Delusional REDD baselines


Published on

Credible baseline setting and accurate and transparent Measurement, Reporting and Verification (MRV) of results are key conditions for the success of REDD (Reducing Emissions from Deforestation and forest Degradation). In this presentation, Britaldo Silveira Soares Filho from Universidade Federal de Minas Gerais discusses the challenges inherent in setting baselines and applying modelling for REDD.

Britaldo Silveira Soares Filho gave this presentation on 8 March 2012 at a workshop organised by CIFOR, ‘Measurement, Reporting and Verification in Latin American REDD+ Projects’, held in Petropolis, Brazil. The workshop aimed to explore important advances, challenges, pitfalls, and innovations in REDD+ methods — thereby moving towards overcoming barriers to meeting MRV requirements at REDD+ project sites in two of the Amazon’s most important REDD+ candidate countries, Peru and Brazil. For further information about the workshop, please contact Shijo Joseph via s.joseph (at)

Published in: Education, Technology, Business
1 Like
  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Delusional REDD baselines

  1. 1. Delusional REDD baselines il ho F Britaldo Silveira Soaress Filho re S oa e ira Si lv al do r it B MEASUREMENT, REPORTING AND VERIFICATION IN LATIN AMERICAN REDD+ PROJECTS CIFOR WORKSHOP – MARCH 8-9, 2012 – PETRÓPOLIS, BRAZIL
  2. 2. Main concepts• Additionality: Proof that reduction in emissionsil ho from REDD is additional to reductions that would F occur if initiative were not in place. ar es So• Leakage: reduction in carbon emissions in one area that results in increasedve ira l emissions in another Si reduced emissions from• Permanence: long-term REDD. Depends on al do vulnerability to deforestation and/or it Br degradation. Concepts and methods borrowed from CDM projects
  3. 3. Measuring performance: the baseline approach The CDM baseline il ho F ar es So e ira Si lv al do r it Historical approach: introducing Forward-looking approach (ex ante): B an innovative process to an installed industrial plant. the building of a new factory with a carbon neutral approach. Depend strongly on the project and less on the context (Lesser risk of hot air).
  4. 4. REDD baselines ho30000 27772 km225000 Fil20000 25396 19600 es ar Baseline So 19014 15405 1822615000 14196 12911 ira Reduction10000 8935 Hitorical lve74645000 Si 6238 6560 REDD 3805 do 0 t al iapproach: Brazil´s Br Change Plan. Historical National Climate Forward-looking approach: Brazil overall GHG target of 39% reduction by 2020 (emissions from agriculture can increase from 480 to 627 MCO2)
  5. 5. What to do with countries or regions with lowdeforestation rates and large extent of forests Acre Madre de Dios Pando il ho F ar es So e ira Si lv al do r it B For more information: visit
  6. 6. Forward-looking baselineC stocks12,000,000 il ho F10,000,000 ar es So 8,000,000 ira 6,000,000 lv e with project Si 4,000,000 Additionality = $$ credits 2,000,000 l do baseline itaProject starts - time Br
  7. 7. Forward-looking baseline il ho F ar es So e ira Si lv al do r it B Project startsBaseline is not solely based on the project itself, but on a referenceregion. Therefore, there is a need to understand the effect of the projecton the deforestation regional trend.
  8. 8. Solution: Deforestation model?? BASIC STEPS OF MODELING il ho F ar es1. understanding historical So trends ira ve2. identifying drivers of deforestation il S3. modelingofuture scenarios ld ita Br A series of what if decisions !!!
  9. 9. Understanding historical trends Modeling approaches for BAU baselines il ho 160000 F ha 140000 average ar es projected Which one ? 120000 So (BAU) simulated 100000 e ira Observed• Historical fixed rates Si lv 80000• Historical trend do 60000 it• Business-as-usual al simulation r B(future 40000 20000 scenarios) 0 time
  10. 10. Modeling deforestation trajectories il ho F ar es So e ira Si lv al do r it B 2010 2007 Soares-Filho et al. 2010. PNAS.
  11. 11. Modeling scenarios Soares-Filho et al. 2006. Nature il ho F ar es So e ira Si lv al do r itB End of Deforestation in the Brazilian AmazonThe Nesptad, Soares-Filho et al. 2009. Science. Trajectories can change drastically
  12. 12. Other aspects• Delimiting the geographic reference o ilh region. F ar es So ira ve Project il S al doReference region r it B
  13. 13. Other aspects• Delimiting the geographic reference o ilh region. F ar es road So ira ve Project il S al doReference region r it B
  14. 14. Other aspects • What if there are other projects???ho il F ar es Project So e ira Si lv Project Project al doReference region r it BMadre de Dios, Peru, have 12 or more REDD projects
  15. 15. identifying drivers of deforestation Probability map of deforestation deforestation up to 1994 distance to main roads ho deforestation up to 1999 overlay F il cross-tabulation buffer zone ar es deforested non-deforested total ---------------------------------------- distance > 10 km | 194187 distance < 10 km | 84228 1005296 | 1199483 199349 | 283577 So ira ---------------------------------------- total | 278415 1204645 | 1483060 deforestation from 1994 to 1999 (gray tone) lv e Si al doThey are in fact spatial determinants r it BMost models only incorporate spatial determinants!!!
  16. 16. modeling future scenarios • Spatially-explicit model of deforestation to quantify where deforestation is likely to occur. o Land cover map of 1986 Land cover map of 1994 Filh Simulated land cover map of 1994 O-09 50` O -09 50` ar esO -09 50` project So project ira O-10 00` O -10 00` O -10 00` lv e O-10 10` O -10 10` Si O -10 10` al do O-10 20` r it deforestation is not random but occurs at locations that have a O -10 20` O -10 20` B combination of bio-geophysical and economic attributes that are more attractive to the agents of deforestation O-10 30` O -10 30` O -10 30` True or false? O O -55 00 ` -54 50` O -55 00` O -54 50` O -55 00` O -54 50`
  17. 17. Stochastic nature of deforestationAbsolute frequency of transition probabilities of observed deforestation 3500 scene 23167 3000 2500 il Performance ho x realismnumber of cells 2000 1500 F 1000 ar esdecreases realism Increasing spatial performance So 500 0 0 50 100 150 200 250 ira probability value Half true, half false lv e Si al do r it B Dinamica EGO simulates landscape structure
  18. 18. Validation only regarding location, but training not spatial structure t1 t2 t3 il ho Land cover map of 1986 Land cover map of 1994 F Simulated land cover map of 1994 O-09 50` O -09 50` ar esO -09 50` So project project Model ira O-10 00` O -10 00` O -10 00` ilvevalidation O-10 10` O -10 10` S O -10 10` al do O-10 20` r it O -10 20` O -10 20` B O-10 30` O -10 30` O -10 30` O O -55 00 ` -54 50` O -55 00` O -54 50` O -55 00` O -54 50`
  19. 19. Model performance assessment• Map comparison methods (how to lie with validation methods) il ho F ar es So e ira Label 1 Costanza Si lv2 Power et al. 3 Pontius Hagen 4 Van Vliet et al. 5 6 Almeida et 7 Almeida et do map author (1989) (2001) (2002) (2003) (2011) al. (2008)1 al. (2008)2 degraded 5x5 0.98 0.83 0.65 0.70 0.66 0.83 1.00 degraded 9x9 degraded 15x15 it al 0.97 0.96 0.81 0.81 0.48 0.34 0.47 0.24 0.48 0.34 0.62 0.46 0.75 0.55 B r degraded 31x31 degraded 51x51 0.95 0.95 0.80 0.80 0.18 0.12 -0.03 -0.13 0.18 0.12 0.26 0.19 0.32 0.23 random 0.90 0.84 0.002 1.00 1.00 0.04 0.05 line1 x line2 1.00 0.81 -0.02 0.59 -0.02 1.00 1.00 Soares-Filho et al. in review And model comparisons are quite subjective (e.g. Vega et al. 2011)
  20. 20. Different models have different assumptions, geographic scales, spatial resolutions, scopes, and, therefore, purposes. il ho F ar es So e ira Si lv al do Thank you r it Reimer BBoth simulations developed using Dinamica EGO, but none of them were designed to fixbaselines
  21. 21. This means trouble il ho F ar es So e ira Si lv al do r itB
  22. 22. What about leakage?In-out, out-out, diffuse leakage il ho(Fearnside, 2009) F ar es So Model wizard for rapid e ira Is modeling for dummies??? assessment ilv S ld o Leakage Enter value Effectiveness = Success rate – Leakage Success ------------ 50% = Effectiveness ------------ 70% - 20% ita Br Representation of a fictitious software Leakage cannot be simulated, but can be evaluated. See Soares-Filho et al. PNAS. 2010
  23. 23. Good job &Business il ho F ar es So • But!!!! ira (although vendors say so) veNo ready solution lfor REDD Si al do software wizard… There is no magic solution, via it Br MODEL must be built from the ground to incorporate local knowledge…
  24. 24. In sum• Although various MRV methodologies for mapping carbon stocks and modeling il ho reductions as well as standards for international certification have been developed, the criteria for establishing baselines for crediting purposes are F unsound; they do not take into account that forward-looking baselines are es questionable because deforestation trajectories can alter drastically in response to ar So changes in a complex set of circumstances (Nunes et al. 2012; Soares-Filho et al. IADB report, 2012).• ira Furthermore, it is very difficult to isolate the local effect of a project from the e overall trajectory of deforestation as well as to assess leakage arising from the Si lv establishment of projects, and none of those projects have attempted to perform such analyses (Soares-Filho et al. IADB report, 2012).• do As more specific standards and indicators are established, more flawed they al become, because they are unfounded anyway (Personal Perception). And for r it financial crediting, every dollar counts. In addition, conflict of interest might lead B to gamming (Fearnside, 2012).
  25. 25. So, how can we apply modeling to REDD? (let’s keep us in business) il ho F ar Knowledge es Intervention So Modeling e ira Si lv Information al do r it B Tool for policy design and planning
  26. 26. Tool for REDD planning and management. Application to Acre Dinamica EGO software il ho F ar es So e ira Si lv al do r it B Simulation as a tool for regional planning Method adopted by SISA
  27. 27. Setting priority areas for REDD• Index of threat (time dependent) (Soares-Filho et al. PNAS, 2010) o lh Fi ar es So e ira Si lv al do r it B Priorização de áreas protegidas
  28. 28. Calculating costs of reduction Amazon opportunity costs il ho F ar es So e ira Si lv Carbon cost along the l doU$ ita Carbon Cost Br30 deforestation frontier2520151050 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 year year (Nespdad, Soares-Filho et al. 2009; Soares-Filho et al. 2010)
  29. 29. Assessing effectiveness and leakage over time Only monitoring deforestation is not enough. il ho F Soares-Filho et al. PNAS . 2010. ar es So Soares-Filho et al. Env. Con. 2012. Nunes, ira Odds ratio adjustedLand use zones 1997 2000 2005 MeanUrban and periurban areas 6.15 2.17 lv e5.40 4.57Mining concessionsSmall-scale cattle ranching 1.26 1.42 Si 1.62 1.20 1.51 1.28 1.46 1.30 doSmall landowner lots 1.02 0.99 1.55 1.19Brazil nut concessionsLogging concessions ital 0.76 2.84 0.67 3.28 0.49 4.29 0.64 3.47Native communities Br 0.77 1.20 1.35 1.11Reforestation concessions 0.19 0.39 0.93 0.51Conservation concessions 3.85 2.78 2.06 2.89Ecotourism concessions 0.08 0.08 0.76 0.31Protected areas 4.85 8.27 5.55 6.22Indigenous reserves 0.20 0.02 x 0.11Unassigned land use 0.96 0.94 1.17 1.02
  30. 30. Which REDD projects in Madre de Dios can make a difference? il ho F ar es So Nunes, Soares-Filho et al. Env. Con. 2012. e ira Si lv al do r itHajek et al. 2011 B
  31. 31. Channeling REDD+ investments il ho F ar es So REDD funds invested to e ira upgrade the production Si lv chain of non- timber forest al do products r it BNunes, Soares-Filho et al. Economic benefits of forest conservation:assessing the potential rents from Brazil nut concessions in Madre de Dios,Peru, to channel REDD+ investments. Environmental Conservation, 2012.
  32. 32. Thank you/Obrigado/Gracias Britaldo Silveira Soares Filho il ho(the culprit for developing deforestation models that are now being applied to REDD+) britaldo@csr.ufmg.brs F re S oa For many more examples, please access ira ve il S al do it Br Modeling in support of sound policyDinamica EGO freeware can be downloaded at