Monitoring Road-Watershed Performance

1
Monitoring Road-Watershed Performance

• An Initiative for Efficient and Effective Road
  Performance Monitoring
• Combine effort to complete DSRs and INFRA to
  achieve road performance monitoring

mj furniss, PNW. 2005
2
Roads are a focus of watershed monitoring

• But roads vary greatly in performance
• Most do not fail
• Failures tend to cluster in areas of inherent
  instability

3
Why?

• Failure sites create a useful dataset for
  defining road performance through time
• Failures define the limits of practice in various
  landscape situations
• When experienced road managers retire,
  mission-critical knowledge could be conserved

4
Why?

• Little added effort for substantial value
  returned
• INFRA in place and working
• DSRs completed
• Related monitoring

5
What you get

• Ability to determine thresholds of performance
• Ability to determine relative risk of failure
• Quantitative description of risks

6
Failure Rate vs Distance from Stream
7
Failure Rate vs Slope Class
8
Slope Position vs Failure Rate
9
Geology and Failure Rate
10
Olympic National Forest
11
ONF Northwest District
12
Bonidu Cr.
13
Use Topograpy to Define Landscape Types for
Chi-square Analysis
Slopelt15, 15-30, 30-45, gt45Slope
Positionlt20, 20-55, 55-85, 85-100Distance
to Streamlt34m, 34-74m, 74-135m, lt135m
14
Example Landscape Units for ?2
15
Chi-Square Results
Landscape types with fewer failures than expected
were generally in gentler slope areas those at
lower slope positions and further from streams.
Types with more failures than expected were
generally at higher slope positions, steeper
slopes, and closer to streams.
16
A Need for More Specific Risk Information
Logistic Regression Modelling

• Combine 509 known failures with 1008 randomly
  selected locations.
• Use slope, slope position, and stream proximity
  to estimate relative risk of road-related
  landslides.

17
Logistic Regression Sample Units
18
Logistic Regression Modelln(odds) -1.8802
0.0238?Slope 0.0192?Slope Position
0.016?Distance 0.0001?Slope?Distance
19
Relative Odds of Road-Related Landslides
20
Relative Odds Compared to 2 Slope, 2 Slope
Position, 200m to Stream
21
Average Relative Odds by Watershed
22
Point swarms show problem areas clearly
23
How you get it

• Add DSR points and attributes to INFRA
• Attributes of failure type, cause, coarse
  magnitude

24
How you get it

• Modify description block in DSR to include
• ? Failure type
• ? Cause
• ? Volume (quantity classes)
• Total
• To stream
• To riparian area (within 50 m)

25
Cause AttributesQuestions

• Perpetrator or innocent bystander?
• Context
• Impact
• Sometimes roads catch and preventsediment
  delivery

26
Other road monitoring

• Use categories created in this effort for
  consistency and combined analysis