Development of a Landslide and Debris Flow Hazard Map for the Old and Grand Prix Fires: San Bernardino National Forest

1
Development of a Landslide and Debris Flow Hazard
Map for the Old and Grand Prix FiresSan
Bernardino National Forest

• Juan de la Fuente
• John Chatoian
• Allen P. King
• Christy B. Till
• Alisha R. Miller
• Robert G. Taylor
• USDA Forest Service

2
Outline

• I. Background
• Chronology
• Context of Project
• Project Goals
• II. Methods
• Remote Sensing Data
• Air Photo Field Assessments 2003-2005 Storm
  Effects
• Air Photo Assessment 1969-1972 Storm Effects
• Literature Review
• Data Analysis
• Hazard Map Development
• III. Summary of Key Slope Processes and Watershed
  Response
• Key Slope Processes After the Fires Painting a
  Picture
• Observations of Watershed Response
• IV. Findings of This Study
• USFS Mapping on 1972 and 2005 Air Photos
• USGS mapping on 2004 Air Photos
• IV. Hazard Map
• Elements

3
I. Background

• Chronology
• Fall 2003- Old Grand Prix Fires occurred
  Assessed by Burned Area Emergency Rehabilitation
  (BAER) teams
• December 2003- Moderate storms, widespread debris
  flows, fatalities BAER team re-convened in
  January 2004 and recommended Debris Flow Hazard
  Map
• January 2005- BAER Team mobilized again to
  evaluate effects of 2005 storms
• March 2005- USFS, NRCS, CGS, USGS personnel met
  in Davis, CA to coordinate work in post fire
  assessments and to discuss USFS hazard map for
  Old Grand Prix Fires
• April 2005-September 2005- New air photos
  acquired, field investigation, photo mapping, GIS
  coverages, analysis, consultation with
  researchers, development of Hazard map.
• Context
• USGS conducting Southern California Debris Flow
  Hazard assessment (Sue Cannon)-
• USFS collaborating with agencies doing landslide
  work in California (USGS, CGS, NRCS. FEMA,
  Counties)
• Will be linked to Southern California USFS
  geomorphic mapping
• Goals
• 1. Determine extent of post-fire landslides
  Debris Flows for Old Grand Prix Fires
• 2. Investigate the response of watersheds in the
  Old Grand Prix Fire area to previous storms
• 3. Summarize current knowledge of post-fire
  response in Old Grand Prix Fire areas
• 4. Explore possible links between bedrock, slope,
  vegetation, land management with fire and
  watershed response
• 5. Develop prototype hazard map for landslides
  and debris flows Utilize sound, state of the art
  techniques, and results of recent studies make
  the product simple and easy to use
• Not to evaluate BAER post-fire mitigation
  measures (see USFS monitoring reports)
• Not to provide site specific assessments or
  mitigations

4
(No Transcript)
5
II. Methods (Goals 1-2)

• Air Photo Field Assessments 2003-2005 Storm
  Effects
• The Forest Service mapped altered channels and
  landslides on 140,00 Color Infrared air photos
  taken in April, 2005. The term Altered Channel
  is used here to identify those channels which
  experienced sufficient disruption of the channel
  bed (scour, deposition, or removal of riparian
  vegetation) to be visible on air photos, and
  distinguished from those in which the bed was not
  mobilized. Some altered channels experienced
  debris flows, some hyperconcentrated flows, and
  some simply mobilization of the bed due to high
  flows. Distinctions between the three were not
  discernible on the air photos. It identified
  features associated with the first two winters
  after the storm, and covered 140,000 acres
  burned during the 2003 Old Grand Prix Fires on
  the San Bernardino National Forest. The mapped
  area completely overlaps 1972 (USFS) and 2004
  (USGS)mapping.
• USFS 2005 Air Photo Mapping- New air photos
  acquired (April 2005) Landslides altered
  channels mapped Limited field verification
• USFS Field Work- Field investigations conducted
  in May, 2005 to calibrate air photo mapping and
  interpretations
• USGS 2004 Air Photo Mapping- USGS is mapping
  landslides and altered channels on 112,000 color
  air photos taken in February and March 2004. Will
  identify 2003-2004 features. Mapping criteria are
  similar to those being applied by the Forest
  Service, and this should facilitate comparison of
  findings. About 100,000 acres of the burn will
  be mapped, focusing on urban and urban interface
  lands. Scheduled completion in 2006.
• USGS Field Work- Field investigations from
  2003-2005 done as part of debris flow hazard
  assessments
• Air Photo Assessment Previous Storm Effects
  (1969)
• USFS 1972 Air Photo Mapping- 1972 air photos
  similarly mapped in a sample area within the Old
  Grand Prix Fire (about 21,000 acres Warm East
  Twin Cr). Time period selected to show effects
  of 1969 storms. Part of mapped area burned in
  1970.

6
II. Methods (Goals 3-4)

• Review of Past Current Work
• Previous Work- A literature review was conducted,
  and relevant articles summarized
• Current Work- Consultation With USGS, CGS,
  County, and USFS geologists
• Data Compilation Analysis-
• Air photo Interpretation Done on overlays,
  ocular transfer to 7 ½ minute topo maps,
  digitized, GIS coverage developed entered into
  geodatabase
• GIS Layer Compilation- Bedrock, 2 meter DEM
  (Slope Gradient, Elevation, Hillshade, 3-D
  Images), Burn Severity, Soil Slip Susceptibility,
  Watershed Response, Vegetation, Landslides
  Altered Channels
• Data Analysis- GIS overlays data base queries
  done to determine density of 2005 landslides and
  altered channels for different slope classes,
  burn intensities, bedrock types, etc. Statistical
  analysis was done on 1972 air photo data
  exploring these same relationships. 3-D images
  examined to evaluate utility for terrain
  stratification
• Remote Sensing Tools Used in Analysis
• 1972 Air Photos- 116,000 Color (mapped
  Landslides and altered channels)
• 2003 Post Fire Imagery and Digital Orthophoto-
  Identified pre-fire landslides altered channels
• 2004 Air Photos- (USGS) 112,000 Color (mapped
  Landslides and altered channels)
• 2005 Air Photos- (USFS) 140,000 Color Infrared
  Plus 2005 Digital Orthophoto (landslides and
  altered channels)
• Digital Elevation Models- (DEMs)- 10 meter, 3
  meter, 2 meter (slope and elevation, stream
  network, 3-Dimensional Imaging Program)

7
II. Methods (Goal 5)

• Hazard Map Development
• Hazard maps were developed to address the
  following Processes
• Debris Flow-
• Debris Slide and Rock Fall-
• Deep Seated Landslide-
• The Following Questions Guided Development of the
  Hazard Maps
• What happened after 2003 fires, where, how, what
  were the effects?
• What happened with first storms after the fires
  how did burned versus unburned land behave? How
  did burned areas respond to subsequent storms?
• What happened in this area during earlier storms
  and fires over the past 50 years?
• What and where are sensitive lands where
  landslides, rock falls, a of debris flows
  occurred?
• What tools are best suited to efficiently and
  systematically delineate these sensitive and
  hazardous lands?

8
III. Summary of Previous Work

• Key Post-Fire Slope Processes Review of BAER
  Team findings, published studies, consultation
  with others, and air photo field work yielded
  the following summary of dominant slope processes
  in the Old Grand Prix Fire areas.
• The fire was followed by the following sequence
  of events and processes
• Removal of Organic Material- The fires removed
  organic ground cover exposed soils to rain drop
  impact loss of organic litter, plants, trees,
  and shrubs removed mechanical support
• Dry Ravel- Loss of mechanical support provided by
  ogranic material on steep slopes facilitated dry
  ravel following fire this process was documented
  in BAER team reports.
• Soil Water Repellancy- Repellancey developed on
  about half of the grass/chaparral lands (BAER
  Report 2003), and probably accentuated runoff
• Root Support- Root support was lost from
  fire-killed vegetation effects were likely not
  felt until the second winter after the fires
  (time lag for delay).
• Channel Material- Channels were filled with dry
  ravel immediately after the fires
• Storms of December 2003- The storm around
  Christmas of 2003 acted on water repellant
  slopes, channels filled with dry ravel, and
  barren soil, producing widespread debris flows
• Source of Debris Flow Material- Channels were
  charged with dry ravel, barren steep hillslopes
  effeciently delivered material by surface
  erosion, rilling and gullying. Some studies
  (Moody 2001) indicate that for non debris slide
  debris flows, about 80 of the sediment generated
  comes from the channel, and 20 from the
  hillslope in Colorado.
• Subsequent Storms- Once a channel network
  experiences a debris flow, and rills strip soil
  from steep hillslopes, the likelihood of of a
  subsequent debris flow in the same channel
  network is much lower Exceptions could occur
  where large landslides develop on channel banks,
  or where significant tributaries within the
  network do not experience the initial debris flow
  episode.
• Debris Replenishment Rate in Channels- The
  replenishment of debris in channels, and in
  gullies on hillslopes occurs over time, but we
  dont have a good picture of how rapidly this
  occurs
• Revegetation- Revegetation effectively stops dry
  ravel and vulnerability of rills and gullies to
  further erosion.
• Loss of Water Repellancy- Fire-associated
  repellancy decreases over time
• Factors Affecting Post-Fire Debris Flows- Key
  factors appear to be time since last fire, time
  since last debris flow, and nature of vegetation
  burned in fire (grass versus chaparral versus oak
  woodland versus conifer stand).
• Debris Flow Risk in 2005- Potential for debris
  flows has apparently decreased since the fall of
  2003 due to a) revegetation (stops dry ravel,
  adds ground cover, replaces lost root support)
  b) Reduction in soil water repellancy c)
  Reduction of debris available in channels and on
  steep hillslopes brought about by the post-fire
  dry ravel, rilling and debris flows. Over time,
  the available debris will increase once again.

9
III. Summary of Previous Work

• Observed Post-Fire Storm Responses From
  Discussions With BAER Teams and USGS Personnel-
• Virtually all of the steep, basins within the Old
  and Grand Prix fire area meeting certain burn
  severity and slope gradient criteria, experienced
  debris flows in response to the relatively low
  intensity storms in December of 2003.
• Models addressing post 2003 fire debris flow
  potential did good job at predicting which basins
  would experience debris flows.
• Most debris flows did not have landslide sources
  
• Large channels which did not experience debris
  flows did experience mobilization of bed material
  and local realignment, resulting in damage to
  facilities.
• Some relatively small drainages (150 acres)
  experienced sizeable debris flows which damaged
  property (eg. Greenwood avenue).
• Most watersheds which experienced debris flows in
  the December 2003 storms did not exhibit repeat
  debris flows in the larger storms of 2004 and
  2005, but rather experienced sediment-laden flood
  flows.
• There were fewer landslide sources for debris
  flows following the December storm than in storms
  the second year (Sue Cannon, personal
  communication, 2005).
• Abundant shallow debris slides (soil slips)
  occurred in the foothills at lower elevations the
  second winter following the fire (2004-2005).
• Debris flows traveling down some alluvial fans
  followed streets, and were confined by block
  walls, allowing them to travel much further than
  they would have on an unaltered natural fan.
• Fatalities were concentrated in alluvial fans at
  mouths of canyons and narrow terraces in confined
  channels.

10
IV. Findings

• Results of 2005 Air Photo Mapping (USFS)
• .
• SUMMARY
• Altered channels and landslides are clearly much
  denser in burned areas than in unburned to a
  lesser degree, they were also denser in
  high/moderate burn severity areas than in low
  severity.
• Shallow debris slides were concentrated in lower
  elevation foothills in granitic rock, and
  densities in high/moderate severity burn areas
  were higher than in low severity.
• There were a small number of reactivated deep
  seated landslides producing debris slides and
  flows.
• Altered channels extend long distances down some
  alluvial fans into lower gradient areas.
• Some bedrock units exhibited elevated landslide
  densities
• Small alluvial terraces within large confined
  canyons appear to be particularly susceptible to
  inundation in many areas, likely involving debris
  flow processes.
• Field verified debris flows and air photo
  identified altered channels occurred with few
  exceptions within the channel network identified
  by the 2 meter DEM (1/2 acre contributing area).
  Debris slides occurred on steep slopes (within a
  relatively narrow slope gradient range when
  classified with 2 meter and 10 meter DEMS).
  There may be some useful links between post-fire
  response and elevation plus pre-fire vegetation.
• Revegetation which occurred from 2003-2005
  (primarily grass) concealed some small features
  on air photos

11
(No Transcript)
12
IV. Findings

• Results of 1972 Air Photo Mapping (USFS)
• The Forest Service mapped approximately 21,000
  acres in the Warm East Twin Creek watersheds
  inside the boundaries of the Old Fire of 2003.
  Altered channels and landslides associated with
  previous storm years (eg. 1969) were mapped on
  color, 116,00 air photos taken in 1972. The
  mapping area included a 7,000 acre fire which
  burned in 1970.
• Majority of features (landslides and altered
  channels) are found near roads or in the 1970
  burned area.
• State Highway 18 runs through the 1972 mapped
  area and appears to have influenced the density
  of features outside the 1970 fire perimeter.
• Majority of the landslides occur in the
  Cretaceous plutonic bedrock (Kcc Monzogranite
  of City Creek or KJc Quartz monzodiorite of
  Crestline), not the Quaternary sedimentary units.
• Landslide features are all in one of two land
  use/vegetation classes chaparral or forest.
• All landslide features occur on 30º-60º slopes
  (majority on slopes between 30º-45º).
• All landslide features are on south facing slopes
  (90º-270º aspect).
• Features within the 1970 burned area show a
  statistical correlation between feature type
  elevation, feature type slope, and feature type
  the presence of a debris flow.
• Bedrock type, slope, aspect, burned areas and
  roads appear to be the key factors controlling
  the location of features.

13
IV. Findings

• Status of USGS Mapping (2004 Air Photos)
• Scheduled for completion 2006 Preliminary
  Observations
• Altered channel density is considerably higher in
  burned relative to unburned areas.
• There is a marked greening of slopes (germination
  of grass) on the March photographs (compared to
  February), which may mask some evidence of
  altered channels and dry ravel.
• There were fewer landslide sources associated
  with debris flows which occurred the first year
  after the fire than in the second year.
• Status of 3-Dimensional Imaging
• Preliminary stratification of watersheds was
  conducted with the 3-D imaging tool available for
  this project. Promising tool, for doing
  identified by slope, dissection, and roughness.
  but came too late in the process to fullyl
  integrate into the Hazard Map development.
• Status of 2005 Digital Orthophoto-
• Map became available at end of project.
• Will use to refine location of landslides and
  altered channels

14
(No Transcript)
15
IV. Hazard Map

• Development of Hazard Classes
• Hazard Classes Susceptibility to debris flows
  and landslides was characterized in three simple
  categories (low, medium or high). Classification
  was based on previous work in the vicinity and
  similar areas, and on the findings of the air
  photo inventories.
• Debris Flow Hazard
• Slope Gradient
• Channelization
• Landform
• Altered channels were assumed to have high
  potential for debris flow
• Altered channels were mostly coincident with
  stream network (2 meter DEM 0.5 acre area) or on
  steep slopes
• Altered Channels on alluvial fans and terraces in
  confined channels
• Therefore All above characteristics assigned
  high hazard for debris flow
• Debris Slide and Rock Fall Hazard
• Slope Gradient
• Landform
• No data available for runout or downslope
  effects on flats
• Debris slides on steep slopes (90 on slopes gt
  40 and 63 on slopes gt 65.
• Limited correlation of 2005 slides to bedrock,
  veg etc.

16
IV. Hazard Map

• Planned Use of Maps
• U.S. Forest Service Use
• Post-fire assessments
• Project Planning for Prescribed Fire, etc.
• TMDLs
• County
• Planning for Future Urbanization
• Disaster preparedness
• Limitations of Maps
• Not site specific
• Post-Fire for only one large fire event is
  analyzed
• Future Updates to Map
• Preliminary Hazard Map in December 2005
• Updates in subsequent years
• Will be done in conjunction with geomorphic
  mapping, and acquisition of high resolution DEMs

17
V. Conclusions

• .
• Determining Extent and Nature of Storm Effects-
  Area-wide air photos or similar imagery, are
  essential to document post fire response. Color
  infrared air photos at a scale of 140,000 scale
  provide a good view of effects over the entire
  burn area.
• DEM- A high resolution Digital Elevation Model is
  essential in Southern California to development
  of a debris flow and landslide hazard map.
  Slope gradient and location of the channel
  network appear to be the best indicators of
  debris flow and debris slide potential.
• Past Response- Understanding of previous
  landscape response to severe storms and fires is
  essential. It must be recognized that the
  pattern of landslides and altered channels
  identified by this study following the 2003 fires
  may not be representative of responses to future
  storms.
• Bedrock, Geomorphic and Soils Mapping- Detailed
  bedrock, geomorphic, and soil mapping may shed
  light on debris slide and debris flow occurrence,
  particularly when examined in concert with slope
  gradient and other factors.
• Acknowledgements
• The following persons provided valuable advice
  and comments
• Sue Cannon USGS, Golden Colorado
• Dave Lidke USGS, Golden, Colorado
• Bill Hicks USFS Retired
• Mark Reid USGS, Menlo Park
• Steve Ellen USGS, Retired
• Jim Mckean USFS Boise Idaho
• Tim McCrink CGS Sacramento, California
• The following persons assisted in photo mapping,
  digitizing, and GIS analysis