1. Better Characterizing Uncertainty in Geologic Paleoflood Analyses
Dean Ostenaa
Ostenaa Geologic LLC
October 24, 2017
Presentation for
2017 GSA Annual Meeting
Seattle, WA

2. Benefits of a Structured, Formal Approach to Estimating Uncertainty
Replaces adhoc and overly simplified uncertainty descriptions
“discharge uncertainty is ±25%”…………..based on the hydraulic model?
“stage uncertainty is ±1 m”………based on the field observation?
Adds transparency
Identifies dependencies within complex conclusions
Brings focus to critical strengths and weaknesses of interpretations and results
Geologic insights are essential to recognition of most epistemic uncertainties (dominant type of uncertainty for paleofloods)

3. Bring Geologic Perspectives into Engineering Analyses
As geologic information is brought forward into this practice, take the opportunity to bring geologic perspectives of how uncertainty related to that information should be conveyed
Geologic experience from conveying paleoseismic information for seismic hazard practice has potentially useful parallels and lessons
Precedents for event identification and site-to-site correlation
Stratigraphic completeness at an individual site
Uncertainties associated with age dating
Sedimentary processes and site-scale depositional/erosional models
Alternatives to continuous distributions for flood series
SSHAC process and models from UCERF3

4. Accepted Practice
Procedures for incorporation of paleoflood data in standard flood frequency estimates, used for flood insurance map estimates and most standard engineering practice, are in the final steps of approval for publication.

5. Uncertainty -Relationships and Dependencies
Topographic inputs affect everything
Resolution – both for hydraulics and for geomorphic characterization
Conveyance
Local hydraulics – is the geologic site likely to see erosion or deposition in a large flood
Slope – both local and reach scale
Stream Power
Stable through time?
Hydraulic models
Different issues for 1D and 2D models
PSI’s and Geomorphic Inputs
Do site characteristics that affect uncertainty change with increasing stage/discharge; or within the study reach?

6. Geologic Paleostage Indicators (PSIs)
Geologic PSI define the contrast between evidence of fluvial modification and evidence of landscape stability

7. PSI Variability and Uncertainty
Types of PSI and Sources of Uncertainty
Flotsam/botanic (tree scars) (Stage)
± depends on size, material, etc.
Could be emplaced above HWM
Koenig et al. (2016) has nice summary
Scour/trim lines (Stage + flow)
Soil vegetation resistance
Erosion criteria; Flow velocity/depth
Measurement uncertainty
Top vs bottom; variability along length
Sediment deposition (Stage + flow)
Always a minimum
Impacted by hydraulic setting (particle size)
Sediment supply
Modification of geomorphic surfaces (Stage + flow)
Erosion criteria
Topographic input resolution
Hydraulic model resolution
Reach scale natural variability
Hydraulic variability
Varies with sub-reach hydraulics
Smaller in backwater areas
Cross channel
Larger in through-flowing reaches
Dependent on stage/discharge?
Captured in hydraulic model outputs?
PSI location relative to hydraulic model cross section or calculation grid
Less in 1D models; more in 2D models
Depends on # of x-sections, topographic resolution, hydraulic model mesh size

8. Illustrating Uncertainties of PSI Water Surface Profiles
510
Elevation, in feet above arbitrary datum
509
508
Left bank HWMs Right bank HWMs
Average water surface
507
100
400
500
600
Distance upstream, in feet
Modified from Koenig et al., 2016

9. Some Initial Questions
Identify the Discharge Uncertainty Associated with Identified PSI
Some Initial Questions
Multiple PSI Locations?
Same Hydraulics at Locations?
Same Models at all Locations?
Yes No No No
Yes
Yes

10. Identify Factors Associated with Hydraulics
Add Branches for Single or Multiple Models as Needed
Hydraulic Model Choices
2D Model Choices
Add Branches for Single or Multiple Models as Needed
Parameter Set 2 (Roughness, etc.)
1D Model Choices
Model A Scenarios and Parameters
Parameter Set 2 (Roughness, etc.)

11. Identify Factors Associated with Characteristics of PSI
Add Branches for Values as Needed
Stage Shift for Location off X-Section
Alternate Values
PSI Stage A
PSI Stage B
PSI Stage C
Stage Uncertainty for PSI

12. Combine Results for Aggregate Discharge Uncertainty Distribution
Discharge A w/ Aleatory Uncertainty Q
PSI Stage A
PSI Stage B
PSI Stage C Q
Discharge B w/ Aleatory Uncertainty Q
Combine Distributions from Multiple Branches
Discharge C w/ Aleatory Uncertainty Q

13. PSI Hydraulics Combine Results
Hydraulic Model Choices
1D Model Choices
2D Model Choices
Add Branches for Single or Multiple Models as Needed
Model A Scenarios and Parameters
Parameter Set 2 (Roughness, etc.)
Hydraulics
Stage Shift for Location off X-Section
Stage Uncertainty for PSI
Alternate Values
Add Branches for Values as Needed
PSI Stage A
PSI Stage B
PSI Stage C
PSI
PSI Stage A
PSI Stage B
PSI Stage C
Discharge A w/ Aleatory Uncertainty
Discharge B w/ Aleatory Uncertainty
Discharge C w/ Aleatory Uncertainty Q
Combine Distributions from Multiple Branches
Combine Results

14. Better Approaches to Exploiting Geologic Data
Minimize Topographic Uncertainty
Lidar; SFM; various digital topographic tools
Exploit Spatial Information
Detailed geomorphic mapping at reach-scale
Maximize geomorphic resolution
Two-dimensional hydraulic modeling
Evaluate key spatial sub-areas defined by geomorphic mapping
Link individual deposits, geomorphic surfaces or landforms
Use multiple empirical relations/approaches as a quantitative definitions for deposition or erosion-based inundation limits
Criteria based on reach-specific soils and vegetation
Evaluate and develop separate uncertainty estimates for each geomorphic sub-area through multiple lines of evidence.
Combine multiple sample areas for global best estimates and uncertainties
Thresholds for deposition by floods are lower than thresholds for erosion
Not all sites of similar geomorphic age will have erosion/deposition at the same discharge

15. Stream Power and Big Lost River Criteria
From Ostenaa and O’Connell, 2005

16. Big Lost River Channel Stream Power Profile
From Ostenaa and O’Connell, 2005

17. Concluding Thoughts
Geologic paleoflood investigations can significantly extend flood history records
Can reduce uncertainty of discharge and probability estimates of infrequent events
Accepted in current practice and regulatory framework
Complimentary to other flood hazard estimation approaches
Structured and formal evaluations of uncertainty
Improve transparency and acceptance by public and regulatory stakeholders
Highlight dependencies and bring focus to strengths and limitations