Transcript Space, Relativity, and Uncertainty in Ecosystem Assessment

Space, Relativity, and Uncertainty in
Ecosystem Assessment of Everglades
Restoration Scenarios
Michael M. Fuller, Louis J. Gross,
Scott M. Duke-Sylvester, and Mark Palmer
Relative Assessment of Management Scenarios
Background and Need
General Approach to Scenario Comparison
Specific Example: Everglades Restudy Area
Multi-Species Approach
Methodology
Results of Analysis
Summary & Conclusions
Balancing the needs of multiple stakeholders
Policy decisions based on inputs from:
• Urban Resource Agencies (water districts, air quality depts, etc)
• Farmers and Industry (fisheries, timber companies, etc).
• Conservation Organizations (Audubon, NRDC, etc.)
• Governing Bodies (municipal, county, state, federal - politcal)
• Natural Area Managers (National Parks, etc)
Problem:
Natural systems are complex; Uncertainty is high
Solution: use models to:
Better understand natural systems.
Forecast future state of system.
Test sensitivity of system to different variables.
Models can have complex structure
EXAMPLE: SESI Model
Alligator & Everglades Hydrology
• Spatially Explicit Species Index
• Reproduction: nesting, mating
• Many parameters
• Many potential sources of error
• Several alternative hydrology plans
Scenario Analysis
Evaluate variation in input data or parameter settings
Investigate changes in conditions or environmental impacts.
Predict effects of alternative management plans.
Incorporate uncertainty into analyses using hypothetical data
SA used to rank alternative
management plans.
Scenario Analysis
Evaluate variation in input data or parameter settings
Investigate changes in conditions or environmental impacts.
PROJECT (forecast) effects of alternative management plans.
Incorporate uncertainty into analyses using hypothetical data
SA used to rank alternative
management plans.
SA assumes that interactions between model components, and
error propagation, is similar for different scenarios,
Uncertainty in how models respond to
alternative scenarios can expose policy
decisions to unexpected system behavior.
How can we reduce uncertainty?
Relative Assessment
A tool for testing the assumptions about model behavior.
Does rank order of management plan change with different
Input data or parameter scenarios?
Reduce uncertainty of model behavior
Comparison requires a specific assessment criterion
A metric for ranking alternatives in order of preference
Using relative assessment for Everglades Restoration
• Compare alternative hydrology plans under consideration.
• Vary the input data and parameter settings.
• Does rank order of alternatives change?
General Approach
• Scenario analysis used to rank alternatives
• Relative assessment to reduce uncertainty.
• Multi-species comparison.
Study System
Central and South Florida Project Comprehensive Review Study
Three natural areas in southern Florida:
• Everglades National Park
• Big Cypress
• Restudy area
Two management alternatives:
F2050 & D13R
F2050 = Baseline
Study System
4 Species + 2 wading bird guilds
Assessment Criterion
ATLSS Spatially Explicit Index Models
Species and Geographic Subregions
Sp ec ies Af fe cted
Am erica n a lligat o r
Sub reg io n Des cript io n
Wa ter Co ns erva tio n areas 3 A and 3 B
a n d Sn ail kite
Sn ail kite
Wa ter Co ns erva tio n areas 1 , 2A, 2 B
Am erica n al ligat o r
Sh ark Ri ve r, N E Sh ark Ri ve r, a n d Ta ylo r Slo u g h s
Wad in g Bird s
Cen tral Ro o keries
Wad in g Bird s
Sou th ern Ro o ke ries
Cape Sable se aside spar ro w
Co re ar ea
Cape Sable se aside spar ro w
Wes ter n ar ea
Cape Sable se aside spar ro w
Eas ter n ar ea
Wh ite - t ailed d eer
Everg lades Na tion al P ark
Wh ite - t ailed d eer
Big Cypress
Constructing Alligator Nest Parameter Scenarios
Average over space & time
Repeat for each hydrology scenario (F2050 & D13R)
9 nesting parameters
Baseline Values
Change values ±20 percent
T est Values
Hydrology
Scenario
30 years of
Historic Water Levels
SESI Model
Spatially Explicit
Habitat Index
(SESI)
Geographic
Subregion
x 100
Repeat for
each subregion
Average SESI Value
Constructing Climate I Data
All species groups.
Criterion: SESI Value
Difference between
F2050 & D13R
Average over space & time
30 y ear historic
base pattern
Select 5 wettest y ears
Random ly reorder selected y ears
Repeat 5 m ore
tim es
Com bine into 30
y ear sequence
Repeat 27 more
tim es
1
2
28
Hydrologic Plan
D13R
F2050
Calculate a mean SESI
value for each 30 y ear
sequence, for each
hy drologic plan, and
a global m ean for each
plan.
Calculate a difference
in global m eans.
Count
-0.0428
Long Wading Bird SESI
1
2
3
0.1366
0.1366
0.1367
0.1535
0.1534
0.1534
28
0.1366
0.1535
0.2291
0.2719
Grand Mean
Constructing Climate II Data
Comparison for all species groups.
Criterion: SESI Value
Difference between F2050 & D13R
D = F2050 - D13R
Baseline Data: 30 year historic pattern.
Experimental Data
Simulate 30 years of water level data.
1) Increase rainfall by 25 percent
2) Decrease rainfall by 25 percent
Summary of Scenarios and Treatments
Sce n ari o
Nes ting P ara m e ter
Ta xon
Allig a to r
Varia tio n
Clim a te C h a n g e I
Tre at m e nt
Para m e ters ran d o m ly ch an g ed i n t he ra n ge of ±20 % a n d
±3 0 % of s ta n dar d se ttin gs.
All 6 T ax a
Res am p ling a n d pe rm ut a tio n of h ist o rical clim a te d a ta to
cre at e 3 clim a te the m es: we t, dr y, a n d a ver age clim at e.
Clim a te C h a n g e II
All 6 T ax a
Baseli n e ± 2 5 per ce nt ra n d om varia tion in wa ter le ve ls.
Effect of Random Variation in Model Parameters
20 P ercent Variat ion
30 P ercent Variat ion
R1
D = F2050 - D13R
R1
40
40
30
30
20
20
10
10
R3
No rank reversals
.026
.028
.030
.032
Difference between Alternative Scenarios (SESI)
R1 = Water conservation areas 3A and 3B
R3 = Shark river, NE Shark river, and Taylor slough
.034
.036
.038
Climate Change Effects
Criterion: SESI Value
Difference between F2050 & D13R
For each species:
D>0
F2050
D<0
D13R
D = F2050 - D13R
Baseline
Increase level
Decrease level
D = 0 (No difference)
Effect of Simulated Climate on Scenario Ranking
White-Tailed
Deer
American
Alligator
Snail Kite
Cape Sable
Long-Legged
Seaside Sparrow Wading Bird
s
Difference in SESI (D13R - F2050 )
Short-Legged
Wading Bird
s
Effect of Simulated Climate Shift on Scenario Ranking
White-Tailed
Deer
American
Alligator
Snail Kite
Cape Sable
Seaside Sparrow
Difference in SESI (D13R - F2050 )
Long-Legged
Wading Birds
Short-Legged
Wading Birds
Summary of Effect on Rank Order of Hydrologic Scenarios
Tax o n
Sce n ario Type
Pe rce nt o f s ce n ario s
Pe rce n t o f s ce n ario s th at reve rse
th at ra nk
th e rank o rde r of F2050 & D13R
F2050 > D13R
rela tive t o ba s e s ce n ario .
Allig a to r
Š2 0 % Para m e ter Vari ation
None
None
Allig a to r
Š3 0 % Para m e ter Vari ation
None
None
All Ta xa
Clim a te C h a n g e I
54.17 %
25.93 %
All Ta xa
Clim a te C h a n g e II
53.71 %
30.55 %
American Alligator
Substantial variation in parameters = no reversals in rank order
All Species Groups
Substantial variation in rainfall pattern = reversals in rank order
Recap
General Approach
• Compare alternative hydrology plans.
• Randomly vary nesting parameters (alligator).
• Construct hypothetical climate patterns from historic data.
• Compare rank order of alternatives.
Results
• Alligator SESI not affected by parameter variation
• Change in climate patterns caused some rank reversals.
Summary
• Scenario analysis useful for ranking alternatives
• Relative assessment helps reduce uncertainty.