Post on 07-Feb-2020
Analysis Of Level Of Analysis: Study Designs For Dam Removal Monitoring
(in 20 minutes or less)
Desiree Tullos, George Pess,
Kelly Kibler, Mike McHenry
RRNW, February 2009
Types of monitoring1. Regulatory
2. Effectiveness
3. Research and river management
Focusing on research: The acid test
Conceptual Recovery
Knighton 1998
• Rates, patterns, extent
• Linking abiotic and biotic recovery
• Mechanisms of recovery– Movement of bed sediments– Change in food web– Habitat and refugia– Thermal regime
Dam removal to inform river management
Cui et al. (2008)
Presentation overviewLearning from dam removals
Unique learning
opportunity
Adequate funding
Rigorous study design
Prioritized monitoring
Rigorous study designs: Primary challenges
– Identifying and defining control
– Removal timelines
– Number and location of samples needed
– Mixed methods
Underwood 1994
Rigorous study design
Overcoming challenges: Principles for study “rigor”
1. Improve sampling design through definition of “impact” and explicit accounting of natural variation
2. Define objective rules for assessing uncertainty in the results
3. Standardization of data formats
4. Articulation of analytical models, underlying assumptions, and study deviations
Principle 1: Defining impact
1. Conceptualization: Potential outcomes and driving processes
2. Baseline assessment to focus study• Establishes common
set of data
• Identifies unique aspects worth studying
Pizutto 2002
Principle 1: Considering variability
Power analysis: 1-
Effect size Variability Number of samplesSignificance level Model
Principle 2: Reducing uncertainty
• Responsive to dam removal
• Easy to measure
• Low natural variability
0 5 10 15 20 25
Mean Substrate dia.
% Canopy Density
Residual Pool Area
% Sand + Fines
Bed Stability
Riparian Agr.
% Undercut Bank (visual)
% Pool Habitat (visual)
"RBP" Habitat Score
Signal:Noise Ratio(ratio of between-site variance/within-site variance)
EM
AP
-U
SE
PA
20
03
Principle 2: “Significance” isn’t only a statistical concept
• Statistics – Are observed changes outside natural variability of the system?
• Practicality – Are observed changes beyond our ability to measure them?
• Ecology – Are observed changes relevant to the ecosystem?
Principle 4: Deviations from study design
Bad timing
Humility
Inadequate resources
Optimization: designing for efficiency
Opportunistic learning at 2 case studies:
Brownsville and Elwha dam removals
Unique learning
opportunity
Brownsville Dam removal
Brownsville DamRiver: CalapooiaHeight: 5’Purpose: mill diversion, estheticsConstructed: 1960’sRemoval: 2007
Unique learning
opportunity
As a small case study, tests our limits of detection.
Conceptualization and hypothesis development
Conceptual model: Dam traps ~1.5 years of coarse sediment that will be released with dam removal
Baseline Assessment
1. Sediment Release Scenario– Volume of sediment stored =14,000cy, D50 = 59mm
– 1.5 years of sediment at normal winter flows
2. Downstream geomorphology– Response areas and types (Channel units, Bed material, Stream power and
channel competence calculations, review of historical aerial photos)
3. Sensitivity of downstream habitat and ecology– Characterized aquatic habitat, sampled benthic macroinvertebrates
Study Design: “controlled”
Practical Significance: Channel change
1250 1500 1750 2000
US
Aerial photos Field Surveys
DS1 to DS2 Dam
Reach Average Total Aerial Error Reach Average Total Field Error
-75
-60
-45
-30
-15
0
15
30
45
-1500 -1250 -1000 -750 -500 -250 0Ch
an
ge
in b
ar
wid
th a
t c
ros
s s
ec
tio
ns
(m
)
DS2 DS1 Dam
Distance from dam (m)
Ecological Significance: Grain size
DS1-Bar1US mean
-10.0
-5.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
1
D5
0 (
mm
)
Brownsville Dam Removal: Monitoring Sites
DS1 surface
DS1 subsurface
US surface
US subsurface
Steelhead spawning preference
Opportunistic Learning: Correcting conceptual models
2319.7
5.7
40
2427.4
0
5
10
15
20
25
30
35
40
45
50
Gr
ain
siz
e (
mm
)
Bars D50
Reach Pools2007 Pools08 Riffles07 Riffles08US 3 4 3 4DS1 1 3 3 4DS2 4 4 2 2
Lateral Pool
Lateral Pool
Riffle
RiffleGlide
GlideRiffle
RiffleGlide
Trench PoolRiffle
Riffle
Glide
Lateral Pool
Riffle
Glide
0
100
200
300
400
500
600
700
800
2007 2008
Le
ng
th (
m)
44.8
53.4
23.0
65.8
28.0
45.2
05
1015202530354045505560657075
Gr
ain
siz
e (
mm
)
Riffles D50
Opportunistic Learning: Investigating detectability
Many changes were within bounds of natural variability and bounds of measurement error
0
5
10
15
20
25
30
35
40
45
19
94
-1995
19
95
-1996
19
96
-1998
19
98
-2000
20
00
-2003
20
03
-2004
20
04
-2005
20
05
-2006
4/2
00
0-7
/2000
Field
Su
rvey
s
19
94
-1995
19
95
-1996
19
96
-1998
19
98
-2000
20
00
-2003
20
03
-2004
20
04
-2005
20
05
-2006
4/2
00
0-7
/2000
Field
Su
rvey
s
19
94
-1995
19
95
-1996
19
96
-1998
19
98
-2000
20
00
-2003
20
03
-2004
20
04
-2005
20
05
-2006
4/2
00
0-7
/2000
Field
Su
rvey
sT
ota
l Err
or
(m)
Average Total Identification Error (m) AverageTotal Position Error (m) Average Wetted Boundary Datum Error (m)
Change in Bar Width Change in Wetted Width Change in Wetted Width
Midpoint
Elwha dam removals
Elwha and Glines Canyon DamsRiver: ElwhaHeight: 108’ and 210’Purpose: hydropowerConstructed: 1913 and 1927Removal: 2012??
90%90 years
Conceptualization and hypothesis development
Conceptual Model: Removal of Elwha dams will promote the re-establishment of self-sustaining
anadromous salmonidpopulations within one to five
generations (2–30 yr) following dam removal
Unique learning
opportunity
Conceptualization and hypothesis development
Wo
od
wa
rd e
t a
l. 2
00
8
Baseline Assessment
1. Sediment Release Scenario– Volume of sediment stored =14,000,000 cy, coarse material
– 2-5 years of elevated sediment loads
2. Downstream geomorphology– Response areas and types (Channel units and floodplain channels, bed
material, review of historical aerial photos)
3. Sensitivity of downstream habitat and ecology– Aquatic habitat, periphyton, benthic macroinvertebrates, fisheries
Study design: “controlled”
Lower
Middle
Upper
Quinault
Sediment
Fish
Yes
Yes
No
No
Statistical significance
McH
enry
an
d P
ess
20
08
Practical significance
Assumption
Recovery rates of salmon are steady state and immediate following dam removal
Uncertainty
• Dam removal may initially cause populations to decline in short term
• Recolonization into some areas may take longer than expected
• Unanticipated barriers to migration may emerge
• Recolonization rates my change
McHenry and Pess 2008
Closing remarks
• Be smart about our analysis and detailed about our questions and hypotheses –BECAUSE EVERY DAM IS DIFFERENT
• Need baseline assessments and link it back to conceptual models
• Always consider and address your uncertainties – errors and stat. power
• Balance your statistics with your ecological significance
• Prioritize monitoring
Unique learning
opportunity
Adequate funding
Rigorous study design
Prioritized monitoring
Yes, every dam is differentBut…general process is the same
Acknowledgements
OSU InvestigatorsCara Walter, Jack Zunka, Trent Carmichael, and an army of undergrads
NWFSC InvestigatorsWatershed Program (EC) – Beechie, Coe, Kloehn, Kiffney, Liermann, Morley, Pess
Genetics & Evolution (CB) - Gary Winans
Estuarine & Ocean Ecology (FE) - Kurt Fresh
Migrational Behavior (FE) - Brian Burke, Kinsey Frick
CollaboratorsLower Elwha Tribe- Larry Ward, Doug Morrill, Mel Elofson, Sonny Sampson, Raymond Moses
NPS - Brian Winter, Jerry Freilich, Steve Acker, Pat Crain, Sam Brenkman
Bureau of Reclamation – Tim Randle
U.S. Fish & Wildlife Service - Roger Peters, Bob Wunderlich
U.S. Geological Survey - Jeff Duda, Pat Shafroth, Chris Konrad, Dave Woodson
University of Idaho - Chris Peery, Nancy Wright, Jeff Braatne
University of Washington - Bob Naiman, Tom Quinn
Peninsula College - Bill Eaton, Jack Ganzhorn, Dwight Barry
Western Washington University - Jim Allaway
University of Montana - Mark Lorang, Ric Hauer
WDFW - Anne Schaffer, Bill Freymond
SupportNOAA Fisheries – NWFSC
NOAA Fisheries – Restoration Center
NOAA – Open Rivers Initiative
USFWS – Coastal Puget Sound Program
USGS
National Fish and Wildlife Foundation
Oregon Watershed Enhancement Board