Randy B. Foltz

Spatial Prediction of Landslide Hazard Using Logistic Regression and ROC Analysis

An empirical modeling of road related and non-road related landslide hazard for a large geographical area using logistic regression in tandem with signal detection theory is presented. This modeling was developed using geographic information system (GIS) and remote sensing data, and was implemented on the Clearwater National Forest in central Idaho. The approach is based on explicit and quantitative environmental correlations between observed landslide occurrences, climate, parent material, and environmental attributes while the receiver operating characteristic (ROC) curves are used as a measure of performance of a predictive rule. The modeling results suggest that development of two independent models for road related and non-road related landslide hazard was necessary because spatial prediction and predictor variables were different for these models. The probabilistic models of landslide potential may be used as a decision support tool in forest planning involving the maintenance, obliteration or development of new forest roads in steep mountainous terrain.

Reopening abandoned forest roads in northern Idaho, USA: Quantification of runoff, sediment concentration, infiltration, and interrill erosion parameters

This study measured runoff and sediment concentration from the tire track and from the non-tire track to determine infiltration, interrill erodibility, and vegetative cover impacts of reopening an abandoned forest road. Runoff was lowest on the non-track portion of the abandoned road and highest on the reopened road. Sediment concentrations were significantly higher on the reopened road. Increased sediment concentrations were attributed to decreased vegetative cover, rather than traffic-induced changes in the physical soil properties of the reopened road. Thirty years of no traffic and vegetation regrowth was not sufficient to allow recovery of infiltration to values similar to an undisturbed forest. The study also found a significant dynamic behavior in interrill erodibility with respect to antecedent rainfall. Forest road erosion models that fail to account for this change will overestimate sediment yields.

Evaluating the efficacy of wood shreds for mitigating erosion

An erosion control product made by shredding on-site woody materials was evaluated for mitigating erosion through a series of rainfall simulations. Tests were conducted on bare soil and soil with 30, 50, and 70% cover on a coarse and a fine-grained soil. Results indicated that the wood product known as wood shreds reduced runoff and soil loss from both soil types. Erosion mitigation ranged from 60 to nearly 100% depending on the soil type and amount of concentrated flow and wood shred cover. Wood shreds appear to be a viable alternative to agricultural straw. A wood shred cover of 50% appears optimal, but the appropriate coverage rate will depend on the amount of expected concentrated flow and soil type.

A Synthesis of Post-Fire Road Treatments for BAER Teams: Methods, Treatment Effectiveness, and Decisionmaking Tools for Rehabilitation

We synthesized post-fire road treatment information to assist BAER specialists in making road rehabilitation decisions. We developed a questionnaire; conducted 30 interviews of BAER team engineers and hydrologists; acquired and analyzed gray literature and other relevant publications; and reviewed road rehabilitation procedures and analysis tools. Post-fire road treatments are implemented if the values at risk warrant the treatment and based on regional characteristics, including the timing of first damaging storm and window of implementation. Post-fire peak flow estimation is important when selecting road treatments. Interview results indicate that USGS methods are used for larger watersheds (>5 mi2) and NRCS Curve Number methods are used for smaller watersheds (2). These methods are not parameterized and validated for post-fire conditions. Many BAER team members used their own rules to determine parameter values for USGS regression and NRCS CN methods; therefore, there is no consistent way to estimate postfire peak flow. Many BAER road treatments for individual stream crossings were prescribed based on road/culvert surveys, without considering capacities of existing road structure and increased post-fire peak flow. For all regions, rolling dips/water bars, culvert upgrading, and ditch cleaning/armoring are the most frequently used road treatments. For Forest Service Regions 1 and 4, culvert upgrading is preferred, especially for fish-bearing streams. For Forest Service Region 3, culvert removal with temporary road closure and warning signs is preferred. Except for culverts, insufficient data is available on other road treatments to estimate their capacity and to evaluate their effectiveness.

A wood-strand material for wind erosion control: effects on total sediment loss, PM10 vertical flux, and PM10 loss.

Fugitive dust from eroding land poses risks to environmental quality and human health, and thus, is regulated nationally based on ambient air quality standards for particulate matter with mean aerodynamic diameter < or = 10 microm (PM10) established in the Clean Air Act. Agricultural straw has been widely used for rainfall-induced erosion control; however, its performance for wind erosion mitigation has been less studied, in part because straw is mobile at moderate wind velocities. A wood-based long-strand material has been developed for rainfall-induced erosion control and has shown operational promise for control of wind-induced erosion and dust emissions from disturbed sites. The purpose of this study was to evaluate the efficacy of both agricultural straw and wood-strand materials in controlling wind erosion and fugitive dust emissions under laboratory conditions. Wind tunnel tests were conducted to compare wood strands of several geometries to agricultural wheat straw and bare soil in terms of total sediment loss, PM10 vertical flux, and PM10 loss. Results indicate that the types of wood strands tested are stable at wind speeds of up to 18 m s(-1), while wheat straw is only stable at speeds of up to 6.5 m s(-1). Wood strands reduced total sediment loss and PM10 emissions by 90% as compared to bare soil across the range of wind speeds tested. Wheat straw did not reduce total sediment loss for the range of speeds tested, but did reduce PM10 emissions by 75% compared to a bare soil at wind speeds of up to 11 m s(-1).

Relative Effects on a Low-Volume Road System of Landslides Resulting from Episodic Storms in Northern Idaho

In late November to early December 1995 and February 1996, northern Idaho was hit by heavy rains on a deep snowpack, resulting in two flood and landslide events of historic magnitude. Each of these storms was larger than the previous significant storm, which occurred in January 1974. A study was initiated by the U.S. Department of Agriculture Forest Service to survey and study the effects of the resultant landslides on the Clearwater National Forest, including the effects on the aquatic ecosystem. The results of this study were compared with the estimated average natural sediment resulting from landslides to evaluate the incremental impacts of these recent episodic landslides. They were also compared with the results of a study conducted on the landslides resulting from the January 1974 storm to determine if the landscape was responding more severely to large storms as a result of Forest Service management activities over the past 21 years. The general results of this study indicate that, of the Forest Service management activities, roads are the major contributor; however, they contribute less sediment than natural landslides. The total resultant sediment appears to be within the transport capacity of the aquatic system, and the landslide response in 1974 was similar to the 1995–1996 response. The results of the aquatic ecosystem study were generally mixed, with some habitat parameters indicating degradation, some unchanged, and some improved as a result of the flooding or flooding with landslide sediment.

A comparison of three erosion control mulches on decommissioned forest road corridors in the northern Rocky Mountains, United States

This study tested the erosion mitigation effectiveness of agricultural straw and two wood-based mulches for four years on decommissioned forest roads. Plots were installed on the loosely consolidated, bare soil to measure sediment production, mulch cover, and plant regrowth. The experimental design was a repeated measures, randomized block on two soil types common in the northern Rocky Mountain area. The control produced the most sediment, while wood strands produced the least during the critical first winter following road decommissioning. Following the first year, there was no statistically significant difference in sediment production among the mulches or control. One year after the three mulches were applied, there was no statistical difference among mulch cover. Further, none of the mulches inhibited plant regrowth. The conservation implications of these research findings demonstrated that wood-based alternatives to agricultural straw were equally effective in reducing sediment production from originally bare, unvegetated soil strips resulting from forest road decommissioning. The amount of effective ground cover provided by mulch, plants, and litter appeared to be more important than the type of mulch.

LOCALLY AVAILABLE AGGREGATE AND SEDIMENT PRODUCTION

Selection of suitable locally available materials to build strong and durable roads with aggregate surfaces is desired to minimize road construction and maintenance costs and to minimize the detrimental effects of sedimentation. Eighteen aggregates were selected from local sources in Idaho, Oregon, South Dakota, and Washington State. Aggregate was placed in shallow metal frames and compacted to simulate a forest road. The levels of runoff and sediment from a highintensity, long-duration simulated rainstorm were measured. The material tests selected for use in the study included ones that define the basic characteristics of the aggregate, along with a number of tests intended to predict susceptibility to erosion. Each of the tests was statistically evaluated to identify those that best predicted the perceived aggregate quality. The two best indicators of aggregate quality were the results of the sand equivalent test and the P20 portion of the Oregon air degradation test. The best indicator of either runoff or sediment production was the fraction passing the 0.6-mm sieve. Acceptable aggregates, both those of good quality and those of marginal quality, exhibited a 2-order-of-magnitude range in both runoff and sediment production.

MODELING EROSION FROM UNPAVED FOREST ROADS AT VARIOUS LEVELS OF GEOMETRIC DETAIL USING THE WEPP MODEL

This study investigates the modeling question: “What level of geometric detail is necessary for modeling surface erosion from a forest road network?” We used the Water Erosion Prediction Project (WEPP) model to predict road sediment yield from a road network in a timber sale on the Boise National Forest, Idaho. Using three levels of detail (high, intermediate, and low), the outsloped 4.4 km road was divided into road segments (145, 91, and 42) of different lengths. The model predicted that 44, 44, and 42 metric tons per year of sediment would be produced from the 4.4 km road network based on the high, intermediate, and low detail methods, respectively. The model also predicted that a corresponding 0.036, 0.058, and 0.061 metric tons of sediment per year would be delivered into the stream system. Most sediment was predicted to be deposited in the buffer. For the predicted sediment produced from the roads, the total difference among the three methods was less than 5%, and regression analysis showed excellent correlation, with no practical differences among the three methods. For the quantity of sediment predicted to be delivered into the stream system from the road, the total difference among the methods was 61%, 67%, and 4% for high vs. intermediate, high vs. low, and intermediate vs. low, respectively. There were no strong correlations between the three methods, and the regression analysis indicated significant differences. More delivered sediment was predicted when using lower detail methods than when using higher detail methods. We conclude from the study results that the low detail method (i.e., considering only road grade reversal) can be used for input requirements of road traveled-way geometry, but that a high level of detail (i.e., dividing a road system into more road sections and considering buffer slope breaks and meanderings) is required for modeling the buffer slope geometry. Considering that the results were not sensitive to road traveled-way but were sensitive to buffer geometry, we believe it is advisable to divide a road network into road segments based on buffer geometry as opposed to road traveled-way geometry.

Infiltration, Erosion, and Vegetation Recovery Following Road Obliteration

Forest roads are obliterated to lower the risks of surface erosion and mass failures. One purpose of the road obliteration is to return the compacted forest roads to productive pre-road conditions, i.e., a forest floor with high infiltration capacity, low interrill erodibility, and high vegetation ground cover. It is important to know how these characteristics recover following road obliteration. Infiltration capacity, interrill erodibility, and vegetation ground cover are essential parameters for modeling erosion from obliterated roads for erosion prediction models such as the Water Erosion Prediction Project (WEPP). We chose three sites located on the Payette National Forest, Idaho. Rainfall simulations were conducted on 1 × 1 m plots with three replications in two consecutive years. Three 30 min storm events with an intensity of 89 mm h-1 were applied to each plot. Photos were taken to determine vegetation ground cover. Infiltration capacity and interrill erodibility in this study were determined as 9.0 mm h-1 for saturated hydraulic conductivity and 3.2 × 106 kg·s m-4 for interrill erodibility. This study postulated a history of saturated hydraulic conductivity on a forest road from prior to road building to years after obliteration. The low elevation (1400 m) site had vegetation ground cover of 27% after three years following road obliteration, while the other high elevation (1800 m and 2200 m) sites had 8% after four years. We conclude that four years was not sufficient time for obliterated roads to return to the pre-road (forest floor) conditions, especially for infiltration capacity.