MYSTERIES OF ROCK SPALLING-- MORE QUESTIONS THAN ANSWERS

Blair, Rob, Fort Lewis College Professor Emeritus, 2409 Delwood Avenue, Durango, CO 81301, blair_r@fortlewis.edu

A METHOD FOR THE RAPID ASSESSMENT OF THE PROBABILITY AND VOLUME OF POST-WILDFIRE DEBRIS FLOWS FROM RECENTLY BURNED BASINS IN THE INTER-MOUNTAIN WEST, U.S.A.   KEYNOTE PRESENTATION

Cannon, S.H.; J.E. Gartner; M.G. Rupert; and J.A. Michael, U.S. Geological Survey, Box 25046, DFC, MS 966, Denver, CO 80225

The increased incidence of catastrophic wildfires in the western United States and the encroachment of human development into fire-prone ecosystems have created a critical need for methods to quantify potential hazards posed by debris flows produced from burned watersheds. Debris flows can be one of the most hazardous consequences of rainfall on recently burned hillslopes, and here we propose an approach to answer two of the fundamental questions in hazard assessments Ð where will events occur? and how big might they be? To identify those basins most prone to post-fire debris-flow activity, we used a series of logistic regression analyses on a database from 401 basins that were burned by 15 recent fires located throughout the intermountain west to develop a basin-scale model for debris flow probability. The model describes debris-flow probability as a function of readily-obtained measures of areal burned extent, soil properties, basin gradients, and rainfall from short-duration convective rainstorms. To determine how large a given event might be, a multiple-regression model was developed to define the range of volumes of material that can potentially be generated from recently-burned basins. The data used in the development of the model consists of measurements from 56 recently burned, and debris-flow producing, basins located throughout the western U.S. for which estimates of debris-flow volume had been obtained. The regression model describes debris flow volume at the basin outlet as a function of basin gradient, burned extent, material properties, and storm rainfall. Combining calculations of debris-flow probability and volume allows for a relative hazard ranking which identifies those basins that are most prone to the largest debris-flow events in response to convective rainstorms. This information is critical to effective post-fire mitigation and evacuation planning, and implementation of the approach in real time can potentially provide information for emergency warnings and evacuations.

A PROTOTYPE FLASH FLOOD AND DEBRIS FLOW EARLY WARNING SYSTEM FOR AREAS RECENTLY BURNED BY WILDFIRE IN SOUTHERN CALIFORNIA

Cannon, Susan H., U.S. Geological Survey, Landslide Hazards Team, cannon@usgs.gov; Pedro Restrepo, NOAA Office of Hydrologic Development; Jayme Laber, NOAA/NWS Los Angeles Weather Forecast Center; and Kevin Werner, NOAA Western Region Headquarters

Flash floods and debris flows are common following wildfires in southern California. On Christmas Day 2003, sixteen people were swept to their deaths by debris flows generated from basins burned the previous fall. In an effort to reduce loss of life by floods and debris flows, the National Oceanic and Atmospheric AdministrationÕs (NOAA) National Weather Service (NWS) and the United States Geological Survey (USGS) established a prototype flash flood and debris flow early warning system for recently burned areas in an eight-county area of southern California. This prototype project builds on the NWS Flash Flood Monitoring and Prediction (FFMP) system and USGS rainfall intensity-duration thresholds. These thresholds are defined for debris flows and flash floods that occur in response to the first few storms to impact a recently burned area, and following a year of vegetative recovery. The FFMP was modified to identify when both flash floods and debris flows are likely to occur based on comparisons between precipitation (including radar estimates, in situ measurements, and short-term forecasts) and the rainfall intensity-duration thresholds developed specifically for burned areas. The FFMP provides a cost-effective and efficient approach to implement a warning system on a 24-hour, 7-day-a-week basis. Advisory outlooks, watches, and warnings are disseminated to emergency management personnel through NOAAÕs Advanced Weather Information Processing System (AWIPS). In addition to operating the prototype system, an area within the southern California study area is dedicated to intense instrumentation and research to develop new geologic, hydrologic, and hydrometeorologic methods for precipitation and debris-flow forecasting, measurement, and analysis techniques. Although the potential exists for enhancing and expanding the warning system to provide spatially and temporally explicit information on debris-flow hazards, significant financial resources and scientific advancements are necessary to realize this potential.

OKANAGAN MOUNTAIN PARK FIRE, 2003, KELOWNA, BRITISH COLUMBIA (THE WORST INTERFACE WILDFIRE IN CANADIAN HISTORY)

Dobson, Don, Dobson Engineering Ltd., #4-1960 Springfield Road, Kelowna, British Columbia, Canada V1Y 5V7, ddobson@dobsoneng.com

In 2003 the Okanagan Mountain Park Fire became the most destructive interface wildfire in Canadian history. Before it was controlled it had razed nearly 26,000 hectares of forest and burned through the southern neighbourhoods of the City of Kelowna consuming 239 homes with estimated losses exceeding $100 million. Fortunately not one human life was lost in the fire.

Immediately following the fire the city initiated a risk analysis to determine the safety of its residents, their property and public infrastructure from the threat of flood and debris flow. The methodology used in the analysis was based on risk management principles adopted in Canada and elsewhere. The study identified and assessed the risks, determined the probability of events occurring and the probable impacts, developed a response to the risks, evaluated the options and finally recommended appropriate options to implement.

It was determined that the greatest risk was from flooding due to the changed conditions in the watersheds, i.e. the loss of forest cover and the damage to soils. The fire partially or completed destroyed the forests in approximately 15,000 hectares of nine major watersheds that flow through the city. The flood risk was confirmed in October 2003 when an intense rainstorm caused flash floods in several of the burned watersheds. Peak flows during the storm, based on actual field data, were calculated at five to 15 times the 200-year pre-fire discharges. Using the impacts from the October storm it is estimated that there are at least 50 homes, 150 residents and 10 roadways at risk from potential flash floods over the next few years.

To mitigate the risk to public safety, property and infrastructure the city proposed to take pre-emptive action in those watersheds presenting the greatest risks. The criteria for the risk assessment was initially based on estimated 200-year flood flows, which is the accepted standard used by the city for normal flood design as well as by both the provincial and federal governments. It was determined however, that there were actually three different flood scenarios that should be considered. First, the 200-year pre-fire flood based on ÒnaturalÓ conditions in the watersheds. Second, the 200-year post-fire flood based on water yields from the high severity burned areas. The third scenario involved the post-fire flood due to rain on hydrophobic soil (water-repellent soil). The significance of this last scenario was clearly illustrated by the October 22/23, 2003 rain event at Kelowna. The design criteria approved by the city was based on runoff generated by the 25-year rainstorm occurring on water-repellent soils. This event has a probability of occurrence of 10% in the next three years (the estimated persistence period for the hydrophobic soil condition).

It was the position of the City of Kelowna that mitigation forms a cornerstone of sound risk and emergency management. The City endorsed cost-effective efforts that reduced the impact of disasters on critical infrastructure, personal property, and the lives of its citizens. Through a thorough understanding of community risks and vulnerabilities, careful assessment of viable options, and a strong commitment to partnerships, the City undertook nearly $3 million of stream crossing upgrading to protect and enhance public safety.

MAPPING LANDSLIDE THRESHOLDS, USING LIDAR IN THE WEST HILLS OF PORTLAND, OREGON

Drazba, Marina, Portland State University, 1429 SW 14th Avenue, Apt. 406, Portland, OR 97201, mcdrazba@pdx.edu

Mapping landslides in western Oregon is problematic due to the heavy vegetation. The West Hills of Portland, Oregon are prone to landslides as there is loess (ML soils) overlying basalt on steep slopes. The geology coupled with high levels of precipitation lead to high numbers of slope failures. Landslide indicators are hard to distinguish due to the heavy vegetation, LIDAR penetrates the vegetation for a view of the ground surface. In this project we attempted to define large-medium-small landslides in the West Hills, Portland using LIDAR. We created contour lines with different intervals (5, 15, 50 feet) and compared the number and area of landslides identified in each. The intervals set at 5 and 15 feet allowed us to see "small" landslides (~5,000 sq ft), as well as the large ones. The 50 foot interval depicted smaller landslides with an area of 20,000 sq ft. While the 5 and 15 feet contour lines were the best, they also took the longest time to load while panning the screen. We found hill shading wasn't as accurate as contour lines. To maximize the image, we draped contour lines over the hill shade. This gave a better view of what the ground morphology was really like. The 5 foot contour was the best. We could see some of the same landslides in the 15 foot contour, especially with the hill shade image below. We recommend using the 50 foot contour when locating larger slides in areas that are heavily populated it was difficult to discern the housing complexes with disturbed slopes.

MODELS TO PREDICT WILDFIRE RELATED DEBRIS-FLOW VOLUMES

Recently burned basins frequently produce debris flows in response to moderate and heavy rainfall. Hazard assessments can benefit from the ability to predict the volume of debris-flow material that may be generated from burned basins. This study develops a set of empirically-based models to predict potential volumes of wildfire-related debris flows in different regions and geologic settings.

For 56 basins in eight recently burned areas located in Colorado, Utah and California, debris-flow volumes were estimated by quantifying the amount of material eroded from the main channels in a basin using surveys of closely-spaced channel cross sections. Measures of basin morphology were calculated from 10- or 30-m digital elevation models. Maps of burn severity were used to quantify the basin areas burned at low, moderate, and high severities. Material properties for each basin were characterized by different measures of the grain-size distribution determined from field samples of burned soil. The dominant rock type underlying the debris-flow producing basin was characterized either from field observations or from geologic maps. Networks of rain gages installed throughout the burned areas provided rainfall amounts and intensities for debris-flow triggering storms.

Stepwise multiple regression analyses were used to generate separate models for basins in the western U.S., southern California, and the Rocky Mountains. Specific models were also generated for basins underlain by sedimentary, metamorphic and granitic rock types. Some of these models can be used to estimate potential volumes of debris-flow material that may deposit at the mouths of recently-burned basins.

THE STEMWINDER MOUNTAIN DEBRIS FLOODS-- AN UNUSUAL EVENT

Giles, Tim, Research Geomorphologist, British Columbia Forest Service, Southern Interior Forest Region, 515 Columbia St., Kamloops, BC, V2C 2T7, Canada, tim.giles@gov.bc.ca

On August 16, 2004, a series of debris floods blocked Highway 3 and reached the Similkameen River just west of Hedley, British Columbia. Three small grassland watersheds experienced a short but intense rainstorm event which caused sheetwash, rilling, gullying and channel bank erosion, and initiated debris floods through the channels and onto the fans. A weather station 2 kilometres west along the Similkameen River valley recorded 30.7 mm of rain in a one hour period just prior to the events.

Intense precipitation exceeded the infiltration capacity of the thin grassland soils in the upper Stemwinder Creek watershed, causing overland surface flow which rapidly concentrated water into small rills. Headward gully erosion occurred on steeper, convergent slopes. In the upper channel, scouring between 2 and 5 metres was common for 1000 metres. In the lower reaches, the mainstem channel runs for 1000 metres through a slaty-shale bedrock-rimmed canyon and undercuts thick, loosely consolidated talus. Where the channel emerges from the canyon, the debris flood spread out and sediment began depositing on the 10 degree gradient fan. The majority of the sediment, estimated to be as much as 30,000 cubic metres, is angular pebble-cobble sized fragments of slaty-shale bedrock sourced from the talus slopes.

There are several large alluvial fans along the Similkameen River in this area and these were generally thought to be inactive relict deposits. The Stemwinder Creek channel was charged with coarse sediment and this event is termed transport-limited (sediment supply unlimited). This is common in the dry southern interior of British Columbia where active colluvial processes occur on slopes, but the long return period of intense precipitation events restricts the occurrence of catastrophic events. Observations in the watershed, along the channel and on the fan suggests that this event may not be as unusual as first anticipated. A better understanding of the frequency and magnitude of this type of occurrence will be instrumental in developing guidelines for future land-use activities on these fans.

LANDSLIDES, LASERS, AND LOGARITHMS: THE ROLE OF EMERGING TECHNOLOGIES IN WATERSHED-SCALE SLOPE STABILITY ASSESSMENT   KEYNOTE PRESENTATION

Haneberg, William C., Haneberg Geoscience, 10208 39th Avenue SW, Seattle WA 98146, bill@haneberg.com

Emerging technologies such as airborne lidar, digital terrain visualization, empirical and rational slope stability models, and virtual geologic mapping are giving practicing engineering geologists new tools for watershed-scale landslide hazard assessment. During the past decade, for example, airborne lidar has been elevated from a research topic to a practical and cost-effective geologic tool. Airborne lidar is likely to have the greatest geologic potential value in areas where it performs the worst: steep, rough, and heavily vegetated terrain. Non-standard processing that is unfamiliar to many lidar vendors can help to optimize and enhance lidar digital elevation models (DEMs) of challenging terrain. These include an avoidance of triangulated irregular networks (TINs) and linear interpolation of DEMs, which commonly produces large triangular facets that obliterate geologic details in areas of low lidar ground strike density. Higher-order techniques such as thin plate splines with tension or nonlinear natural neighbors can be used to interpolate smooth and continuously differentiable DEM surfaces. The DEM grid spacing and interpolation parameters must be chosen with care and geologic insight, however, because higher order techniques can also give rise to unwanted interpolation artifacts. This work is best performed under the supervision of an experienced geologist. Once an optimal DEM has been created, geomorphic derivative maps showing slope angle, residual topography, topographic roughness, and other measures can be created and layered with a suite of shaded relief images to create a virtual landscape conducive to office-based landslide hazard mapping. When supplemented by field verification, this approach is in many ways superior to traditional field-based approaches and can help to identify subtle geomorphic features that would otherwise go unnoticed. A new generation of commercial interferometric radar DEMs with 5 m elevation postings, currently available for all of California, is also well suited for this kind of work. In addition to providing detailed terrain models for virtual mapping, high resolution lidar and radar DEMs can be used as input for empirical and rational landside hazard models. Although they tend to be more mathematically complicated and difficult to parameterize than empirical models, process-based landslide models, most of which are currently based on infinite slope assumptions, have the potential to evaluate the effects of rare or unprecedented events affecting watersheds. Empirical models, in contrast, are generally limited to conditions similar to those under which their supporting data were collected. Both classes of models can be susceptible to the effects of DEM elevation errors, which can produce slope angle errors of ±2 to ±3 degrees (1σ) for lidar DEMs and ±3 to ±5 degrees for conventional 10 m DEMs. The steep terrain and large area of many watersheds also makes geotechnical and hydrogeologic characterization difficult and unrealistically expensive. Rational probabilistic models can be used to incorporate the effects of input variability on calculated results and potentially produce results that are more amenable to quantitative risk assessment than traditional geologic maps.

WATERSHED SCALE LANDSLIDE HAZARD ASSESSMENT USING PROBABILISTIC INFINITE SLOPE ANALYSIS (PISA) FOR STATIC AND SEISMIC CONDITIONS

Haneberg, William C., Haneberg Geoscience, 10208 39th Avenue SW, Seattle WA 98146, bill@haneberg.com

PISA-m is a computer program that calculates deterministic and probabilistic results based on an infinite slope approximation distributed across watershed scale slope angle, soil, and forest cover maps. Slope angles are calculated from digital elevation models (DEMs) and include an estimate of the effects of DEM elevation errors. Geotechnical input is given as a soil unit map, a forest cover unit map, and a parameter file. The soil map is divided into units with unique combinations of soil-based geotechnical values (soil cohesion, angle of internal friction, unit weight, soil thickness, and pore water pressure). The forest cover map is divided into units with unique sets of tree root cohesive strength and tree surcharge. Each of the variables in each of the map units can be specified as zero, a constant, an empirical mean and variance, or a theoretical probability distribution (uniform, normal, triangular, exteme value, or β-PERT). Static output options are grids of:

• Factor of safety means (FS)
• Factor of safety standard deviations (s_FS)
• Lognormal probabilities of sliding (Prob FS ≤ 1)
• Non-parametric slope reliability indices ((FS-1)/s_FS)

Seismic output options are grids of:

• Newmark acceleration means (a_N)
• Newmark acceleration standard deviations (s_aN)
• Normally distributed Newmark non-exceedance probabilities (Prob a_N ≤ a_threshold)
• Non-parametric Newmark reliability indices ((a_N - a_threshold)/s_aN)
• Newmark displacement means (D_N)
• Newmark displacment exceedance probabilities (Prob D_N ≥ D_threshold)

PISA-m uses an analytical first-order, second-moment (FOSM) expansion to estimate the variance of the results based on the variance of the input. Comparison with numerical Monte Carlo simulation results shows that the FOSM approximation works well for distributions that are symmetric or nearly so. FOSM is also non-iterative and produces relatively compact output, which can be important when using high resolution LiDAR DEMs containing tens of millions of values. The current implementation is written in Fortran 95 using dynamic memory allocation, allowing input/output maps of any size. It does not include machine-specific language extensions and uses a simple command line interface, so it can be made to run on any operating system with a Fortran 95 compiler (using features such as auto-vectorization for parallel computing if available). Map input and output is via Arc ASCII grid or Surfer ASCII grid files, which can be displayed in a variety of 2D and 3D visualization programs. Possible future enhancements may include separate soil depth and/or pore water pressure maps and binary input/output for increased speed.

FIRE-CLIMATE RELATIONSHIPS ON THE COLORADO PLATEAU: A HOLOCENE FIRE-RELATED DEBRIS FLOW CHRONOLOGY FOR KENDRICK MOUNTAIN, ARIZONA

Jenkins, Sara E., Quaternary Sciences Program, Northern Arizona University, 303 W. Fir, Flagstaff, AZ 86001, sej22@NAU.edu

The dynamic, non-random relationship between climate, wildfire, and multi-scale geomorphic response over various timescales has become an important theme for research in forested ecosystems of the western United States. Wildfire recurrence in ponderosa pine and mixed conifer forests has been linked to climate perturbations on multiple timescales; fire history reconstructions on the Colorado Plateau, however, suggest a Holocene regime of highly recurrent, low severity fires independent of climate perturbations. In this study, a late Holocene fire-related geomorphic history for Kendrick Mountain, Arizona, reveals episodes of high severity fire in ponderosa pine-mixed conifer landscapes of the southern Colorado Plateau. Radiocarbon dating of 28 fire-related debris flow deposits in four basins and associated burn layers yielded dates for these events. Morphometric relationship thresholds for debris flow initiation within study basins are similar to values calculated for neighboring regions and flow types. Basin slope and percent of basin burned appear significant for debris flow initiation in this region. Peaks in cumulative event probability for the mountain are centered around 250-500, 725-850, 1100-1175, and 1650-1850 cal yr BP, with a minor peak at 3350 cal yr BP. A comparison with regional climate proxies reveals little correlation between climate trends and inferred wildfire events. This suggests that local climate factors on short timescales are the dominant forcing mechanism for stand-replacing fires. Throughout the Southwest, short-term moisture fluctuations due to El Ni–o-Southern Oscillation (ENSO) patterns have been linked to fire occurrence and a similar relationship may be present on the southern Colorado Plateau. Multi-decadal error bars on the ages of Kendrick debris flow events, however, preclude comparison with local climate shifts. In the future, a more complete geomorphic record may refine assumptions about the relationship between fire and climate for this region. One important implication of the Kendrick research is that refinement of reference conditions in Colorado Plateau ponderosa pine-mixed conifer forests should include mention of the potential for high severity fires resulting from short-term moisture fluctuations.

LANDSLIDES CAUSED BY FORESTRY OPERATIONS IN BRITISH COLUMBIA-- RESEARCH, PREDICTIVE MAPPING, AND WATERSHED-SCALE IMPACTS

Jordan, Peter, Southern Interior Forest Region, British Columbia Ministry of Forests and Range, 1907 Ridgewood Rd., Nelson, BC, V1L 6K1, peter.jordan@gov.bc.ca

Landslides caused by logging and forest roads in British Columbia have long been recognized as an important issue, affecting fish habitat, water quality, stream channel stability, and sometimes public safety. In 1995, the provincial Forest Practices Code (FPC) legislation was introduced, requiring forest licensees to plan for landslide hazard, and to have terrain stability assessments done by licensed professionals in potentially unstable terrain. This prompted activity in two areas: terrain stability mapping was done by the forest industry over large areas of the provincial forest; and a number of research and inventory studies were done on forestry-related landslides, mainly by Forest Service researchers.

Terrain stability mapping is a procedure used in British Columbia to identify areas subject to landslide hazard. It is based on a surficial geology map, to which a 5-level hazard scale is applied, largely on the judgment of the mapper. One of the purposes of the landslide research projects described here is to develop improved criteria for assigning hazard ratings, based on the statistical relations between landslide frequency and various terrain attributes.

Research projects have been conducted in three hydrologic regions of the province: the hyper-maritime outer west coast of Vancouver Island and the Queen Charlotte Islands; the less maritime but still rainfall-dominated coastal region of the Coast and Cascade Mountains, and the interior, snowmelt-dominated Columbia and Rocky Mountains. The results of these studies show a clear trend to higher landslide frequencies, both natural and development-related, in the more maritime environments. All regions show a typical one order of magnitude increase in landslide frequencies in susceptible terrain following forest development. At the coast, many or most landslides are caused by clearcutting, with root strength loss and local hydrologic changes implicated as contributing factors. In the interior, most landslides are caused by roads, either unstable road fills or road drainage diversions. Certain terrain types, such as stratified glacial deposits and gullied terrain, have higher than average susceptibility to landslides.

In 2004, the FPC was replaced by the Forest and Range Practices Act (FRPA). The new act has fewer and less detailed regulations, and the number of plans and assessments which must be approved by the Forest Service are greatly reduced. It is less prescriptive and more "results-based"; for example, forest operations must not cause a landslide which has a "material adverse effect" on certain specified values. This more risk-based approach will require more attention to the impacts of landslides and erosion events on streams. Relatively little research has been done in British Columbia on watershed-scale effects of landslides; however, there are some documented examples which provide guidance in assessing the impacts of forestry-related mass movement, both at the short-term single-event scale, and the cumulative watershed scale.

PROGRESS IN FOREST ENGINEERING GEOLOGY IN THE LATE 20TH CENTURY AND EARLY 21ST CENTURY-- THE INTEGRATION OF ENGINEERING GEOLOGY WITHIN ECOSYSTEM MANAGEMENT OF FORESTED LANDS   KEYNOTE PRESENTATION

Koler, Tom, USFS El Dorado National Forest, 100 Forni Road, Placerville, CA 95667, tkoler@fs.fed.us

The role of engineering geologists working for private, industrial, and public land managers has been influenced by the exponential increase in timber demand after World War II and the parallel adoption of environmental legislation to protect and conserve natural resources. Through the 1940s, the selection of timber harvest units in gentle terrain was simple, but by the 1950s logging in rugged terrain made selection more difficult. The relationship between harvesting methods, slope instability, and flooding became a concern for managers after the 1955 and 1964 floods in the American West.

In the 1960s and 1970s, state and federal environmental legislation helped to mitigate natural resource damages resulting from timber management. The Forest Service increased the number of engineering geologists it employed during late 1970s and 1980s. Engineering geologists remained rare in industry until habitat conservation plans were developed during the Reagan administration and implemented during the Clinton administration. By the late 1980s, American values shifted to emphasize resource protection and conservation, resulting in a decrease in timber harvesting on public lands and a standoff between environmental groups and the timber industry.

During the early 1990s, the Forest Service shifted from multi-use management to ecosystem management, which marked the end of large timber harvest targets and an emphasis in resource protection and conservation. The first habitat conservation plans were adopted by Plum Creek Timber Company in Washington State and The Pacific Lumber Company in Northern California in the late 1990s. The early years of this decade have seen a continuation of this policy shift and the number of engineering geologists working in forestry has continued to grow. Engineering geologists working in forestry today must have skills to accommodate management decision making within an ecosystem approach.

EFFECTS OF URBANIZATION/DEVELOPMENT ON MASS WASTING   KEYNOTE PRESENTATION

Laprade, William T., Vice President, Shannon & Wilson, Inc., 400 N. 34th Street, Suite 100, P.O. Box 300303, Seattle, WA 98103, WTL@shanwil.com

Experiences and much of the literature document the deleterious effects of human modification of the land; however, land development and engineering works can also increase the stability of unstable slopes before instability occurs. Using Seattle as an example, it can be shown how (1) poor engineering practices, particularly in the early days of development, contributed to instability, and (2) proper engineering practices, particularly in the last 20 years, stabilized slopes that were chronic areas of instability.

Types of detrimental actions include: deranged drainage, septic drainfield installation, concentration of storm runoff, uncontrolled or unengineered fill for streets and yards, unretained cuts, and clearing of vegetation without a reforestation plan. Beneficial types of action during development include: sewer pipe installation, interception of natural drainage, interception of groundwater by deep interceptor trenches or residential footing drains, surface water collection into storm drains, retention of natural weak colluvium by walls, removal of weak colluvium, and planned revegetation on steep hillsides.

Urban areas go through life stages much the same as a person: youth, adolescence and maturity. This can also be applied to mass wasting in the urban environment. During its youthful days, a townÍs residences and infrastructure fit into the topography with a soft footprint, avoiding the most obvious geologically hazardous areas. In its teenage years, the adolescent city makes mischief by cutting and filling, and expanding into areas where it should not. In maturity, the city recognizes its wayward ways, fixes its errors and requires newcomers to conform to new rules that minimize mass wasting. What age is your city?

MASS WASTING FOLLOWING FOREST FIRES: PROCESSES AT MULTIPLE SCALES   KEYNOTE PRESENTATION

Luce, Charles H., Research Hydrologist and Watershed Research Team Leader, USDA Forest Service, Rocky Mountain Research Station, Boise Aquatic Sciences Lab, 322 E. Front Street, Suite 401, Boise, ID 83702, cluce@fs.fed.us

Substantial changes to vegetation and soil properties after wildfire temporarily increase risks of mass wasting through multiple processes. While the loss of root strength following tree mortality has received substantial attention as one mechanism for debris flow initiation, there is a growing recognition of the role of increased runoff from severely burned areas in initiating gullies and bulking debris flows. These two processes yield very different temporal patterns of risk following wildfire at both seasonal and multi-annual scales. While root strength changes yield a maximum in risk several years after wildfire, water repellency imposes maximum risk immediately after the fire. Further, in many climates shallow landslides are most likely to occur during passage of large storm fronts during winter, sometimes during rain-on-snow, whereas bulking debris flows generally initiate during summer thunderstorms, when water repellency is most pronounced. The different initiation mechanisms also impart differences in debris flow rheology, with bulking debris flows traveling on lower slopes than landslide-initiated debris flows. These differences can also yield different spatial scaling of synoptic mass wasting events, either controlled by thunderstorm scales or snowmelt patterns. Climatic characteristics, including seasonal timing of precipitation and precipitation type, play a key role in determining the regional patterns of process dominance. Understanding the key controls on these processes can lead to better assessments of risk and improved planning for at-risk resources following wildfire. This presentation will describe the difference in these two debris flow initiation processes and illustrate the implications of the differences at local, basin, and regional scales.

MASS-WASTING IN VOLCANICALLY DISTURBED WATERSHEDS: PRIMARY EVENTS, LANDSCAPE RESPONSES, AND GEOLOGICAL-ECOLOGICAL INTERACTIONS   KEYNOTE PRESENTATION

Major, Jon J., U.S. Geological Survey, Cascades Volcano Observatory, 1300 SE Cardinal Court, Suite 100, Vancouver, WA 98683, jjmajor@usgs.gov

Mass wasting at volcanoes and in volcanically disturbed watersheds occurs both as primary and secondary events, encompasses a variety of scales, triggers assorted landscape responses, and is influenced by geological-ecological interactions. Primary landslides and debris flows from volcanoes can occur with or without an eruption, flow tens to hundreds of km at speeds of tens of km/hr, destroy or damage all structures along their flow paths, and claim thousands of lives in a single event. In contrast, landslides and debris flows that originate in nearby volcanically disturbed watersheds are post-eruptive events that are smaller and chiefly less destructive than those from volcanoes; nonetheless, they, too, can have significant geomorphic and land-management ramifications.

The style of volcanic disturbance strongly affects the nature and frequency of mass-wasting in disturbed watersheds. Shallow landslides and debris flows are more common in areas where trees are killed and uprooted compared to areas less ecologically afflicted. Following the 1980 Mount St. Helens (MSH) eruption, numerous small to moderate (102-105 m3) slides occurred in steep areas of the volcanic blast zone, mainly within 5 years of the eruption. Between 1980 and 1984 the frequency of slides in that disturbance zone was 35 times greater than that in undisturbed forested areas and 9 times greater than that in areas clear-cut elsewhere in the Cascade Range.

The scale of mass wasting strongly influences subsequent landscape response. Small-scale mass movements typically produce locally transient perturbations having relatively short-lived geomorphic impacts. Large volcanic mass movements, however, can broadly perturb drainage networks and leave geomorphic legacies that persist for decades to centuries. At MSH, for example, suspended-sediment fluxes from watersheds subject chiefly to the volcanic blast and subsequent small landslides returned to pre-eruption levels within 5 years, whereas fluxes from watersheds affected by large (107-108 m³) volcanic debris flows and a gigantic (2.5 km³) volcanic debris avalanche remain 10-100 times greater than pre-eruption fluxes even after 25 years.

Geological-ecological interactions in volcanically disturbed watersheds can also affect mass-movement activity. For example, in landscapes where trees have been killed but not uprooted mass movements can occur years later when root decay weakens the stability of the soil mantle. Also, post-eruptive evolution of infiltration capacities can trigger evolution from shallow landsliding shortly after an eruption to more deep-seated landsliding with time. Thus, while mass-movements in volcanically disturbed watersheds can trigger assorted landscape responses, their occurrence and temporal evolution also represent a response to volcanic disturbance.

SLOPE STABILITY PROBLEMS IN COLORADO SKI AREAS

McCalpin, James P., GEO-HAZ Consulting, Inc., Crestone, CO

Because most ski area operations occur on land leased from the US Forest Service, proposals to develop new trails, roads, or snowmaking trigger an Environmental Impact Statement, and thus involve a geologist. Often the "fatal flaw" of a proposal (at least as seen by the USFS and opposition groups) is the presumed destabilizing effects of tree clearing and snowmaking on quasi-stable hillslopes. As a result, careful mapping of historic and prehistoric landslides, and of the present network of ski area runoff ditches, is necessary before remediation measures can be designed. In the ski areas below, slope stability concerns have been critical.

• Aspen Highlands, 1994; client wanted to place a restaurant complex in topographic depressions at the summit, but they turned out to be sackungs.
• Aspen Mountain, 1996; debris flow initiated on Aspen Mountain and damaged Aspen Music School at base of mountain.
• Buttermilk, 1995; creeping landslide slowly stretched Tiehack lift until it had to be replaced.
• Burnt Mountain (proposed, 1995); large areas mapped as glacial till by USGS turned out to be prehistoric landslides.
• Keystone, 1997; major expansion into Jones Gulch area was torpedoed by widespread landsliding, helped along by endangered lynx corridor. Ski Tip lift line revised to avoid landslides.
• Powderhorn, 2000; spring landslides moved lift towers on both active lifts.
• Durango Mountain Resort, 2005; Forest Service noticed hummocky topography in area of proposed snowmaking, but it was caused by differential erosion of bedrock; no landslide problems!
• Snowmass, 2005; active slump area in Funnel Trail required drainage rerouting before adjacent snowmaking and terrain park could be approved.
• Crested Butte (in progress); opening of new ski mountain at Crested Butte is currently held up by slope stability concerns.

WHY DO WE NEED TO CHARACTERIZE MINE WASTE-ROCK PILES, TAILINGS, AND HEAP LEACH FACILITIES?

McLemore, Virginia T., New Mexico Bureau of Geology and Mineral Resources, New Mexico Institute of Mining and Technology, Socorro, NM 87801, ginger@gis.nmt.edu

Characterization of mine waste-rock piles, tailings dams, heap leach facilities and naturally exposed alteration areas is important 1) to establish pre-mining background conditions, 2) to characterize and predict stability, weathering, and erosion, 3) to predict acid-rock drainage and other chemical releases, 4) to properly dispose of and manage mine wastes, and 5) to develop mine closure plans. Waste rock is the unprocessed overburden material that is excavated and disposed of in order to access valuable ore bodies. Tailings are the waste materials produced from the extraction process used for obtaining ore. Natural alteration areas are important to characterize to begin to differentiate between natural and man-made contamination within a given watershed. Maps and aerial photographs are essential tools for the planning and execution of any characterization project. Most site characterizations are based upon drilling, shallow surface test pits, or surface sampling. At some mines, remediation or re-mining of the rock piles and tailings has resulted in partial or complete removal of the rock pile or tailings material. Multidisciplinary characterization studies (geologic, mineralogical, geochemical, biological, geotechnical, and hydrologic studies) have become increasingly important to analyze the effects on nearby water quality and the stability of rock piles, tailings, and hydrothermal alteration areas. Mineralogy and petrology are especially important. Numerous sampling methods are available and different methods are employed based on the specific site and purpose of the characterization study. Most characterization studies to date focus on the stability of rock piles and tailings or acid drainage from the piles. By understanding why rock piles and tailings fail and how acid drainage is formed, perhaps we can build them to be more stable and prevent acid drainage. More recent research is examining the effect of weathering on the stability or rock piles. Several case studies will be briefly mentioned addressing these issues.

DISTURBANCE REGIME MODELING FOR WATERSHED ANALYSIS & STATE FOREST MANAGEMENT IN OREGON

Michael, David L., Oregon Department of Forestry, 801 Gales Creek Road, Forest Grove, OR 97116, DMICHAEL@ODF.STATE.OR.US

In the process of conducting watershed analysis Oregon Department of Forestry (ODF) has moved toward computer modeling of disturbance regime "punctuated" landslide process for management of slope stability concerns in Oregon State Forests. The State Forests Program (SFP) under the Northwest Oregon Plan began conducting watershed analysis in 2004. In the second watershed analysis performed, the slope stability portion was sub-contracted to Earth Systems Institute who applied "Terrain Resource Inventory and Analysis Database" (TRIAD) a disturbance regime computer modeling technique to the watershed analysis. Further investigation into this approach to data collection and management, has lead ODF, SFP to contract the terrain analysis of 800,000 acres. Terrain databases, in conjunction with software tools, can provide new types of information for landscape management and can support existing watershed programs including habitat inventories, watershed analysis, habitat conservation plans, land acquisitions, cumulative effects studies, restoration plans, and TMDL assessments. The application of such computer modeling in conjunction with site specific field based geotechnical input will undoubtedly provide better management decisions and thus greater resource protection in all aspects of OregonÕs State Forest management.

DEBRIS-FLOW RUNOUT ESTIMATIONS USING TOPOGRAPHIC PARAMETERS

Prochaska, Adam B., aprochas@mines.edu; Paul M. Santi; and Jerry D. Higgins, Department of Geology and Geological Engineering, Colorado School of Mines, Golden, CO 80401; and Susan H. Cannon, U.S. Geological Survey, Golden, CO 80401

Prediction of the runout length of a debris flow is important for the delineation of potentially hazardous areas on the fan and for the siting of mitigation structures. Conventional runout estimation methods are often sensitive to input parameters that may be difficult to estimate with certainty, such as volume, velocity, and frictional parameters. In order to avoid the use of these unreliable input parameters, this research developed a runout estimation model that is based on easily-measured topographic parameters. Based on a runout estimation method developed for snow avalanches, this model predicts debris-flow runout as an angle of reach from a fixed point in the drainage channel to the end of the runout zone. The most accurate fixed point to use was found to be the mid-point elevation of the drainage channel, measured from the apex of the alluvial fan to the top of the drainage basin. The accuracy of this model compares well to accuracies reported in the technical literature for existing runout estimation methods. The robustness of this model was tested by applying it to three debris-flow events not used in its development; predicted runout ranged from 82 to 131 percent of the actual runout for these three events. Prediction interval multipliers were also developed so that the user may calculate predicted runout within specified confidence limits.

PERFORMANCE OF FLEXIBLE DEBRIS FLOW BARRIERS IN FIRE BURNED AREAS, STATE ROUTE 18, SAN BERNARDINO COUNTY, CA

Rorem, Erik J., President & General Manager, erik.rorem@geobrugg.com, and John Kalejta II, john.kalejta@geobrugg.com, Project Engineer, Geobrugg North America, LLC, 551 W. Cordova Road, PMB 730, Santa Fe, NM

As presented at the 55th Annual Highway Geology Symposium in 2004, various flexible debris flow barriers were installed near San Bernardino, CA in the summer of 2004 in anticipation of debris flows originating from fire burned slopes.These debris flow barriers were installed in ten distinct debris flow channels upslope from and opening onto state route 18. Each of the barriers was dimensioned based on a unique dimensioning model using data provided by Caltrans including anticipated debris volumes and velocities, and a broad characterization of the expected debris flow compositions, channel geometry and barrier orientations.Each site required a unique barrier design with differing barrier heights, capacities and support infrastructure.

Construction of these barriers was completed in June, 2004. As anticipated, heavy rains in October and winter of 2004/2005 resulted in significant debris flows in all the identified channels. During the October events, all the barriers were impacted to various degrees. The barriers performed exactly as intended, and were subsequently cleaned of debris. As a result of these events, some aspects of designs were identified that could be adjusted to better facilitate cleanout maintenance and improve performance in general. In the winter of 2004/2005, the barriers were impacted again by debris during storm events. Some barriers were completely filled with debris and even somewhat overtopped, and some damage to the barriers was evident due to the greater than expected debris volumes. However, the drainage culverts immediately below the barriers remained clear and effective as intended, thus channeling water flow underneath rather than over the road. Ironically, this protected section of road was now used as a detour for other sections of road that were closed due to problems created by debris flow and rockfall; an opposite scenario from one year earlier.

The application of these barriers can be considered a complete success, performing exactly as intended. Some minor modifications to the barriers will be undertaken to prevent subsequent damage and to better facilitate maintenance.

MODELING RELATIVE LANDSLIDE SUSCEPTIBILITY WITH THE WEIGHT-OF-EVIDENCE MODEL

Rosenberg, Lewis I., San Luis Obispo County Planning Dept., P.O. Box 1693, Tijeras, NM 87059, Lrosenberg@co.slo.ca.us

The weight-of-evidence (w-o-e) model, originally developed by the Geological Survey of Canada (GSC) and the U.S. Geological Survey (USGS) for mineral potential mapping, can be adapted to landslide susceptibility mapping. The w-o-e model uses a set of map databases and the hypothesis of Òthis location is susceptible to landsliding.Ó Weights are estimated from known landslide locations and the variables used as predictors. The w-o-e model is run as an extension using ArcGIS geographic information system (GIS) software with the GSC/USGS ARCSDM3 extension.

Predictor variables for the w-o-e model included slope inclination, slope aspect, slope curvature, stratigraphic units, fault zones, and vegetation type. These predictor variables were converted from vector databases to raster files in ArcGIS. The rasters were combined in ARCSDM3 and analyzed to evaluate which variables correlated with known landslide localities. The strongest correlations with mapped landslides were slope inclination, slope curvature, and stratigraphic unit.

The results of the w-o-e model are used to focus on the areas of highest relative landslide susceptibility in San Luis Obispo County. These areas are zoned as Geologic Study Areas. The combination of more accurate landslide susceptibility mapping and zonation, combined with peer-review of site-specific geotechnical studies, help to reduce losses from landslide damage in San Luis Obispo County.

SOURCES OF DEBRIS FLOW MATERIAL IN BURNED AREAS

Santi, Paul M., psanti@mines.edu; Victor G. deWolfe; Jerry D. Higgins, Department of Geology and Geological Engineering, Colorado School of Mines, Golden, CO 80401; and Susan H. Cannon and Joseph E. Gartner, U.S. Geological Survey, Golden, CO 80401

The vulnerability of recently burned areas to debris flows has been well established. Likewise, it has been shown that many, if not most, post-fire debris flows are initiated by runoff and erosion and grow in size through erosion and scour by the moving debris flow, as opposed to landslide-initiated flows with little growth.

To better understand the development and character of these flows, a study has been completed encompassing 46 debris-flow events in California, Utah, and Colorado, in nine different recently burned areas.

For each debris flow, progressive debris production was measured at intervals along the length of the canyon, and from these measurements graphs were developed showing cumulative debris volume as a function of channel length. All 46 debris flows showed significant bulking by scour and erosion, with average yield rates for each channel ranging from 0.4 to 12 cubic yards of debris produced for every yard of channel length, with an overall average value of 3. Significant increases in yield rate partway down the channel were identified in 83% of the channels, with an average of a three-fold increase in yield rate. Yield rates for short reaches of channels (up to several hundred yards) ranged as high as 26.7 cy/yd. Debris was contributed from side channels into the main channels for 52% of the flows, with an average of 23% of the total debris coming from those side channels. Rill erosion was identified for 30% of the flows, with rills contributing between 0.1 and 10.5 % of the total debris, with an average of 3%. Debris was deposited as levees in 87% of the flows, with most of the deposition occurring in the lower part of the basin. A median value of 10% of the total debris flow was deposited as levees for these cases, with a range from near zero to nearly 100%. These results show that the channel erosion and scour are the dominant sources of debris in burned areas, with yield rates increasing after some threshold is exceeded partway down the canyon. Side channels are much more important sources of debris than rills. Levees are very common, but their size and effect on the amount of debris that reaches a canyon mouth is highly variable.

THE FLORIDA RIVER LANDSLIDE AREA-- INFLUENCE OF THE MISSIONARY RIDGE FIRE FIVE YEARS LATER

Schulz, William H., wschulz@usgs.gov; Jeffrey A. Coe, William L. Ellis, and John D. Kibler, Geologic Hazards Team, U.S. Geological Survey, Box 25046, MS-966, Denver, CO 80225

Snowmelt during Spring 2005 triggered over a dozen landslides on a hillslope burned during the Missionary Ridge fire and located along the Florida River, about one kilometer downstream from Lemon Dam. The landslides were generally translational debris slides confined to a two-meter thick colluvial mantle. These landslides spawned debris flows, several of which flowed down to the Florida River floodplain. Our observations suggest that incineration of vegetation on the hillslope aided initiation of these landslides by removing root strength and evapotranspiration. Wildfire was not the only major disturbance on the hillslope; the 2005 landslides occur upon an approximately twenty million cubic meter, apparently dormant translational rockslide. Unlike the many dip-slope bedrock landslides in the area, the dormant landslide cuts through a 325-meter-thick section of bedrock, primarily of the Morrison Formation. This landslide shows evidence of having dammed the Florida River and of renewed activity during the recent past. We identified an apparent bedrock thrust over soil within the landslide and that soil contained abundant charcoal, perhaps suggesting a wildfire-related origin for the thrust event. We are analyzing the causes of recent landsliding and potential future landslide activity through radiometric dating, finite-element modeling, and surface displacement and shallow groundwater monitoring.

LANDSLIDE HAZARD ASSESSMENT AND PREDICTION-- APPROPRIATE OPTIONS AT DIFFERENT SCALES IN MANAGED TERRAIN   KEYNOTE PRESENTATION

Sidle, Roy C., Professor and Head of the Slope Conservation Section, Geohazards Division, Disaster Prevention Research Institute, Kyoto University, Uji, Kyoto, 611-0011, Japan

In the past two decades, a large number of landslide models and assessment procedures have been developed. These can be roughly divided into four categories: (1) terrain stability mapping; (2) simple rainfall-landslide and earthquake-landslide relationships; (3) multi-factor, empirical landslide hazard assessments; and (4) distributed, physically-based landslide models. Some of these methods can be readily used to assess relative landslide hazard at regional scales, others can be used as predictive tools for more specific sites, and yet others can be used to develop real-time warning systems. With each increment of specificity, the intensity of required data increases. Here, I wish to focus on the important attributes of each type of landslide assessment procedure with an emphasis on those that are useful for evaluating effects of land use, as well as the appropriate scales of application. Many of the landslides associated with widespread land use are numerous but typically small. Additionally, many of these occur in regions of the world where resources and expertise are limited. Thus, it is necessary to develop and implement landslide assessment methods for areas where certain critical data may be lacking and high-cost technology is unavailable. Terrain hazard mapping represents a somewhat general and qualitative level of landslide hazard assessment in which topographic, geomorphic, and geologic information are utilized together with data on pre-existing landslides to generate maps with broad categories of landslide hazards. Typically, such hazard mapping is developed or at least implemented by management agencies or regional governing bodies to evaluate the effect of various land uses on the occurrence of landslides. Thus, the main focus may not be to predict landslide occurrence, but rather to reduce the risk of landslide hazard related to a particular land use. If the assessment includes quantitative factors related to vulnerability of the elements at risk, then it can be considered a landslide risk assessment. Other types of terrain hazard maps may identify specific landscape or geologic features that are susceptible to slope failure.

Landslide hazard assessment based on simple relationships with rainfall characteristics has been applied at global and regional scales. When coupled with real-time rainfall data, such analyses can provide the basis for early warning systems for shallow landslides. For earthquake-triggered landslides, simple relations between earthquake magnitude and distance to the epicenter have proven useful for landslide hazard assessment. The major problems with earthquake analysis are uncertainties associated with future earthquake location, magnitude, and timing.

Empirical landslide hazard assessments share some common attributes with terrain hazard mapping, but generally differ with respect to the number of factors considered, their relationship to past landsliding, and how the factors are evaluated in the context of the assessment. In empirical landslide analysis, the factors contributing to landslide initiation are typically established based on characteristics of existing landslides. The end product is focused on producing maps, or at least useable decision tools, that relate landslide hazards to measurable environmental attributes Ð similar to terrain hazard analysis. Both bivariate and multivariate statistical analyses can be employed to examine each factor or several factors together, respectively, in combination with the presence or absence of landslides. Most approaches assume that landslides are more likely to occur under conditions similar to those of previous failures.

Physically-based landslide models typically assess stability in terms of a factor of safety for slope failure. Distributed, physically-based landslide models have two unique requirements: (1) spatially and, in some cases, temporally distributed model parameters are needed; and (2) the model output must be spatially and temporally explicit because of the need to know the locations and timing of landslides. Recent advances in incorporating advanced GIS and DEM technology into distributed, physical-based modeling has facilitated the prediction of landslides at the catchment scale. Major advantages of these models is their ability to simulate catchment-scale subsurface flow (and pore water pressures) and to assess dynamic vegetation attributes (e.g., rooting strength, species distribution) related to forest management and land cover change. Thus, it is possible to generate useful detailed spatial and temporal landslide scenarios for managed areas. As a theoretical advance from empirical landslide models based solely on rainfall characteristics, numerous infiltration-based landslide models have been developed for individual sites in both 2 and 3 dimensions. These models, however, are difficult to apply in topographically complex catchments. There is a conspicuous absence of hazard assessment and prediction methods that address the attributes and processes related to deep-seated landslides. Empirical methods that focus on rainfall trigger mechanisms apply only to shallow, rapid failures; other empirical multi-factor analyses use the same criteria for all failure types in a region. At present, terrain hazard mapping appears to be the only suitable strategy for deep-seated landslide hazard assessment, albeit of a very general nature.

NATURAL VARIATION LIMITS TO LANDSLIDE MODELS FOR WESTERN NORTH AMERICAN FORESTED WATERSHEDS

Spittler, Thomas E., California Geological Survey, 135 Ridgway, Santa Rosa, CA 94502, tom.spittler@conservation.ca.gov

Forested landscapes in western margin of North American watersheds are controlled by the complex interaction of geology, hydrology/climatology, and biology, each of which is in constant flux. Earthquakes, uplift and subsidence, sea level change, storms, droughts, floods, fires, and the growth, death, and evolution of species affect the region. Over-printed on the natural systems are the effects of 150 years of land management. Deterministic models can be useful in generally defining regional-scale mass wasting potential. However, because of natural variations within dynamic western watersheds, if the assumptions and uncertainties of individual models are clear the models may be inappropriately used.

Temporal variability of seven factors, including tectonic setting, geomorphic evolution, relative seal level (hydrologic base level) changes, soil formation processes, species evolution, climate and weather, and anthropogenic management need to be considered in developing and using slope stability models. Similarly, six spatial variables, including geologic rock units, geomorphic units, soil units, vegetation types, bioturbation, and anthropogenic management may affect landslide models. The cumulative range in time for the temporal variables spans 12 orders of magnitude, from seconds for individual storm pulses to tens of millions of years for tectonics, while the cumulative range in size for the spatial variables span 16 orders of magnitude, from sub-millimeter for micorrhizal hyphae to hundreds of square kilometers for major rock units. Each identified variable may range from 3 to 12 orders of magnitude.

Because of the dynamic complexity of forested watersheds and the changes in land management that have occurred since Euro-American settlement of the western margin of the continent, where the limits of use are clearly identified regional deterministic models may aid in identifying sites that need specific review but they do not supercede site-specific observations and measurements.

POST-FIRE BURN SITE EVALUATION

Spittler, Thomas E., California Geological Survey, 135 Ridgway, Santa Rosa, CA 94502, tom.spittler@conservation.ca.gov

Fires are natural periodic disturbances in many western regions, particularly where the dominant vegetation is chaparral. Following wildfires the sediment production rates for early high-intensity storms may increase above background rates by more than an order of magnitude. However, over a millennial timeframe the denudation rate is about the same order of magnitude as the tectonic uplift rate. Although the long-term denudation rate in mountainous areas is relatively constant, for the short term, post-fire debris flows and debris floods are catastrophic events that destroy lives and property.

The change in the size and intensity of wildfires following suppression management by Euro-American immigrants into western chaparral areas, combined with the cessation of the Native American vegetation management, has resulted in fewer, larger, and more severe wildfires that may be accompanied by fewer but more sever post fire debris flows and debris floods.

The factors that contribute to potential life-threatening debris flows and floods include bedrock geology and associated regolith, steep slopes, youthful geomorphology, high burn severity, loss of in-channel vegetation, rainfall intensity or rate of snow melt, and time since the burn. These factors may be used to model the potential for watersheds to experience debris flows or debris floods. They are also used, in combination with observations suggestive of past debris flow activity, including the presence of alluvial fans, to evaluate the potential risk to life and property for individual sites that are not within larger, modeled watersheds.

MASS WASTING AND ITS CONTROL ON CHANNEL BEHAVIOR AND VALLEY FORMATION IN THE PARTRIDGE CREEK WATERSHED, ILLINOIS RIVER VALLEY, ILLINOIS

Stumpf, Andrew J., Quaternary Geology Section, Illinois State Geological Survey, Champaign, IL; Andrew C. Phillips, Quaternary Geology Section, Illinois State Geological Survey, Champaign, IL; Geoff E. Pociask, Wetlands Geology Section, Illinois State Geological Survey, Champaign, IL; Lisa R. Smith, Geospatial Analysis and Modeling Section, Illinois State Geological Survey, Champaign, IL; and William P. White, Center for Watershed Science, Illinois State Water Survey, Peoria, IL

Detailed field and remotely-sensed data collected from the Partridge Creek watershed suggest that mass wasting occurring since the end of the last (Wisconsin) glaciation has in part controlled the development of the drainage. Geologic, pedologic, morphometric, and hydrologic data were compiled to support ongoing studies of stream dynamics and sedimentation to achieve ecosystem restoration goals.

Post-glacial runoff partly dissected morainal and loess-covered uplands and carried large volumes of sediment to the Illinois River. At least three post-glacial mass wasting episodes have been identified based upon landform type, landscape position, material texture, and radiocarbon dating. These deposits delineate the configuration of the drainages at that time. High colluvial benches lying 20–30 feet above the modern channel are remnants of an early post-glacial valley system when base levels were higher than today. These deposits range from silt to silt loam in texture, and are composed of materials eroded from sparsely vegetated side-slopes and loess-covered uplands. Later infilling of valleys following incision established the modern drainage system.

Recent mass wasting can be subdivided into two events: (1) slope instability accelerated by clearing of forested land beginning in the mid 19^th century and (2) later mass wasting following widespread conversion of poorly-drained land to agriculture and construction of flood prevention measures during the late 19^th and early 20^th centuries. These deposits are generally coarser in texture than the old deposits. Preliminary investigations suggest that the current mass wasting is controlled by the internal structure of geologic materials coupled with changes in hydrology/land use. A trigger for these events may be recent base-level change due to channelizing, mining, and damming of the Partridge Creek and its tributary channels.

STABILITY CONCERNS AT THE PARKIN QUARRY, WASHINGTON COUNTY, OREGON

Theule, Joshua, jtheule@pdx.edu; Margaret Russell; and Faculty Sponsor: Dr. Scott Burns, Department of Geology, burnss@pdx.edu, Portland State University, Portland, OR

An earthflow occurred during heavy rains at the Parkin Quarry. The earthflow event initiated an engineering geologic evaluation of the failure and the existing quarry area. Preliminary earthflow mitigation measures and slope stability concerns encountered in the quarry were evaluated. The basis for the geologic mapping was an aerial photo and new topography prepared at 2 foot contours.

The Parkin Quarry is located on the eastern flank of a northwesterly plunging anticline. The quarry extracts diabase sill within shales associated in the Eocene. The sill and shale dip out of slope 45 to 55 degrees.

Placement of the fills was in a small drainage above the quarry highwall. Heavy rains in December 2005 triggered the failure of the fills; the earthflow flowed down the canyon and onto an adjoining neighbor's property. A preliminary evaluation indicated that the existing fills and earthflow materials will need to be stabilized.

By extracting the diabase sill in the quarry, portions of the shale have been destabilized located behind the sill. The stability of the shale and diabase vary due to the variations of weathering, dip slope of the shale, and the steep cut into the slope.

Preliminary stability analysis was performed on the shale. To stabilize the existing highwall will either entail, flattening of the highwall slopes or placement of fills in the quarry to buttress the existing highwall slope. Additional engineering geology and geotechnical evaluation may be needed depending on the final configuration of the quarry desired by the owner.

WASHINGTON'S LANDSLIDE HAZARD ZONATION PROJECT: A PROCESS FOR ADDRESSING RISK TO RESOURCES   KEYNOTE PRESENTATION

Vaugeois, Laura, Washington Department of Natural Resources, Forest Practices Division, 1111 Washington Street SE, Olympia, WA 98504-7012, laura.vaugeois@wadnr.gov

In nearly all forested watersheds of Washington where forest management activities have occurred, landslides, as a result of that management have provided the dominant sediment input to the associated aquatic system, far outpacing the naturally occurring rates. The goal of Washington's forest practice unstable slope rules is to prevent an increase in the naturally occurring rate of landslides due to forest management. This project was designed to assist land managers and regulators in determining the potential risk to public resources from forest management activities by mapping landslides and landforms that produce landslides. This type of approach produces a "no surprises" working environment between landowners, stakeholders, and regulators.

The project is designed as an integrated screening tool that exists and is utilized from GIS, uses existing information whenever possible, follows a standardized mapping protocol, is useful and accessible to all stakeholder groups, can serve as the data repository for unstable slope studies in the state, and can be used as the starting point for continued slope stability monitoring. At the beginning of this project, all available digital landslide inventories and mass wasting map units were collected and compiled into two GIS databases. These databases include both spatial and tabular data, to allow both spatial analyses of landslides and landforms interaction as well as statistical analyses. New data, as it is collected, is posted to the compilation and these GIS databases are updated on the internet quarterly for free download. Two websites were created to facilitate public involvement in the review process and timely access to the documents. To date, over 35,000 individual landslides have been mapped or compiled from previous mapping and over 1,100,000 acres have been mapped in this project.

• For more details, see: www.dnr.wa.gov/ forestpractices/lhzproject
• To download the compiled databases, go to: www.dnr.wa.gov/ forestpractices/data "Landslides" and "Landslide Hazard Zones"

MASS WASTING IN A DISTURBED TERRANE, ABANDONED COAL MINE SITE, SOUTHWEST INDIANA

West, T.R. and Fairfax, S.J., Dept. Earth & Atmospheric Sciences, Purdue University, West Lafayette, IN, trwest@purdue.edu

The Friar Tuck Abandoned Mine Lands site in southwest Indiana near Dugger, consisted of 1400 acres of unreclaimed castover strip piles and coal processing refuse deposits. Six major gob piles (coarse refuse), hundreds of acres of spoil ridges and slurry ponds remained from surface and underground mining conducted from 1929-1957. The Antioch Power plant operated from 1935-1952, but coal processing continued until 1965. Mining predated significant state and federal reclamation laws. Springfield, Hymera, and Danville coals (Indiana Coal V, VI and VII) were mined by area stripping and underground methods (room and pillar).

Natural soils consisted of loess, alluvium and till. Loess blanketed the upland, undisturbed area with Illinoian till below; alluvium occurred along Mud Creek on the northern boundary and a tributary stream on the west. Present were a 100 acre tailings pond (northwest corner), over 800 acres of spoil piles and 80 acres of gob piles; four major ones with the following acreage; Northwest=18, Northeast=14, Southeast=11 and Southwest=35.

Reclamation-design and construction are now completed for the site. The large tailings pond was vegetated by applying limestone and overseeding for three years. Northwest and Southwest gob piles were reclaimed by grading, construction of terraces and promoting vegetation. Two smaller gob piles were disposed of in isolated basins. Some grading was employed on the Northeast and Southeast gob piles, plus check dam construction and significant limestone application. Acid ponds were drained and several acid seeps were treated with crushed apatite.

Extensive erosion developed on the gob piles as acid conditions reduced vegetative cover so that increased runoff encouraged deep gully erosion. The Northwest gob pile had been previously reclaimed with a cover of loess which proved highly susceptible to erosion. In this study it was regraded and covered with mine spoil, a clayey material much less subject to erosion.

The mostly-barren soil exposure covering two square miles has been reclaimed and vegetative cover established. Work was accomplished through funding from the Abandoned Mine Lands Program under SMCRA, the Surface Mining Control and Reclamation Act, 1977.

LANDSLIDE POTENTIAL MAP FOR THE ISLAND OF POHNPEI, FEDERATED SATES OF MICRONESIA

White, Robin S., R.G. U.S. Department of Agriculture- Natural Resources Conservation Service, 6200 Jefferson NE, Room 305, Albuquerque, NM 87109-3734; Kevin Kinvig OÕahu RC&D, 99-193 Aiea Heights Drive, Suite 207, Aiea, HI 96701; Robyn L. Myers, Ph.D. U.S. Department of Agriculture- Natural Resources Conservation Service, 430 G Street, #4164, Davis, CA 95616-4164

The U.S. Department of Agriculture-Natural Resources Conservation Service (USDA-NRCS) was requested by the local government of the Island of Pohnpei, Federated States of Micronesia (FSM) to prepare a landslide potential map after back to back storms triggered 30 landslides resulting in property damage and 19 fatalities in April 1997. The local government to prepare a landslide hazard map could use the landslide potential map. The challenge was to find a method using available digital data and desktop Geographical Information System (GIS) technology instead of paper methods. Two approaches were used for the GIS system: a relative hazard/potential technique based on soils, vegetation and U.S. Geological Survey (USGS) Digital Elevation Model (DEM) data; and the Stability Index Mapping (SINMAP) methodology. SINMAP is a GIS extension based on the infinite slope model, which is used to predict areas of landslide origination. The resultant map was color coded to low, medium and high landslide potential areas. Both methods correlated well to landslides plotted by the USGS and other known landslides. The decision tree models tended to be more conservative and to group in the higher categories. The SINMAP result produced a finer distinction of stability for landslide potential and is a more scientifically approved method. The analysis was delivered as a set of final maps and a report to the USDA-NRCS Field Office in Pohnpei, and the Pacific Basin Office in Guam.

DEBRIS FLOWS AND FLOODS IN THE WAKE OF THE WILLOW FIRE, CENTRAL ARIZONA, 2004

Youberg, Ann, ann.youberg@azgs.az.gov, and Philip A. Pearthree, Arizona Geological Survey, 416 W. Congress, Suite 100, Tucson, AZ 85701

During late June and early July, 2004, 120,000 acres were burned by the Willow Fire in the Tonto National Forest, northeast of Phoenix, Arizona. As part of control and containment efforts back-burns were lit along the southeast side of the fire, using State Route 87 (SR 87) as a fire break. Much of the rugged area immediately north of SR 87 experienced medium or high burn severity. We participated in the Burned Area Emergency Response team to assess the potential for mass movement (landslides, debris flows, rockfalls) along SR 87. We reviewed pre-fire aerial photographs, burn-intensity maps, and topographic maps and then conducted a one day field survey of the burn area with Forest Service personnel. We recognized old debris flow levees on two small alluvial fans below steep, moderately burned drainages. The toes of these fans are cut by SR 87, so we inferred that post-fire debris flows could pose a threat to the highway. One week after our field reconnaissance, the area was subject to a monsoon thunderstorm and debris flows did occur in these drainages, temporarily closing the southbound lane of SR87. Larger drainages nearby exhibited rilling on sideslopes and large volumes of coarse bedload were deposited in ponded areas on the upstream side of SR 87. It is likely that debris flows in the upper watersheds of these drainages contributed coarse bedload to the larger trunk streams. We will present results from our field reconnaissance, details about the storm, and information regarding these post-fire debris flows and floods.