Post-Wildfire Landslide Hazards
1. Define conditions for fire-related debris-flow occurrence and develop a basin-scale model for debris-flow probability
In a study of the erosional response of 398 recently burned basins in the intermountain west, we found that not all basins produce debris flows; most burned watersheds respond to even heavy rainfall events by flooding. However, those watersheds that do produce destructive debris flows can be readily identified by a combination of geologic, topographic, and rainfall characteristics.
We performed a detailed statistical analyses of a database consisting of 35 independent variables that describe basin gradient, burn severity, soil properties, and storm rainfall conditions from basins that did, and did not, produce debris flows. Using logistic regression, we determined that the variables that best determine the probability of debris-flow occurrence are:
- The percent of area burned in each basin at both high and moderate severities,
- The average storm rainfall intensity,
- The percent of the basin with slopes great than or equal to 30%,
- The basin ruggedness,
- The percent of clay in the soil, and the soil liquid limit
These variables are part of a logistic statistical model that, when incorporated into a Geographical Information System (GIS), can be used to estimate the probability of post-fire debris flow from individual drainage basins.
An example of a map showing the probability of debris-flow occurrence from basins burned by the 2002 Missionary Ridge Fire near Durango, Colorado. Map generated by incorporating the logistic regression statistical model into a GIS.
Locations in western U.S. where data on basin response, basin gradient burn severity, and soils properties was collected. Open-File Report 2005-1218
2. Develop basin-scale models for estimating debris-flow peak discharge
An example of a map showing estimates of potential debris flow peak discharges from basins burned by the 2002 Missionary Ridge Fire near Durango, CO. Map generated by incorporation of the multi-variate statistical model into a GIS.
Data collected from our monitoring network and from the literature have provided the means to identify factors that strongly affect post-fire debris flow peak discharges, and to develop a multi-variate statistical model that can be used to estimate the potential peak discharge of individual drainage basins for a given storm. The analyses indicated that the area of the basin burned, the area of the basin with slopes greater than or equal to 30%, and the average storm rainfall intensity strongly affect debris-flow peak discharge.
3. Develop models for fire-related debris flow initiation processes
In our study of burned basins, we have found that the great majority of fire-related debris flows initiate by a process of progressive sediment bulking of storm runoff, rather than by the mobilization of discrete landslide failures (the most common mechanism in unburned terrains). In a study of the mechanics of fire-related debris-flow initiation at the Cerro Grande fire in New Mexico, we installed a series of sediment/runoff traps, established channel cross-sections and hillslope transects and a rain gage network in a severely burned 1st- order basin before the onset of the summer monsoon.
In addition, we made a series of detailed maps of debris-flow producing basins in different settings. We walked down basins, mapping features that documented the transition from clear water flow, to sediment-laden flow, to debris flow. By comparing and contrasting the features we mapped in the different settings we gained some ideas about the controls on this process (like sediment availability, contributing area, degree of channel confinement, and channel gradient).
These projects provided information on sediment and runoff delivery rates from burned hillsides, as well as data and information necessary to evaluate in a physical characterization of this process. Using this information, we are currently working on developing physical models of post-fire runoff, erosion, and debris-flow generation.
Field observations and measurements show that that most fire-related debris flows initiate from bulking of surface runoff with material eroded from hillslopes and from channels. Photo by John Moody, USGS
Configuration of monitoring array installed in basins burned by the Cerro Grande fire near Los Alamos, New Mexico. Data was collected through first summer monsoon season after each storm that impacted the area.
Detailed map of a debris-flow producing basin burned by the Sula Complex in 2000 in Montana. Evidence is shown of the progression from concentrated flow to the generation of debris flows. Features are somewhat exaggerated and schematized for illustrative purposes.
4. Evaluate the effects of wildfire on the stability of pre-existing landslide deposits
Boundary Peak landslide burned by the Dome fire in 1996 in Bandelier National Monument.
We also conducted a study to evaluate the effect of removal of vegetation by wildfires on the stability of pre-existing landslide deposits. Instrument sites consisting of continuously-recording soil-moisture probes installed at varying depths and recording rain gages were established for a comparison of the infiltration characteristics of burned and unburned sites on a deep-seated landslide deposit at Bandelier National Monument, New Mexico. A survey-quality GPS array was also installed to detect movement. This study showed that 1) throughout the summer monsoon season moisture infiltrates to greater depths at the burned site than at the unburned site, and 2) the soil at the burned site stays relatively moist throughout the summer, while the unburned site dries out rapidly with the onset of warm temperatures. These differences may be due to the lack of vegetation-induced transpiration at the burned sites. No movement was detected from the survey array over a period of three years. Further evaluation is necessary to determine how much increased infiltration could potentially destabilize the landslide deposits.
The USGS Landslide Hazard Program; the National Fire Plan Adaptive Management Program; CINDI, the Center for Integrated Disaster Information; and the Joint Fire Sciences Program have provided funding for these efforts.
