus contorta) with scattered Douglas fir ( Pseudotoga menziesii), sandy loam soils and a mean of 99% ground cover (litter, duff and live vegetation). Soil cores were collected before burning, immediately after burning and after the burned areas had been subjected to simulated rainfall at an intensity of 80 mm hr-1 for 1 hour. The cores were impregnated with resin from which thin sections were made and microscopically analyzed to determine the vertical distribution of organic material, ash, mineral soil and porosity. Burning consumed all of the surface litter and duff and formed a <1cm layer of black and gray ash above the mineral soil, indicating a moderate severity burn. The mean soil temperature in the upper 1 cm of the mineral soil was 70° C, and there was no detectable increase in water repellency. Rainfall simulations conducted before and after the fires indicated that burning reduced the infiltration capacity from a pre-fire mean of 87 mm hr-1 to a post-fire mean of 35 mm hr-1. Prior to burning the upper 1 cm of the soil was comprised of 41% non- ash organic material, 4% clastic material and 55% pore space. After burning the porosity in the upper 1 cm decreased to 36% and the solid component consisted primarily of black and white ash (34% and 7% respectively). The biggest decrease in porosity was in the upper 2 mm of the soil where porosity decreased from 62 to 22% with a corresponding increase in the proportion of ash. Following the rainfall simulations the black ash content of the upper 1 cm decreased to 21%, suggesting that ash particles were removed in the runoff. However the remaining ash particles became noticeably more aligned parallel to the soil surface. Our observations indicate that ash may contribute to reduced infiltration after fire in two ways: 1) by filling pore space, and 2) by orienting parallel to the soil surface, so creating a thin water repellent organic layer in the upper few millimeters of the soil.
H43F-1696
Long term dynamic changes in soil organic matter following forest fires
Soil is an important component of the ecological system, providing a substrate for vegetation growth, habitat for animals and a site for regulation of eco-hydrological processes. After a disturbance, such as a forest fire, the soil undergoes changes that have not all been studied yet. Soil organic matter (OM) is one of the fundamental components of soils. After a fire, OM is destroyed and consequently, physical, chemical and biological properties related to OM are altered. Mt. Carmel, Israel, represents a typical Mediterranean ecosystem, with increasing number of forest fires. In the past decades, 9 large fires (> 100 hectare) and more than 350 smaller ones consumed thousand of hectares of natural and planted vegetation. The study is aimed at a) assessing long term rehabilitation processes of soil properties following forest fires; b) defining pedological indices of soil rehabilitation. Research methods included field and laboratory techniques - soil surveys and sampling in the field in addition to chemical analyses of soil properties. Soil survey and sampling were conducted in regions burnt during 1983, 1989, 1999, and 2005 and in non-burnt control areas. Laboratory analyses included: soil texture, organic matter content (OM), electrical conductivity (EC), phosphorus and other nutrients. Recovery rates of soil properties are related to fire severity and might be divided according to their rehabilitation rates (long or short-term processes). During the time scale in which the research was conducted it is apparent that following fires of all severities OM content remained significantly lower compared to the control values. Patterns of phosphorus levels were similar to the observed OM patterns, implying that these two factors are interconnected. Other components, however, such as pH and EC, which are also influenced by OM, exhibit different recovery patterns - pH values increased abruptly after the fire, and recovered within 1-7 years period (depending on fire severity), while the electric connectivity (EC) significantly increased after the fire and exhibited a prolonged decrease. According to our findings, we suggest that the primary impact of fires is mainly in A horizon (soil surface 0-10 cm). OM content plays a key role affecting various nutrients response to the fire and their recovery rates. Thus, in contrast to many studies demonstrating a rapid recovery of the soil system in Mediterranean ecosystems, our findings, based on a 23 years time span, indicates a long term rehabilitation of most soil compounds after forest fires.
H43F-1697
Post-fire Erosion: Long-term Recovery and Effectiveness of Rehabilitation Techniques
Millions of dollars are spent each year to reduce runoff and erosion rates after high-severity wildfires, but there are relatively few data evaluating their effectiveness over longer time periods relative to natural recovery. In this study we summarize six years of hillslope erosion data from 18 untreated plots and five replicated post-fire rehabilitation treatments applied after the 2002 Hayman and Schoonover wildfires in the ponderosa pine zone in the central Colorado Front Range. The rate of recovery and effectiveness of these treatments are compared to the results from the 2000 Bobcat fire 130 km to the north (Wagenbrenner et al., 2006). On the Hayman fire straw mulching reduced hillslope-scale sediment yields by more than 90% in the first two summers after burning and by 77% in the third summer after burning. By the fourth summer the straw mulch had no significant effect on the amount of ground cover or sediment yields relative to the untreated controls. Aerial hydromulching had a similar effectiveness and longevity, but a ground-based hydromulch treatment was poorly formulated and did not significantly reduce sediment yields. Scarification and seeding did not increase the amount of surface cover or reduce sediment yields. The initial wet application of a polyacylamide appeared to reduce sediment yields, but subsequent applications did not have any significant effect. None of the treatments was effective for more than three summers after burning. The results confirm the fundamental importance of ground cover and are consistent with the mulching and seeding results from the Bobcat fire. The primary difference is the slower vegetative recovery rates for the coarser-textured soils in the Hayman and Schoonover fires, and this has caused post-fire sediment yields to remain elevated for six summers after burning as compared to a maximum of four summers for the Bobcat fire. Land managers must evaluate the duration of treatment effectiveness and post-fire erosion as well as the shorter-term efficacy of their proposed treatments, and these evaluations must be done on a site-specific basis. http://www.warnercnr.colostate.edu/frws/people/faculty/macdonald/macdonald.html
H43F-1698
Predicting Fire Severity and Hydrogeomorphic Effects for Wildland Fire Decision Support
The Wildland Fire Decision Support System (WFDSS) uses the Fire Spread Probability (FSPro) model to predict the spatial extent of fire, and to assess values-at-risk within probable spread zones. This information is used to support Appropriate Management Response (AMR), which involves decision making regarding fire-fighter deployment, fire suppression requirements, and identification of areas where fire may be safely permitted to take its course. Current WFDSS assessments are generally limited to a binary prediction of whether or not a fire will reach a given location and an assessment of the infrastructure which may be damaged or destroyed by fire. However, an emerging challenge is to expand the capabilities of WFDSS so that it also estimates the probable fire severity, and hence the effect on soil, vegetation and on hydrologic and geomorphic processes such as runoff and soil erosion. We present a conceptual framework within which derivatives of predictive fire modelling are used to predict impacts upon vegetation and soil, from which fire severity and probable post-fire watershed response can be inferred, before a fire actually occurs. Fire severity predictions are validated using Burned Area Reflectance Classification imagery. Recent tests indicate that satellite derived BARC images are a simple and effective means to predict post-fire erosion response based on relative vegetation disturbance. A fire severity prediction which reasonably approximates a BARC image may therefore be used to assess post-fire erosion and flood potential before fire reaches an area. This information may provide a new avenue of reliable support for fire management decisions.