SlideShare a Scribd company logo
1 of 17
www.helsinki.fi/yliopisto
Assessment of Scenario Generation
Approaches for Forest Management
Planning through Stochastic
Programming
Kyle Eyvindson and Annika Kangas
2.2.2015Kyle Eyvindson 1
www.helsinki.fi/yliopisto
• Aim is to:
• Integrate uncertainty into the development of forest
plans
‒ Inventory, growth models, climate change...
• Produce a robust solution which meets the demands of
the decision maker(s), and can accommodate
preferences towards risks
• One method is through stochastic programming
‒ issues of tractability can become an issue
2.2.2015 2Kyle Eyvindson
Introduction
www.helsinki.fi/yliopisto
• Mathematical optimization where some parameters are
uncertain.
• Depending on the structure of the problem, different
problem formulation alternatives are available
‒ simple recourse
‒ two-stage (multi stage) recourse
2/2/2015 3Kyle Eyvindson
Stochastic programming:
Briefly
Determine
optimal time
to conduct
inventory to
maximize ...
Maximize First
period harvest
volume, s.t.
non-declining
harvest.
www.helsinki.fi/yliopisto
• From LP to SP – a 2 stand example:
H – harvest, T – Thin, N – Do nothing.
2/2/2015 4Kyle Eyvindson
SP formulated through a deterministic
approximation of the uncertainties.
(Birge and Louveaux 2011)
t=0
t=1
t=2
H NT
N N H T N
NT
N H T N
H NT
N N H T N
NT
N H T N
H NT
N N H T N
NT
N H T N
H NT
N N H T N
NT
N H T N
H NT
N N H T N
NT
N H T N
H NT
N N H T N
NT
N H T N
Each scenario is a representation of the current and future forest resources
www.helsinki.fi/yliopisto
• This requires the known (or estimated) distribution of
the error.
• A number of scenarios are developed to approximate
the distribution. (King and Wallace 2012)
‒ A need for balance:
 too many scenarios – tractability issues
 too few scenarios – problem representation issues
Kyle Eyvindson
Incorporating uncertainty into
the planning problem
www.helsinki.fi/yliopisto
• It depends on:
• the formulation used,
• the risk preferences involved,
• the amount of uncertainty under consideration
• the accuracy required
• One way to determine an appropriate number of scenarios
is through the sample average approximation (SAA,
Kleywegt et al. 2001.)
Kyle Eyvindson
How many scenarios is enough?
www.helsinki.fi/yliopisto
• A method for evaluating the quality of a stochastic
solution.
• The algorithm simply:
‒ Select the size of the samples (N and N’), and number of
replications (M)
‒ For each m in M:
‒ Solve the problem
‒ This provides an estimate of the objective function (using N), and with
this solution, evaluate the problem using N’
‒ Evaluate the optimality gap and variance of the estimator – if
gap is too high, increase N and/or N’
Kyle Eyvindson
Sample Average Approximation
(Kleywegt et al. 2001)
N’>>N
www.helsinki.fi/yliopisto
• A forest where the DM wishes
to
• maximize first period income
‒ subject to:
‒ even flow constraints;
‒ and an end inventory constraint.
• Small forest holding
‒ 47.3 hectares, 41 stands
Forest planning problem
2.2.2015 8Kyle Eyvindson
22%
17%
20%
9%
32%
Age Class Distribution (years)
0-20
20-40
40-60
60-80
80+
30%
8%
9%
6%
31%
16%
Diameter Distribution (m)
0-5
5-10
10-15
15-20
20-25
25+
0
10
20
30
40
50
60
Pine Spruce Birch
Wood Volume (m3/ha)
www.helsinki.fi/yliopistoKyle Eyvindson
• Two cases are studied:
• The case where only the inventory uncertainty is
included
• and where both inventory uncertainty and growth model
errors are included.
• A few assumptions were made:
1. A recent inventory was conducted
2. The inventory method was assumed to have an error
which was normally distributed, mean zero and a standard
deviation of 20% of the mean height and basal area.
Scenario generation approach:
www.helsinki.fi/yliopisto
• For each inventory error, a set of 50 growth model error
scenarios were simulated.
• The growth model errors were generated using a one
period autoregressive process [AR(1)], using the same
models as Pietilä et al. 2010.
• Forest simulation was done using SIMO (Rasinmäki et al.
2009)
• Created a set of 528 schedules for the 41 stands (~13 schedules per
stand) for each scenario.
2.2.2015 10Kyle Eyvindson
Scenario generation approach:
(2)
www.helsinki.fi/yliopistoKyle Eyvindson
• A standard even flow problem.
• Maximize: 1st period incomes
‒ subject to even flow and end inventory constraints
 Using both hard and soft constraints
• For application in a stochastic setting this problem needs slight
modification:
• Maximize: Expected 1st period incomes – sum of scenario
based negative deviations
‒ subject to soft even flow an end inventory constraints
 Having strict constraints is not the real intention behind the even-flow
problem.
 The soft constraints allow for a ‘more or less’ even flow in all scenarios.
Sample problem:
www.helsinki.fi/yliopisto
• Deterministic solution
Kyle Eyvindson
A visualization:
Soft constraints Hard constraints
www.helsinki.fi/yliopisto
• Stochastic solution
Kyle Eyvindson
A visualization:
Light weight on negative deviations Strong weight on negative deviations
www.helsinki.fi/yliopistoKyle Eyvindson
Results of the SAA:
Light weight on negative deviations:
Only inventory errors Inventory and Growth model errors
www.helsinki.fi/yliopistoKyle Eyvindson
Results of the SAA:
Strong weight on negative deviations:
Only inventory errors Inventory and Growth model errors
www.helsinki.fi/yliopisto
• The size of the stochastic problem need not be
enormous.
• The size of the problem depends upon:
‒ the amount of uncertainty under consideration,
‒ the importance the uncertainty has in the problem
formulation, and
‒ the acceptability of selecting a ‘sub-optimal’ solution.
• A stochastic program with a sizable optimality gap still
outperform the deterministic equivalent.
2.2.2015 16Kyle Eyvindson
Conclusions:
www.helsinki.fi/yliopisto
• Birge, J.R., and Louveaux, F. 2011. Introduction to stochastic programming. Second
edition. Springer, New York. 499 p.
• Kangas, A., Hartikainen, M., and Miettinen, K. 2013. Simultaneous optimization of
harvest schedule and measurement strategy. Scand. J. Forest Res.(ahead-of-print),
1-10. doi: 10.1080/02827581.2013.823237.
• Kleywegt, Shapiro, Homem-de-Mello. 2001. The sample average approximation for
stochastic discrete optimization. SIAM. J. OPTIM. (12:2) 479-502.
• King, A.J., and Wallace, S.W. 2012 Modeling with Stochastic Programming,
Springer, New York
• Krzemienowski, A. & Ogryczak W. 2005. On extending the LP computable risk
measures to account downside risk. Computational Optimization and Applications
32:133-160.
• Rasinmäki, J., Mäkinen, A., and Kalliovirta, J. 2009. SIMO: an adaptable simulation
framework for multiscale forest resource data. Comput. Electron. Agric. 66(1): 76–
84. doi: 10.1016/j.compag.2008.12.007.
• Pietilä, Kangas, Mäkinen, Mehtätalo. 2010. Influence of Growth Prediction Errors on
the Expeced Loses from Forest Decisions. Silva Fennica 44(5). 829:843.
2.2.2015 17Kyle Eyvindson
References:

More Related Content

Similar to Eyvindson iufro

Risk Event Modeling and Event Chains
Risk Event Modeling and Event ChainsRisk Event Modeling and Event Chains
Risk Event Modeling and Event ChainsIntaver Insititute
 
Operations Research.pptx
Operations Research.pptxOperations Research.pptx
Operations Research.pptxbanhi.guha
 
Machine Learning part 3 - Introduction to data science
Machine Learning part 3 - Introduction to data science Machine Learning part 3 - Introduction to data science
Machine Learning part 3 - Introduction to data science Frank Kienle
 
Week12 slides
Week12 slidesWeek12 slides
Week12 slideshenry KKK
 
Risk Management Best Practices
Risk Management Best PracticesRisk Management Best Practices
Risk Management Best PracticesPMILebanonChapter
 
Stochastic Analysis of Resource Plays: Maximizing Portfolio Value and Mitiga...
Stochastic Analysis of Resource Plays:  Maximizing Portfolio Value and Mitiga...Stochastic Analysis of Resource Plays:  Maximizing Portfolio Value and Mitiga...
Stochastic Analysis of Resource Plays: Maximizing Portfolio Value and Mitiga...Portfolio Decisions
 
Forecasting and Quantification - Presentation - EMBA Degree Individual Assign...
Forecasting and Quantification - Presentation - EMBA Degree Individual Assign...Forecasting and Quantification - Presentation - EMBA Degree Individual Assign...
Forecasting and Quantification - Presentation - EMBA Degree Individual Assign...WilfredGitaari2
 
Quantitative Project Risk Analysis
Quantitative Project Risk AnalysisQuantitative Project Risk Analysis
Quantitative Project Risk AnalysisIntaver Insititute
 
Online machine learning in Streaming Applications
Online machine learning in Streaming ApplicationsOnline machine learning in Streaming Applications
Online machine learning in Streaming ApplicationsStavros Kontopoulos
 
Download-manuals-surface water-software-48appliedstatistics
 Download-manuals-surface water-software-48appliedstatistics Download-manuals-surface water-software-48appliedstatistics
Download-manuals-surface water-software-48appliedstatisticshydrologyproject001
 
Download-manuals-surface water-software-48appliedstatistics
 Download-manuals-surface water-software-48appliedstatistics Download-manuals-surface water-software-48appliedstatistics
Download-manuals-surface water-software-48appliedstatisticshydrologyproject0
 
Download-manuals-surface water-software-48appliedstatistics
 Download-manuals-surface water-software-48appliedstatistics Download-manuals-surface water-software-48appliedstatistics
Download-manuals-surface water-software-48appliedstatisticshydrologyproject0
 
Risk analysis and management
Risk analysis and managementRisk analysis and management
Risk analysis and managementIvo Andreev
 
Lession 4 Qualitative Risk Analysis .pdf
Lession 4 Qualitative Risk Analysis .pdfLession 4 Qualitative Risk Analysis .pdf
Lession 4 Qualitative Risk Analysis .pdff1002753214
 
Quantitative Project Risk Analysis
Quantitative Project Risk AnalysisQuantitative Project Risk Analysis
Quantitative Project Risk AnalysisIntaver Insititute
 
David-Jan Jansen, DNB - OECD Workshop on “Climate change, Assumptions, Uncert...
David-Jan Jansen, DNB - OECD Workshop on “Climate change, Assumptions, Uncert...David-Jan Jansen, DNB - OECD Workshop on “Climate change, Assumptions, Uncert...
David-Jan Jansen, DNB - OECD Workshop on “Climate change, Assumptions, Uncert...OECD Environment
 

Similar to Eyvindson iufro (20)

cas_washington_nov2010_web
cas_washington_nov2010_webcas_washington_nov2010_web
cas_washington_nov2010_web
 
Risk Event Modeling and Event Chains
Risk Event Modeling and Event ChainsRisk Event Modeling and Event Chains
Risk Event Modeling and Event Chains
 
Operations Research.pptx
Operations Research.pptxOperations Research.pptx
Operations Research.pptx
 
Kommunikasjon: A tool for managing product quality
Kommunikasjon: A tool for managing product qualityKommunikasjon: A tool for managing product quality
Kommunikasjon: A tool for managing product quality
 
Machine Learning part 3 - Introduction to data science
Machine Learning part 3 - Introduction to data science Machine Learning part 3 - Introduction to data science
Machine Learning part 3 - Introduction to data science
 
Week12 slides
Week12 slidesWeek12 slides
Week12 slides
 
Risk Management Best Practices
Risk Management Best PracticesRisk Management Best Practices
Risk Management Best Practices
 
Stochastic Analysis of Resource Plays: Maximizing Portfolio Value and Mitiga...
Stochastic Analysis of Resource Plays:  Maximizing Portfolio Value and Mitiga...Stochastic Analysis of Resource Plays:  Maximizing Portfolio Value and Mitiga...
Stochastic Analysis of Resource Plays: Maximizing Portfolio Value and Mitiga...
 
Forecasting and Quantification - Presentation - EMBA Degree Individual Assign...
Forecasting and Quantification - Presentation - EMBA Degree Individual Assign...Forecasting and Quantification - Presentation - EMBA Degree Individual Assign...
Forecasting and Quantification - Presentation - EMBA Degree Individual Assign...
 
Quantitative Project Risk Analysis
Quantitative Project Risk AnalysisQuantitative Project Risk Analysis
Quantitative Project Risk Analysis
 
Online machine learning in Streaming Applications
Online machine learning in Streaming ApplicationsOnline machine learning in Streaming Applications
Online machine learning in Streaming Applications
 
Download-manuals-surface water-software-48appliedstatistics
 Download-manuals-surface water-software-48appliedstatistics Download-manuals-surface water-software-48appliedstatistics
Download-manuals-surface water-software-48appliedstatistics
 
Download-manuals-surface water-software-48appliedstatistics
 Download-manuals-surface water-software-48appliedstatistics Download-manuals-surface water-software-48appliedstatistics
Download-manuals-surface water-software-48appliedstatistics
 
Download-manuals-surface water-software-48appliedstatistics
 Download-manuals-surface water-software-48appliedstatistics Download-manuals-surface water-software-48appliedstatistics
Download-manuals-surface water-software-48appliedstatistics
 
Risk analysis and management
Risk analysis and managementRisk analysis and management
Risk analysis and management
 
Lession 4 Qualitative Risk Analysis .pdf
Lession 4 Qualitative Risk Analysis .pdfLession 4 Qualitative Risk Analysis .pdf
Lession 4 Qualitative Risk Analysis .pdf
 
bpm-risk-analysis
bpm-risk-analysisbpm-risk-analysis
bpm-risk-analysis
 
Quantitative Project Risk Analysis
Quantitative Project Risk AnalysisQuantitative Project Risk Analysis
Quantitative Project Risk Analysis
 
David-Jan Jansen, DNB - OECD Workshop on “Climate change, Assumptions, Uncert...
David-Jan Jansen, DNB - OECD Workshop on “Climate change, Assumptions, Uncert...David-Jan Jansen, DNB - OECD Workshop on “Climate change, Assumptions, Uncert...
David-Jan Jansen, DNB - OECD Workshop on “Climate change, Assumptions, Uncert...
 
Linear Programming-1.ppt
Linear Programming-1.pptLinear Programming-1.ppt
Linear Programming-1.ppt
 

More from questRCN

Campbell 2014 esa workshop
Campbell 2014 esa workshopCampbell 2014 esa workshop
Campbell 2014 esa workshopquestRCN
 
Roberti esa 2014 quantifying measurement uncertainty
Roberti esa 2014 quantifying measurement uncertaintyRoberti esa 2014 quantifying measurement uncertainty
Roberti esa 2014 quantifying measurement uncertaintyquestRCN
 
Taylor workshop esa_2014
Taylor workshop esa_2014Taylor workshop esa_2014
Taylor workshop esa_2014questRCN
 
Yanai esa workshop 2014
Yanai esa workshop 2014Yanai esa workshop 2014
Yanai esa workshop 2014questRCN
 
Roberti esa 2014_workshop_slides
Roberti esa 2014_workshop_slidesRoberti esa 2014_workshop_slides
Roberti esa 2014_workshop_slidesquestRCN
 
Campbell 2015 lter asm v0.1 quest
Campbell 2015 lter asm v0.1 questCampbell 2015 lter asm v0.1 quest
Campbell 2015 lter asm v0.1 questquestRCN
 
Coweeta ppt cd_ms
Coweeta ppt cd_msCoweeta ppt cd_ms
Coweeta ppt cd_msquestRCN
 
See asm 2015_gaps
See asm 2015_gapsSee asm 2015_gaps
See asm 2015_gapsquestRCN
 
Yanai quest asm 2015 part 1
Yanai quest asm 2015 part 1Yanai quest asm 2015 part 1
Yanai quest asm 2015 part 1questRCN
 
Lter 2015 levine
Lter 2015 levineLter 2015 levine
Lter 2015 levinequestRCN
 
Roberti lter asm_september2015_jr_ds_jt_jc
Roberti lter asm_september2015_jr_ds_jt_jcRoberti lter asm_september2015_jr_ds_jt_jc
Roberti lter asm_september2015_jr_ds_jt_jcquestRCN
 
Nicolas picard
Nicolas picardNicolas picard
Nicolas picardquestRCN
 
Pierre bernier iufro2014
Pierre bernier iufro2014Pierre bernier iufro2014
Pierre bernier iufro2014questRCN
 
Yanai iufro subplenary
Yanai iufro subplenaryYanai iufro subplenary
Yanai iufro subplenaryquestRCN
 
Stochastic iufro kangas
Stochastic iufro kangasStochastic iufro kangas
Stochastic iufro kangasquestRCN
 
Iufro2014 strimbu
Iufro2014 strimbuIufro2014 strimbu
Iufro2014 strimbuquestRCN
 
Taylor: Estimating uncertainty for continental scale measurements.
Taylor:  Estimating uncertainty for continental scale measurements.Taylor:  Estimating uncertainty for continental scale measurements.
Taylor: Estimating uncertainty for continental scale measurements.questRCN
 
Hobbs: Better ignorant than misled: Including uncertainty in forecasts suppo...
Hobbs:  Better ignorant than misled: Including uncertainty in forecasts suppo...Hobbs:  Better ignorant than misled: Including uncertainty in forecasts suppo...
Hobbs: Better ignorant than misled: Including uncertainty in forecasts suppo...questRCN
 
Campbell, Quantifying uncertainty in ecology: Examples from small watershed s...
Campbell, Quantifying uncertainty in ecology: Examples from small watershed s...Campbell, Quantifying uncertainty in ecology: Examples from small watershed s...
Campbell, Quantifying uncertainty in ecology: Examples from small watershed s...questRCN
 
Aiello-Lammens: Global Sensitivity Analysis for Impact Assessments.
Aiello-Lammens:  Global Sensitivity Analysis for Impact Assessments.Aiello-Lammens:  Global Sensitivity Analysis for Impact Assessments.
Aiello-Lammens: Global Sensitivity Analysis for Impact Assessments.questRCN
 

More from questRCN (20)

Campbell 2014 esa workshop
Campbell 2014 esa workshopCampbell 2014 esa workshop
Campbell 2014 esa workshop
 
Roberti esa 2014 quantifying measurement uncertainty
Roberti esa 2014 quantifying measurement uncertaintyRoberti esa 2014 quantifying measurement uncertainty
Roberti esa 2014 quantifying measurement uncertainty
 
Taylor workshop esa_2014
Taylor workshop esa_2014Taylor workshop esa_2014
Taylor workshop esa_2014
 
Yanai esa workshop 2014
Yanai esa workshop 2014Yanai esa workshop 2014
Yanai esa workshop 2014
 
Roberti esa 2014_workshop_slides
Roberti esa 2014_workshop_slidesRoberti esa 2014_workshop_slides
Roberti esa 2014_workshop_slides
 
Campbell 2015 lter asm v0.1 quest
Campbell 2015 lter asm v0.1 questCampbell 2015 lter asm v0.1 quest
Campbell 2015 lter asm v0.1 quest
 
Coweeta ppt cd_ms
Coweeta ppt cd_msCoweeta ppt cd_ms
Coweeta ppt cd_ms
 
See asm 2015_gaps
See asm 2015_gapsSee asm 2015_gaps
See asm 2015_gaps
 
Yanai quest asm 2015 part 1
Yanai quest asm 2015 part 1Yanai quest asm 2015 part 1
Yanai quest asm 2015 part 1
 
Lter 2015 levine
Lter 2015 levineLter 2015 levine
Lter 2015 levine
 
Roberti lter asm_september2015_jr_ds_jt_jc
Roberti lter asm_september2015_jr_ds_jt_jcRoberti lter asm_september2015_jr_ds_jt_jc
Roberti lter asm_september2015_jr_ds_jt_jc
 
Nicolas picard
Nicolas picardNicolas picard
Nicolas picard
 
Pierre bernier iufro2014
Pierre bernier iufro2014Pierre bernier iufro2014
Pierre bernier iufro2014
 
Yanai iufro subplenary
Yanai iufro subplenaryYanai iufro subplenary
Yanai iufro subplenary
 
Stochastic iufro kangas
Stochastic iufro kangasStochastic iufro kangas
Stochastic iufro kangas
 
Iufro2014 strimbu
Iufro2014 strimbuIufro2014 strimbu
Iufro2014 strimbu
 
Taylor: Estimating uncertainty for continental scale measurements.
Taylor:  Estimating uncertainty for continental scale measurements.Taylor:  Estimating uncertainty for continental scale measurements.
Taylor: Estimating uncertainty for continental scale measurements.
 
Hobbs: Better ignorant than misled: Including uncertainty in forecasts suppo...
Hobbs:  Better ignorant than misled: Including uncertainty in forecasts suppo...Hobbs:  Better ignorant than misled: Including uncertainty in forecasts suppo...
Hobbs: Better ignorant than misled: Including uncertainty in forecasts suppo...
 
Campbell, Quantifying uncertainty in ecology: Examples from small watershed s...
Campbell, Quantifying uncertainty in ecology: Examples from small watershed s...Campbell, Quantifying uncertainty in ecology: Examples from small watershed s...
Campbell, Quantifying uncertainty in ecology: Examples from small watershed s...
 
Aiello-Lammens: Global Sensitivity Analysis for Impact Assessments.
Aiello-Lammens:  Global Sensitivity Analysis for Impact Assessments.Aiello-Lammens:  Global Sensitivity Analysis for Impact Assessments.
Aiello-Lammens: Global Sensitivity Analysis for Impact Assessments.
 

Recently uploaded

Call Girls Moshi Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Moshi Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Moshi Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Moshi Call Me 7737669865 Budget Friendly No Advance Bookingroncy bisnoi
 
VVIP Pune Call Girls Moshi WhatSapp Number 8005736733 With Elite Staff And Re...
VVIP Pune Call Girls Moshi WhatSapp Number 8005736733 With Elite Staff And Re...VVIP Pune Call Girls Moshi WhatSapp Number 8005736733 With Elite Staff And Re...
VVIP Pune Call Girls Moshi WhatSapp Number 8005736733 With Elite Staff And Re...SUHANI PANDEY
 
VIP Model Call Girls Viman Nagar ( Pune ) Call ON 8005736733 Starting From 5K...
VIP Model Call Girls Viman Nagar ( Pune ) Call ON 8005736733 Starting From 5K...VIP Model Call Girls Viman Nagar ( Pune ) Call ON 8005736733 Starting From 5K...
VIP Model Call Girls Viman Nagar ( Pune ) Call ON 8005736733 Starting From 5K...SUHANI PANDEY
 
Proposed Amendments to Chapter 15, Article X: Wetland Conservation Areas
Proposed Amendments to Chapter 15, Article X: Wetland Conservation AreasProposed Amendments to Chapter 15, Article X: Wetland Conservation Areas
Proposed Amendments to Chapter 15, Article X: Wetland Conservation Areas💥Victoria K. Colangelo
 
CSR_Tested activities in the classroom -EN
CSR_Tested activities in the classroom -ENCSR_Tested activities in the classroom -EN
CSR_Tested activities in the classroom -ENGeorgeDiamandis11
 
Types of Pollution Powerpoint presentation
Types of Pollution Powerpoint presentationTypes of Pollution Powerpoint presentation
Types of Pollution Powerpoint presentationmarygraceaque1
 
Training Of Trainers FAI Eng. Basel Tilapia Welfare.pdf
Training Of Trainers FAI Eng. Basel Tilapia Welfare.pdfTraining Of Trainers FAI Eng. Basel Tilapia Welfare.pdf
Training Of Trainers FAI Eng. Basel Tilapia Welfare.pdfBasel Ahmed
 
Call Girls Jejuri Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Jejuri Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Jejuri Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Jejuri Call Me 7737669865 Budget Friendly No Advance Bookingroncy bisnoi
 
VIP Model Call Girls Wagholi ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Wagholi ( Pune ) Call ON 8005736733 Starting From 5K to ...VIP Model Call Girls Wagholi ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Wagholi ( Pune ) Call ON 8005736733 Starting From 5K to ...SUHANI PANDEY
 
Booking open Available Pune Call Girls Yewalewadi 6297143586 Call Hot Indian...
Booking open Available Pune Call Girls Yewalewadi  6297143586 Call Hot Indian...Booking open Available Pune Call Girls Yewalewadi  6297143586 Call Hot Indian...
Booking open Available Pune Call Girls Yewalewadi 6297143586 Call Hot Indian...Call Girls in Nagpur High Profile
 
Call Now ☎ Russian Call Girls Connaught Place @ 9899900591 # Russian Escorts ...
Call Now ☎ Russian Call Girls Connaught Place @ 9899900591 # Russian Escorts ...Call Now ☎ Russian Call Girls Connaught Place @ 9899900591 # Russian Escorts ...
Call Now ☎ Russian Call Girls Connaught Place @ 9899900591 # Russian Escorts ...kauryashika82
 
VIP Model Call Girls Chakan ( Pune ) Call ON 8005736733 Starting From 5K to 2...
VIP Model Call Girls Chakan ( Pune ) Call ON 8005736733 Starting From 5K to 2...VIP Model Call Girls Chakan ( Pune ) Call ON 8005736733 Starting From 5K to 2...
VIP Model Call Girls Chakan ( Pune ) Call ON 8005736733 Starting From 5K to 2...SUHANI PANDEY
 
BOOK Call Girls in (Dwarka) CALL | 8377087607 Delhi Escorts Services
BOOK Call Girls in (Dwarka) CALL | 8377087607 Delhi Escorts ServicesBOOK Call Girls in (Dwarka) CALL | 8377087607 Delhi Escorts Services
BOOK Call Girls in (Dwarka) CALL | 8377087607 Delhi Escorts Servicesdollysharma2066
 
$ Love Spells 💎 (310) 882-6330 in Pennsylvania, PA | Psychic Reading Best Bla...
$ Love Spells 💎 (310) 882-6330 in Pennsylvania, PA | Psychic Reading Best Bla...$ Love Spells 💎 (310) 882-6330 in Pennsylvania, PA | Psychic Reading Best Bla...
$ Love Spells 💎 (310) 882-6330 in Pennsylvania, PA | Psychic Reading Best Bla...PsychicRuben LoveSpells
 
Call Girls In Bloom Boutique | GK-1 ☎ 9990224454 High Class Delhi NCR 24 Hour...
Call Girls In Bloom Boutique | GK-1 ☎ 9990224454 High Class Delhi NCR 24 Hour...Call Girls In Bloom Boutique | GK-1 ☎ 9990224454 High Class Delhi NCR 24 Hour...
Call Girls In Bloom Boutique | GK-1 ☎ 9990224454 High Class Delhi NCR 24 Hour...rajputriyana310
 
Call Girls Ramtek Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Ramtek Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Ramtek Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Ramtek Call Me 7737669865 Budget Friendly No Advance Bookingroncy bisnoi
 
VIP Model Call Girls Bhosari ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Bhosari ( Pune ) Call ON 8005736733 Starting From 5K to ...VIP Model Call Girls Bhosari ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Bhosari ( Pune ) Call ON 8005736733 Starting From 5K to ...SUHANI PANDEY
 
Call Girls Magarpatta Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Magarpatta Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Magarpatta Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Magarpatta Call Me 7737669865 Budget Friendly No Advance Bookingroncy bisnoi
 

Recently uploaded (20)

Call Girls Moshi Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Moshi Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Moshi Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Moshi Call Me 7737669865 Budget Friendly No Advance Booking
 
VVIP Pune Call Girls Moshi WhatSapp Number 8005736733 With Elite Staff And Re...
VVIP Pune Call Girls Moshi WhatSapp Number 8005736733 With Elite Staff And Re...VVIP Pune Call Girls Moshi WhatSapp Number 8005736733 With Elite Staff And Re...
VVIP Pune Call Girls Moshi WhatSapp Number 8005736733 With Elite Staff And Re...
 
VIP Model Call Girls Viman Nagar ( Pune ) Call ON 8005736733 Starting From 5K...
VIP Model Call Girls Viman Nagar ( Pune ) Call ON 8005736733 Starting From 5K...VIP Model Call Girls Viman Nagar ( Pune ) Call ON 8005736733 Starting From 5K...
VIP Model Call Girls Viman Nagar ( Pune ) Call ON 8005736733 Starting From 5K...
 
Proposed Amendments to Chapter 15, Article X: Wetland Conservation Areas
Proposed Amendments to Chapter 15, Article X: Wetland Conservation AreasProposed Amendments to Chapter 15, Article X: Wetland Conservation Areas
Proposed Amendments to Chapter 15, Article X: Wetland Conservation Areas
 
CSR_Tested activities in the classroom -EN
CSR_Tested activities in the classroom -ENCSR_Tested activities in the classroom -EN
CSR_Tested activities in the classroom -EN
 
Types of Pollution Powerpoint presentation
Types of Pollution Powerpoint presentationTypes of Pollution Powerpoint presentation
Types of Pollution Powerpoint presentation
 
Training Of Trainers FAI Eng. Basel Tilapia Welfare.pdf
Training Of Trainers FAI Eng. Basel Tilapia Welfare.pdfTraining Of Trainers FAI Eng. Basel Tilapia Welfare.pdf
Training Of Trainers FAI Eng. Basel Tilapia Welfare.pdf
 
Call Girls Jejuri Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Jejuri Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Jejuri Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Jejuri Call Me 7737669865 Budget Friendly No Advance Booking
 
VIP Model Call Girls Wagholi ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Wagholi ( Pune ) Call ON 8005736733 Starting From 5K to ...VIP Model Call Girls Wagholi ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Wagholi ( Pune ) Call ON 8005736733 Starting From 5K to ...
 
Booking open Available Pune Call Girls Yewalewadi 6297143586 Call Hot Indian...
Booking open Available Pune Call Girls Yewalewadi  6297143586 Call Hot Indian...Booking open Available Pune Call Girls Yewalewadi  6297143586 Call Hot Indian...
Booking open Available Pune Call Girls Yewalewadi 6297143586 Call Hot Indian...
 
Call Now ☎ Russian Call Girls Connaught Place @ 9899900591 # Russian Escorts ...
Call Now ☎ Russian Call Girls Connaught Place @ 9899900591 # Russian Escorts ...Call Now ☎ Russian Call Girls Connaught Place @ 9899900591 # Russian Escorts ...
Call Now ☎ Russian Call Girls Connaught Place @ 9899900591 # Russian Escorts ...
 
VIP Model Call Girls Chakan ( Pune ) Call ON 8005736733 Starting From 5K to 2...
VIP Model Call Girls Chakan ( Pune ) Call ON 8005736733 Starting From 5K to 2...VIP Model Call Girls Chakan ( Pune ) Call ON 8005736733 Starting From 5K to 2...
VIP Model Call Girls Chakan ( Pune ) Call ON 8005736733 Starting From 5K to 2...
 
Call Girls In Yamuna Vihar꧁❤ 🔝 9953056974🔝❤꧂ Escort ServiCe
Call Girls In Yamuna Vihar꧁❤ 🔝 9953056974🔝❤꧂ Escort ServiCeCall Girls In Yamuna Vihar꧁❤ 🔝 9953056974🔝❤꧂ Escort ServiCe
Call Girls In Yamuna Vihar꧁❤ 🔝 9953056974🔝❤꧂ Escort ServiCe
 
BOOK Call Girls in (Dwarka) CALL | 8377087607 Delhi Escorts Services
BOOK Call Girls in (Dwarka) CALL | 8377087607 Delhi Escorts ServicesBOOK Call Girls in (Dwarka) CALL | 8377087607 Delhi Escorts Services
BOOK Call Girls in (Dwarka) CALL | 8377087607 Delhi Escorts Services
 
$ Love Spells 💎 (310) 882-6330 in Pennsylvania, PA | Psychic Reading Best Bla...
$ Love Spells 💎 (310) 882-6330 in Pennsylvania, PA | Psychic Reading Best Bla...$ Love Spells 💎 (310) 882-6330 in Pennsylvania, PA | Psychic Reading Best Bla...
$ Love Spells 💎 (310) 882-6330 in Pennsylvania, PA | Psychic Reading Best Bla...
 
Call Girls In Bloom Boutique | GK-1 ☎ 9990224454 High Class Delhi NCR 24 Hour...
Call Girls In Bloom Boutique | GK-1 ☎ 9990224454 High Class Delhi NCR 24 Hour...Call Girls In Bloom Boutique | GK-1 ☎ 9990224454 High Class Delhi NCR 24 Hour...
Call Girls In Bloom Boutique | GK-1 ☎ 9990224454 High Class Delhi NCR 24 Hour...
 
Call Girls Ramtek Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Ramtek Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Ramtek Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Ramtek Call Me 7737669865 Budget Friendly No Advance Booking
 
VIP Model Call Girls Bhosari ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Bhosari ( Pune ) Call ON 8005736733 Starting From 5K to ...VIP Model Call Girls Bhosari ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Bhosari ( Pune ) Call ON 8005736733 Starting From 5K to ...
 
Sustainable Packaging
Sustainable PackagingSustainable Packaging
Sustainable Packaging
 
Call Girls Magarpatta Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Magarpatta Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Magarpatta Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Magarpatta Call Me 7737669865 Budget Friendly No Advance Booking
 

Eyvindson iufro

  • 1. www.helsinki.fi/yliopisto Assessment of Scenario Generation Approaches for Forest Management Planning through Stochastic Programming Kyle Eyvindson and Annika Kangas 2.2.2015Kyle Eyvindson 1
  • 2. www.helsinki.fi/yliopisto • Aim is to: • Integrate uncertainty into the development of forest plans ‒ Inventory, growth models, climate change... • Produce a robust solution which meets the demands of the decision maker(s), and can accommodate preferences towards risks • One method is through stochastic programming ‒ issues of tractability can become an issue 2.2.2015 2Kyle Eyvindson Introduction
  • 3. www.helsinki.fi/yliopisto • Mathematical optimization where some parameters are uncertain. • Depending on the structure of the problem, different problem formulation alternatives are available ‒ simple recourse ‒ two-stage (multi stage) recourse 2/2/2015 3Kyle Eyvindson Stochastic programming: Briefly Determine optimal time to conduct inventory to maximize ... Maximize First period harvest volume, s.t. non-declining harvest.
  • 4. www.helsinki.fi/yliopisto • From LP to SP – a 2 stand example: H – harvest, T – Thin, N – Do nothing. 2/2/2015 4Kyle Eyvindson SP formulated through a deterministic approximation of the uncertainties. (Birge and Louveaux 2011) t=0 t=1 t=2 H NT N N H T N NT N H T N H NT N N H T N NT N H T N H NT N N H T N NT N H T N H NT N N H T N NT N H T N H NT N N H T N NT N H T N H NT N N H T N NT N H T N Each scenario is a representation of the current and future forest resources
  • 5. www.helsinki.fi/yliopisto • This requires the known (or estimated) distribution of the error. • A number of scenarios are developed to approximate the distribution. (King and Wallace 2012) ‒ A need for balance:  too many scenarios – tractability issues  too few scenarios – problem representation issues Kyle Eyvindson Incorporating uncertainty into the planning problem
  • 6. www.helsinki.fi/yliopisto • It depends on: • the formulation used, • the risk preferences involved, • the amount of uncertainty under consideration • the accuracy required • One way to determine an appropriate number of scenarios is through the sample average approximation (SAA, Kleywegt et al. 2001.) Kyle Eyvindson How many scenarios is enough?
  • 7. www.helsinki.fi/yliopisto • A method for evaluating the quality of a stochastic solution. • The algorithm simply: ‒ Select the size of the samples (N and N’), and number of replications (M) ‒ For each m in M: ‒ Solve the problem ‒ This provides an estimate of the objective function (using N), and with this solution, evaluate the problem using N’ ‒ Evaluate the optimality gap and variance of the estimator – if gap is too high, increase N and/or N’ Kyle Eyvindson Sample Average Approximation (Kleywegt et al. 2001) N’>>N
  • 8. www.helsinki.fi/yliopisto • A forest where the DM wishes to • maximize first period income ‒ subject to: ‒ even flow constraints; ‒ and an end inventory constraint. • Small forest holding ‒ 47.3 hectares, 41 stands Forest planning problem 2.2.2015 8Kyle Eyvindson 22% 17% 20% 9% 32% Age Class Distribution (years) 0-20 20-40 40-60 60-80 80+ 30% 8% 9% 6% 31% 16% Diameter Distribution (m) 0-5 5-10 10-15 15-20 20-25 25+ 0 10 20 30 40 50 60 Pine Spruce Birch Wood Volume (m3/ha)
  • 9. www.helsinki.fi/yliopistoKyle Eyvindson • Two cases are studied: • The case where only the inventory uncertainty is included • and where both inventory uncertainty and growth model errors are included. • A few assumptions were made: 1. A recent inventory was conducted 2. The inventory method was assumed to have an error which was normally distributed, mean zero and a standard deviation of 20% of the mean height and basal area. Scenario generation approach:
  • 10. www.helsinki.fi/yliopisto • For each inventory error, a set of 50 growth model error scenarios were simulated. • The growth model errors were generated using a one period autoregressive process [AR(1)], using the same models as Pietilä et al. 2010. • Forest simulation was done using SIMO (Rasinmäki et al. 2009) • Created a set of 528 schedules for the 41 stands (~13 schedules per stand) for each scenario. 2.2.2015 10Kyle Eyvindson Scenario generation approach: (2)
  • 11. www.helsinki.fi/yliopistoKyle Eyvindson • A standard even flow problem. • Maximize: 1st period incomes ‒ subject to even flow and end inventory constraints  Using both hard and soft constraints • For application in a stochastic setting this problem needs slight modification: • Maximize: Expected 1st period incomes – sum of scenario based negative deviations ‒ subject to soft even flow an end inventory constraints  Having strict constraints is not the real intention behind the even-flow problem.  The soft constraints allow for a ‘more or less’ even flow in all scenarios. Sample problem:
  • 12. www.helsinki.fi/yliopisto • Deterministic solution Kyle Eyvindson A visualization: Soft constraints Hard constraints
  • 13. www.helsinki.fi/yliopisto • Stochastic solution Kyle Eyvindson A visualization: Light weight on negative deviations Strong weight on negative deviations
  • 14. www.helsinki.fi/yliopistoKyle Eyvindson Results of the SAA: Light weight on negative deviations: Only inventory errors Inventory and Growth model errors
  • 15. www.helsinki.fi/yliopistoKyle Eyvindson Results of the SAA: Strong weight on negative deviations: Only inventory errors Inventory and Growth model errors
  • 16. www.helsinki.fi/yliopisto • The size of the stochastic problem need not be enormous. • The size of the problem depends upon: ‒ the amount of uncertainty under consideration, ‒ the importance the uncertainty has in the problem formulation, and ‒ the acceptability of selecting a ‘sub-optimal’ solution. • A stochastic program with a sizable optimality gap still outperform the deterministic equivalent. 2.2.2015 16Kyle Eyvindson Conclusions:
  • 17. www.helsinki.fi/yliopisto • Birge, J.R., and Louveaux, F. 2011. Introduction to stochastic programming. Second edition. Springer, New York. 499 p. • Kangas, A., Hartikainen, M., and Miettinen, K. 2013. Simultaneous optimization of harvest schedule and measurement strategy. Scand. J. Forest Res.(ahead-of-print), 1-10. doi: 10.1080/02827581.2013.823237. • Kleywegt, Shapiro, Homem-de-Mello. 2001. The sample average approximation for stochastic discrete optimization. SIAM. J. OPTIM. (12:2) 479-502. • King, A.J., and Wallace, S.W. 2012 Modeling with Stochastic Programming, Springer, New York • Krzemienowski, A. & Ogryczak W. 2005. On extending the LP computable risk measures to account downside risk. Computational Optimization and Applications 32:133-160. • Rasinmäki, J., Mäkinen, A., and Kalliovirta, J. 2009. SIMO: an adaptable simulation framework for multiscale forest resource data. Comput. Electron. Agric. 66(1): 76– 84. doi: 10.1016/j.compag.2008.12.007. • Pietilä, Kangas, Mäkinen, Mehtätalo. 2010. Influence of Growth Prediction Errors on the Expeced Loses from Forest Decisions. Silva Fennica 44(5). 829:843. 2.2.2015 17Kyle Eyvindson References:

Editor's Notes

  1. One scenario represents one deterministic version of the possible future. Inventory, growth errors and climate change errors can be incorporated into the scenario development.
  2. Here I’ll describe the difference between the result and the scenarios.
  3. Here I’ll describe what is going on with the solutions. A much steadier flow of income over the periods – some negative deviations, yes, but a much flatter profile.