SlideShare una empresa de Scribd logo
1 de 23
Descargar para leer sin conexión
Fuzzy Multiple Criteria Evaluation of Conservation
Buffer Placement Strategies in Landscapes
Zeyuan Qiu, New Jersey Institute of Technology
Jin Zou, Kunming University of Science and Technology
Yang Kang, Columbia University
July 29, 2014
Backgrounds
 Conservation buffer restoration is among
the best management practices for
repairing impaired streams and restoring
ecosystem functions in degraded
watersheds.
 There are different buffer placement
strategies:
 Fixed-width riparian buffers: many existing
riparian protection rules and ordinances.
 Variable width riparian buffers (Phillips, 1989;
Xiang, 199; Herron and Hairsine, 1998;
Basnyat et al., 1999)
 Critical source areas (Bren, 2000; Tomer et al.,
2003; Doskey et al., 2006; Qiu, 2003 and
2009)
 Multiple criteria evaluation (Qiu et al., 2010)
Four Evaluation Criteria (Qiu,
2010)
 Hydrological Sensitivity measured by a
topographic index (TI)
 Benefits for controlling runoff generation
 Soil Erodibility measured by NRCS soil erodibility
index (Wischmeier and Smith, 1978)
 Benefits for reducing soil erosion
 Wildlife Habitat measured by the potential
presence of the species of concern as evaluated
by the New Jersey Landscape Project.
 Benefits for enhancing wildlife habitats
 Impervious Surface measured by the percentage
of impervious surface rate
 Benefits for mitigating stormwater impacts
T
RKLSEI 
Multiple Criteria Evaluation Framework
(Qiu, 2010)
 Following the simple additive utility function,
the total weighted score for strategy j
 where i represents the evaluation criteria and j the
buffer strategies, wi the criterion weights and aij
standardized criteria values
 cij is the strategy j’s class score for criterion i and
reflects the combined impacts of the criterion
weights and the criteria values (wiaij)
 Using the average criterion class scores of
the prioritized agricultural lands
 

I
i
ij
I
i
ijij cawC
11
Research Problems
 It is very critical to assign the proper
criterion weights that represent the
decision makers’ preferences;
 Using the class score to represent the
combined impacts of the criterion weights
and values as done in Qiu (2010) is overly
simplified;
 Deterministic way of eliciting the decision
makers’ preferences may not be consistent
with the nature of vagueness of decision
preferences.
Research Objectives
 To implement a fuzzy
analytic hierarchy
process (AHP) to illicit
the multiple criterion
weights that reflect
the vague nature of
the decision makers’
preferences for
placing conservation
buffer in agricultural
lands in Raritan River
Basin in New Jersey.
Evaluation Criterion Classification:
Hydrological Sensitivity
Evaluation Criterion Classification:
Soil Erodibility
Evaluation Criterion Classification:
Wildlife Habitat
Evaluation Criterion Classification:
Impervious Surface
Fuzzy Analytical Hierarchy
Process (AHP)
 An AHP survey was conducted in December 2012 to
the participants of the NJ SWCS Chapter Fall
Meeting to pairwise compare the decision criteria in
a scale of nine linguistic terms;
 Fifteen questionnaires were distributed and 15 were
returned. Thirteen correspondents indicated that
they are familiar or very familiar with conservation
buffers and their benefits;
 In addition to the standard comparison, we also ask
the respondent to rank his or her level of confidence
in giving such comparison statement in a Likert
scale of 1-5, which corresponds to five linguistic
label;
 In the end, the respondent was also asked to rank
the overall level of confidence in giving all
comparison statements in the 1-5 Likert scale.
Sample of Survey Questions
To control soil erosion is _____________ than to reduce surface runoff
and runoff-related pollutants.
 Absolutely more important
 Much more important
 Somewhat more important
 More important
 Equally important
 Less important
 Somewhat less important
 Much less important
 Absolutely less important
Please rank your level of confidence in giving the statement above at 1-
5 scale: ______
1 2 3 4 5
Confident Very
Confident
Absolutely
Confident
Somewhat
Confident
Not
Confident
Summary of the Survey Results
Questions Average Min. Max.
1a. To control soil erosion is ___ than to reduce surface runoff and runoff-related pollutants 4.27 4 6
1b. Please rank your level of confidence in giving the statement 4.15 3 5
2a. To control soil erosion is ___ than to enhance wildlife habitat 5.53 4 8
2b. Please rank your level of confidence in giving the statement 3.69 3 5
3a. To control soil erosion is ___ than to mitigate storm water impacts 4.60 2 8
3b. Please rank your level of confidence in giving the statement 3.85 3 5
4a. To reduce surface runoff and runoff-related pollutants is ___ than to enhance wildlife habitat 5.40 3 8
4b. Please rank your level of confidence in giving the statement 3.69 2 5
5a. To reduce surface runoff and runoff-related pollutants is ___ than to mitigate storm water
impacts
4.80 4 7
5b. Please rank your level of confidence in giving the statement 3.92 3 5
6a. To enhance wildlife habitat is ___ than to mitigate storm water impacts 3.00 0 5
6b. Please rank your level of confidence in giving the statement 3.69 3 5
7. Please rank your overall level of confidence in giving all statements 3.87 3 5
Borda Count for Fuzzy and
Linguistic Decision Making
 The Borda count was originally developed as a
single-winner election method in which voters rank
candidates in order of preference and gradually
evolved to be an appropriate procedure for group
decision making to select the most preferred
alternative when facing multiple decision
alternatives.
 The linguistic labels can be represented through
trapezoid fuzzy numbers (TFNs), which can capture
the vagueness of such linguistic assessments
Semantics With Nine Linguistic Labels
For Comparing Decision Criterion
Label Meaning TFN V(t)
l0 Absolutely less important (0, 0, 0, 0) 0.0000
l 1 Much less important (0, 0.02, 0.05, 0.11) 0.0417
l 2 Somewhat less important (0.05, 0.11, 0.17, 0.25) 0.1433
l 3 Less important (0.17, 0.25, 0.34, 0.44) 0.2983
l 4 Equally important (0.34, 0.44, 0.56, 0.66) 0.5000
l 5 More important (0.56, 0.66, 0.75, 0.83) 0.7017
l 6 Somewhat more important (0.75, 0.83, 0.89, 0.95) 0.8567
l 7 Much more important (0.89, 0.95, 0.98, 1) 0.9583
l 8 Absolutely more important (1, 1, 1, 1) 1.0000
Semantics with Five Linguistic Labels
for Ranking Confidence Level
Label Meaning TFN V(t)
d1 Not confident (0, 0.1, 0.2, 0.3) 0.15
d 2 Somewhat confident (0.2, 0.3, 0.4, 0.5) 0.35
d 3 Confident (0.4, 0.5, 0.7, 0.8) 0.60
d 4 Very confident (0.7, 0.8, 0.9, 1) 0.85
d 5 Absolutely confident (1, 1, 1, 1) 1.00
Broad Borda Counts


n
j
k
ij
k
ijik rcxr
1
)( 

m
k
ikki xrpxr
1
)()(
Where is i, j are the index for the decision criteria and k is
the index for the decision makers


n
j
k
ijik rxr
1
)( 

m
k
iki xrxr
1
)()(
Garcia-Lapresta et al. (2009)
Derived Criterion Weights
Control
Soil
Erodibility
Reduce
Runoff and
Runoff-
related
Pollutants
Enhance
Wildlife
Habitat
Mitigate
Stormwater
Impacts
Original Borda Count 0.199 0.212 0.331 0.258
Considering the
confidence in ranking
the pairs
0.201 0.214 0.323 0.261
Considering the
overall confidence
0.196 0.212 0.326 0.265
Comparison of Three MCDM
Scenarios
 MCDM-C: using the sum of the
criterion classes as proposed
originally by Qiu (2010)
 MCDM-O: using the original data
with the derived weights.
 MCDM-R: using the revised data
with the derived weights.
Comparison of the Areas Ranked
by Any Two Scenarios (%)
MCDM-C MCDM-O MCDM-R MCDM-C MCDM-O MCDM-R
Top 5% Top 10%
MCDM-C
MDCM-O 25.89 43.67
MCDM-R 40.25 55.18 49.05 60.80
Top 15% Top 20%
MCDM-C
MDCM-O 59.14 65.84
MCDM-R 52.17 43.70 49.18 35.12
Kappa Values for Comparing Any
Two Scenarios
Top 5% Top 10% Top 15% Top 20%
MCDM_C
vs
MCDM_O 0.256 0.432 0.586 0.652
MCDM_C
vs
MCDM_R 0.400 0.486 0.515 0.482
MCDM_O
vs
MCDM_R 0.550 0.635 0.430 0.340
Conclusion
 Fuzzy MCDM didn’t appear to be superior to
the simple classification-based MCDM
approach. There are tradeoffs among all
scientifically defensible MCDM alternative
approaches
 MCDM-C loses the originality of the raw data
 MCDM-O introduces the bias of the raw data
 MCDM-R is not superior in all cases
 All MCDM approaches give compatible
targeted areas for buffer restoration and
placement with the Kappa Value ranging
from 0.4 to 0.7.
 Always use not fancy, but understandable
approach
Acknowledgement
 USDA Natural Resources
Conservation Services CCPI
(Cooperative Conservation
Partnership Initiative) program
 USDA Forest Service National
Agroforestry Center

Más contenido relacionado

Similar a Fuzzy multiple criteria evaluation of conservation buffer placement

Draftvegetationprotocol2
Draftvegetationprotocol2Draftvegetationprotocol2
Draftvegetationprotocol2
John Marshall
 
15 Reasons Not to Use Percent Removals
15 Reasons Not to Use Percent Removals15 Reasons Not to Use Percent Removals
15 Reasons Not to Use Percent Removals
Marcus Quigley
 
Tang nccsc 2013 04 05
Tang nccsc 2013 04 05Tang nccsc 2013 04 05
Tang nccsc 2013 04 05
NC_CSC
 
4. Characterisation Approach
4. Characterisation Approach4. Characterisation Approach
4. Characterisation Approach
Environmental Protection Agency, Ireland
 
Birr - Identifying Critical Portions of the Landscape
Birr - Identifying Critical Portions of the LandscapeBirr - Identifying Critical Portions of the Landscape
Birr - Identifying Critical Portions of the Landscape
Jose A. Hernandez
 
Tokyo presentation august 15 2011
Tokyo presentation august 15 2011Tokyo presentation august 15 2011
Tokyo presentation august 15 2011
Tri-S Environmental
 
Juan Land Conservation Policies
Juan Land Conservation PoliciesJuan Land Conservation Policies
Juan Land Conservation Policies
a95osksj
 

Similar a Fuzzy multiple criteria evaluation of conservation buffer placement (20)

Evolution of M&E from TDA/SAP to Current Country Strategies: A Science and Po...
Evolution of M&E from TDA/SAP to Current Country Strategies: A Science and Po...Evolution of M&E from TDA/SAP to Current Country Strategies: A Science and Po...
Evolution of M&E from TDA/SAP to Current Country Strategies: A Science and Po...
 
Annals of Agricultural & Crop Sciences
Annals of Agricultural & Crop Sciences Annals of Agricultural & Crop Sciences
Annals of Agricultural & Crop Sciences
 
Deforestation diagnostics
Deforestation diagnosticsDeforestation diagnostics
Deforestation diagnostics
 
Draftvegetationprotocol2
Draftvegetationprotocol2Draftvegetationprotocol2
Draftvegetationprotocol2
 
Arvai Cc
Arvai CcArvai Cc
Arvai Cc
 
Targeting for diverse ecosystem zimmerman
Targeting for diverse ecosystem   zimmermanTargeting for diverse ecosystem   zimmerman
Targeting for diverse ecosystem zimmerman
 
15 Reasons Not to Use Percent Removals
15 Reasons Not to Use Percent Removals15 Reasons Not to Use Percent Removals
15 Reasons Not to Use Percent Removals
 
Tang nccsc 2013 04 05
Tang nccsc 2013 04 05Tang nccsc 2013 04 05
Tang nccsc 2013 04 05
 
Common Chemometric Indicators for Prediction of Soil Organic Matter Content a...
Common Chemometric Indicators for Prediction of Soil Organic Matter Content a...Common Chemometric Indicators for Prediction of Soil Organic Matter Content a...
Common Chemometric Indicators for Prediction of Soil Organic Matter Content a...
 
Tomer - Targeted Monitoring
Tomer - Targeted MonitoringTomer - Targeted Monitoring
Tomer - Targeted Monitoring
 
4. Characterisation Approach
4. Characterisation Approach4. Characterisation Approach
4. Characterisation Approach
 
Birr - Identifying Critical Portions of the Landscape
Birr - Identifying Critical Portions of the LandscapeBirr - Identifying Critical Portions of the Landscape
Birr - Identifying Critical Portions of the Landscape
 
Total Maximum Daily Load - Hawaii Status Report
Total Maximum Daily Load - Hawaii Status ReportTotal Maximum Daily Load - Hawaii Status Report
Total Maximum Daily Load - Hawaii Status Report
 
PHD -Kumar
PHD -Kumar PHD -Kumar
PHD -Kumar
 
Tokyo Presentation August 15 2011
Tokyo Presentation August 15 2011Tokyo Presentation August 15 2011
Tokyo Presentation August 15 2011
 
Tokyo presentation august 15 2011
Tokyo presentation august 15 2011Tokyo presentation august 15 2011
Tokyo presentation august 15 2011
 
Tokyo presentation august 15 2011
Tokyo presentation august 15 2011Tokyo presentation august 15 2011
Tokyo presentation august 15 2011
 
UBV_soutenance_v5
UBV_soutenance_v5UBV_soutenance_v5
UBV_soutenance_v5
 
Juan Land Conservation Policies
Juan Land Conservation PoliciesJuan Land Conservation Policies
Juan Land Conservation Policies
 
Using Novel Geospatial Tools and Approaches for Identifying Critical Nutrient...
Using Novel Geospatial Tools and Approaches for Identifying Critical Nutrient...Using Novel Geospatial Tools and Approaches for Identifying Critical Nutrient...
Using Novel Geospatial Tools and Approaches for Identifying Critical Nutrient...
 

Más de Soil and Water Conservation Society

Más de Soil and Water Conservation Society (20)

September 1 - 0939 - Catherine DeLong.pptx
September 1 - 0939 - Catherine DeLong.pptxSeptember 1 - 0939 - Catherine DeLong.pptx
September 1 - 0939 - Catherine DeLong.pptx
 
September 1 - 830 - Chris Hay
September 1 - 830 - Chris HaySeptember 1 - 830 - Chris Hay
September 1 - 830 - Chris Hay
 
August 31 - 0239 - Yuchuan Fan
August 31 - 0239 - Yuchuan FanAugust 31 - 0239 - Yuchuan Fan
August 31 - 0239 - Yuchuan Fan
 
August 31 - 0216 - Babak Dialameh
August 31 - 0216 - Babak DialamehAugust 31 - 0216 - Babak Dialameh
August 31 - 0216 - Babak Dialameh
 
August 31 - 0153 - San Simon
August 31 - 0153 - San SimonAugust 31 - 0153 - San Simon
August 31 - 0153 - San Simon
 
August 31 - 0130 - Chuck Brandel
August 31 - 0130 - Chuck BrandelAugust 31 - 0130 - Chuck Brandel
August 31 - 0130 - Chuck Brandel
 
September 1 - 1139 - Ainis Lagzdins
September 1 - 1139 - Ainis LagzdinsSeptember 1 - 1139 - Ainis Lagzdins
September 1 - 1139 - Ainis Lagzdins
 
September 1 - 1116 - David Whetter
September 1 - 1116 - David WhetterSeptember 1 - 1116 - David Whetter
September 1 - 1116 - David Whetter
 
September 1 - 1053 - Matt Helmers
September 1 - 1053 - Matt HelmersSeptember 1 - 1053 - Matt Helmers
September 1 - 1053 - Matt Helmers
 
September 1 - 1030 - Chandra Madramootoo
September 1 - 1030 - Chandra MadramootooSeptember 1 - 1030 - Chandra Madramootoo
September 1 - 1030 - Chandra Madramootoo
 
August 31 - 1139 - Mitchell Watkins
August 31 - 1139 - Mitchell WatkinsAugust 31 - 1139 - Mitchell Watkins
August 31 - 1139 - Mitchell Watkins
 
August 31 - 1116 - Shiv Prasher
August 31 - 1116 - Shiv PrasherAugust 31 - 1116 - Shiv Prasher
August 31 - 1116 - Shiv Prasher
 
August 31 - 1053 - Ehsan Ghane
August 31 - 1053 - Ehsan GhaneAugust 31 - 1053 - Ehsan Ghane
August 31 - 1053 - Ehsan Ghane
 
August 31 - 1030 - Joseph A. Bubcanec
August 31 - 1030 - Joseph A. BubcanecAugust 31 - 1030 - Joseph A. Bubcanec
August 31 - 1030 - Joseph A. Bubcanec
 
September 1 - 130 - McBride
September 1 - 130 - McBrideSeptember 1 - 130 - McBride
September 1 - 130 - McBride
 
September 1 - 0216 - Jessica D'Ambrosio
September 1 - 0216 - Jessica D'AmbrosioSeptember 1 - 0216 - Jessica D'Ambrosio
September 1 - 0216 - Jessica D'Ambrosio
 
September 1 - 0153 - Mike Pniewski
September 1 - 0153 - Mike PniewskiSeptember 1 - 0153 - Mike Pniewski
September 1 - 0153 - Mike Pniewski
 
September 1 - 0130 - Johnathan Witter
September 1 - 0130 - Johnathan WitterSeptember 1 - 0130 - Johnathan Witter
September 1 - 0130 - Johnathan Witter
 
August 31 - 1139 - Melisa Luymes
August 31 - 1139 - Melisa LuymesAugust 31 - 1139 - Melisa Luymes
August 31 - 1139 - Melisa Luymes
 
August 31 - 1116 - Hassam Moursi
August 31 - 1116 - Hassam MoursiAugust 31 - 1116 - Hassam Moursi
August 31 - 1116 - Hassam Moursi
 

Fuzzy multiple criteria evaluation of conservation buffer placement

  • 1. Fuzzy Multiple Criteria Evaluation of Conservation Buffer Placement Strategies in Landscapes Zeyuan Qiu, New Jersey Institute of Technology Jin Zou, Kunming University of Science and Technology Yang Kang, Columbia University July 29, 2014
  • 2. Backgrounds  Conservation buffer restoration is among the best management practices for repairing impaired streams and restoring ecosystem functions in degraded watersheds.  There are different buffer placement strategies:  Fixed-width riparian buffers: many existing riparian protection rules and ordinances.  Variable width riparian buffers (Phillips, 1989; Xiang, 199; Herron and Hairsine, 1998; Basnyat et al., 1999)  Critical source areas (Bren, 2000; Tomer et al., 2003; Doskey et al., 2006; Qiu, 2003 and 2009)  Multiple criteria evaluation (Qiu et al., 2010)
  • 3. Four Evaluation Criteria (Qiu, 2010)  Hydrological Sensitivity measured by a topographic index (TI)  Benefits for controlling runoff generation  Soil Erodibility measured by NRCS soil erodibility index (Wischmeier and Smith, 1978)  Benefits for reducing soil erosion  Wildlife Habitat measured by the potential presence of the species of concern as evaluated by the New Jersey Landscape Project.  Benefits for enhancing wildlife habitats  Impervious Surface measured by the percentage of impervious surface rate  Benefits for mitigating stormwater impacts T RKLSEI 
  • 4. Multiple Criteria Evaluation Framework (Qiu, 2010)  Following the simple additive utility function, the total weighted score for strategy j  where i represents the evaluation criteria and j the buffer strategies, wi the criterion weights and aij standardized criteria values  cij is the strategy j’s class score for criterion i and reflects the combined impacts of the criterion weights and the criteria values (wiaij)  Using the average criterion class scores of the prioritized agricultural lands    I i ij I i ijij cawC 11
  • 5. Research Problems  It is very critical to assign the proper criterion weights that represent the decision makers’ preferences;  Using the class score to represent the combined impacts of the criterion weights and values as done in Qiu (2010) is overly simplified;  Deterministic way of eliciting the decision makers’ preferences may not be consistent with the nature of vagueness of decision preferences.
  • 6. Research Objectives  To implement a fuzzy analytic hierarchy process (AHP) to illicit the multiple criterion weights that reflect the vague nature of the decision makers’ preferences for placing conservation buffer in agricultural lands in Raritan River Basin in New Jersey.
  • 11. Fuzzy Analytical Hierarchy Process (AHP)  An AHP survey was conducted in December 2012 to the participants of the NJ SWCS Chapter Fall Meeting to pairwise compare the decision criteria in a scale of nine linguistic terms;  Fifteen questionnaires were distributed and 15 were returned. Thirteen correspondents indicated that they are familiar or very familiar with conservation buffers and their benefits;  In addition to the standard comparison, we also ask the respondent to rank his or her level of confidence in giving such comparison statement in a Likert scale of 1-5, which corresponds to five linguistic label;  In the end, the respondent was also asked to rank the overall level of confidence in giving all comparison statements in the 1-5 Likert scale.
  • 12. Sample of Survey Questions To control soil erosion is _____________ than to reduce surface runoff and runoff-related pollutants.  Absolutely more important  Much more important  Somewhat more important  More important  Equally important  Less important  Somewhat less important  Much less important  Absolutely less important Please rank your level of confidence in giving the statement above at 1- 5 scale: ______ 1 2 3 4 5 Confident Very Confident Absolutely Confident Somewhat Confident Not Confident
  • 13. Summary of the Survey Results Questions Average Min. Max. 1a. To control soil erosion is ___ than to reduce surface runoff and runoff-related pollutants 4.27 4 6 1b. Please rank your level of confidence in giving the statement 4.15 3 5 2a. To control soil erosion is ___ than to enhance wildlife habitat 5.53 4 8 2b. Please rank your level of confidence in giving the statement 3.69 3 5 3a. To control soil erosion is ___ than to mitigate storm water impacts 4.60 2 8 3b. Please rank your level of confidence in giving the statement 3.85 3 5 4a. To reduce surface runoff and runoff-related pollutants is ___ than to enhance wildlife habitat 5.40 3 8 4b. Please rank your level of confidence in giving the statement 3.69 2 5 5a. To reduce surface runoff and runoff-related pollutants is ___ than to mitigate storm water impacts 4.80 4 7 5b. Please rank your level of confidence in giving the statement 3.92 3 5 6a. To enhance wildlife habitat is ___ than to mitigate storm water impacts 3.00 0 5 6b. Please rank your level of confidence in giving the statement 3.69 3 5 7. Please rank your overall level of confidence in giving all statements 3.87 3 5
  • 14. Borda Count for Fuzzy and Linguistic Decision Making  The Borda count was originally developed as a single-winner election method in which voters rank candidates in order of preference and gradually evolved to be an appropriate procedure for group decision making to select the most preferred alternative when facing multiple decision alternatives.  The linguistic labels can be represented through trapezoid fuzzy numbers (TFNs), which can capture the vagueness of such linguistic assessments
  • 15. Semantics With Nine Linguistic Labels For Comparing Decision Criterion Label Meaning TFN V(t) l0 Absolutely less important (0, 0, 0, 0) 0.0000 l 1 Much less important (0, 0.02, 0.05, 0.11) 0.0417 l 2 Somewhat less important (0.05, 0.11, 0.17, 0.25) 0.1433 l 3 Less important (0.17, 0.25, 0.34, 0.44) 0.2983 l 4 Equally important (0.34, 0.44, 0.56, 0.66) 0.5000 l 5 More important (0.56, 0.66, 0.75, 0.83) 0.7017 l 6 Somewhat more important (0.75, 0.83, 0.89, 0.95) 0.8567 l 7 Much more important (0.89, 0.95, 0.98, 1) 0.9583 l 8 Absolutely more important (1, 1, 1, 1) 1.0000
  • 16. Semantics with Five Linguistic Labels for Ranking Confidence Level Label Meaning TFN V(t) d1 Not confident (0, 0.1, 0.2, 0.3) 0.15 d 2 Somewhat confident (0.2, 0.3, 0.4, 0.5) 0.35 d 3 Confident (0.4, 0.5, 0.7, 0.8) 0.60 d 4 Very confident (0.7, 0.8, 0.9, 1) 0.85 d 5 Absolutely confident (1, 1, 1, 1) 1.00
  • 17. Broad Borda Counts   n j k ij k ijik rcxr 1 )(   m k ikki xrpxr 1 )()( Where is i, j are the index for the decision criteria and k is the index for the decision makers   n j k ijik rxr 1 )(   m k iki xrxr 1 )()( Garcia-Lapresta et al. (2009)
  • 18. Derived Criterion Weights Control Soil Erodibility Reduce Runoff and Runoff- related Pollutants Enhance Wildlife Habitat Mitigate Stormwater Impacts Original Borda Count 0.199 0.212 0.331 0.258 Considering the confidence in ranking the pairs 0.201 0.214 0.323 0.261 Considering the overall confidence 0.196 0.212 0.326 0.265
  • 19. Comparison of Three MCDM Scenarios  MCDM-C: using the sum of the criterion classes as proposed originally by Qiu (2010)  MCDM-O: using the original data with the derived weights.  MCDM-R: using the revised data with the derived weights.
  • 20. Comparison of the Areas Ranked by Any Two Scenarios (%) MCDM-C MCDM-O MCDM-R MCDM-C MCDM-O MCDM-R Top 5% Top 10% MCDM-C MDCM-O 25.89 43.67 MCDM-R 40.25 55.18 49.05 60.80 Top 15% Top 20% MCDM-C MDCM-O 59.14 65.84 MCDM-R 52.17 43.70 49.18 35.12
  • 21. Kappa Values for Comparing Any Two Scenarios Top 5% Top 10% Top 15% Top 20% MCDM_C vs MCDM_O 0.256 0.432 0.586 0.652 MCDM_C vs MCDM_R 0.400 0.486 0.515 0.482 MCDM_O vs MCDM_R 0.550 0.635 0.430 0.340
  • 22. Conclusion  Fuzzy MCDM didn’t appear to be superior to the simple classification-based MCDM approach. There are tradeoffs among all scientifically defensible MCDM alternative approaches  MCDM-C loses the originality of the raw data  MCDM-O introduces the bias of the raw data  MCDM-R is not superior in all cases  All MCDM approaches give compatible targeted areas for buffer restoration and placement with the Kappa Value ranging from 0.4 to 0.7.  Always use not fancy, but understandable approach
  • 23. Acknowledgement  USDA Natural Resources Conservation Services CCPI (Cooperative Conservation Partnership Initiative) program  USDA Forest Service National Agroforestry Center